<%BANNER%>

Development of a Management Focused Decision Support Tool for Okeechobee Basin Beef Cattle Agroecosystems

Permanent Link: http://ufdc.ufl.edu/UFE0021840/00001

Material Information

Title: Development of a Management Focused Decision Support Tool for Okeechobee Basin Beef Cattle Agroecosystems
Physical Description: 1 online resource (120 p.)
Language: english
Creator: Jagannathan, Sudarshan
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: Agricultural and Biological Engineering -- Dissertations, Academic -- UF
Genre: Agricultural and Biological Engineering thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Agricultural enterprises require resource management that involve many trade-offs within a complex ecological and financial environment. As an example enterprise within South Central Florida, the MacArthur Agro-Ecology Research Center (MAERC) located on the Buck Island Ranch (BIR) in Lake Placid Florida has a major objective to optimize its long term sustainability in both ecological and economic facets. MAERC/BIR combines a research facility with a commercial-scale, beef cattle enterprise (10,300 acres) to explore the role of long-term ecological and social dynamics within sub-tropical grazing systems (www.maerc.org). In order to maintain long term viability and sustainability, a balance between ranch profitability and reduction of non point source pollution effects needs to be established and studied. A possible solution to this challenge is to create a Decision Support System (DSS) for beef cattle enterprises. Such a DSS could serve to communicate simulation results and metrics effectively to the ranch operators, whose focus would be on profitability, as well as the researchers and conservationists, whose focus would be on limiting the effects of non point source pollution. Thus, the objective of this research project is to design and construct a decision support model of a beef cattle ranch system to simulate selected beef cattle and ranch management operations on a southern Florida beef cattle enterprise and to explore the management decisions with respect to water resource factors such as runoff and nutrient loading. The Questions and Decisions ? (QnD?) model system was created to provide an effective and efficient tool to integrate ecosystem, management, economic and socio-political factors into a user-friendly model/game framework. This model is a unique and new development since no other model before has modeled scenarios on a ranch-scale. The model is also good in that it is more than just a hydrological model but also a decision support tool for managers with a user interface that helps them in real-time decision making. The QnD model links spatial components within geographic information system (GIS) files to the abiotic (climatic) and biotic interactions that exist in an environmental system. QnD can be constructed with any combination of detailed technical data or estimated interactions of the ecological/management/social/economic forces influencing an ecosystem. The specific QnD version has been developed for the BIR (QnD:BIR) using the conceptual diagram which shows the integrated ecological and economic factors at the ranch-scale. QnD:BIR uses elements of the Standardized Performance Analysis (SPA) method applied to BIR to simulate elements of beef cattle production and economic dynamics. QnD:BIR uses simplified water and phosphorous dynamics at a monthly time step generated from the long term research from southern Florida beef and dairy cattle research. QnD:BIR utilizes existing geographic information systems (GIS) coverages and monitoring data available from the MAERC/BIR facility. QnD:BIR was tested on environmental data from BIR for the period of 2000 - 2003 for sixteen experimental pastures including both improved and native pastures. Specifically, QnD:BIR simulation results of monthly runoff, phosphorus load and forage production were compared with comparable field-scale data. Given the coarse monthly time step, simulations of these factors were generally acceptable for use in the whole ranch simulations. Given potential climate data for the area, specific scenarios were constructed to test different management scenarios in terms of P loading and cattle production metrics. The development of QnD: BIR provides a useful and modular system, capable of running various scenarios depending on the setup for simulating both environmental and enterprise functions, within an easy to use graphical interface with the ability to move cows and manage the enterprise hands-on. Further model development and simulation could be expanded to allow more detail in cattle response to temperature and surface water availability. Qnd: BIR is a simple model that uses empirical relations with acceptable levels of accuracy (and a Nash-Sutcliffe coefficient of at least 0.5 mostly). The model also takes into account rainfall, water table depth, temperature, and soil characteristics for its hydrology and phosphorus cycle. However, since the model uses empirical relations, it cannot be applied in conditions that differ vastly from the conditions present in BIR. Also, due to its simple nature it does not take into account factors such as light, for ET, or drainage within and across pastures in the ranch and into the canals. Considering the significant positive qualities and certain limitations of the model, it can be said that the model is to be used more as a guideline to point the manager in the right direction for decision making than as a tool to provide exact values or measures for runoff or phosphorus load in the long term.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Sudarshan Jagannathan.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Kiker, Gregory.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021840:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021840/00001

Material Information

Title: Development of a Management Focused Decision Support Tool for Okeechobee Basin Beef Cattle Agroecosystems
Physical Description: 1 online resource (120 p.)
Language: english
Creator: Jagannathan, Sudarshan
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: Agricultural and Biological Engineering -- Dissertations, Academic -- UF
Genre: Agricultural and Biological Engineering thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Agricultural enterprises require resource management that involve many trade-offs within a complex ecological and financial environment. As an example enterprise within South Central Florida, the MacArthur Agro-Ecology Research Center (MAERC) located on the Buck Island Ranch (BIR) in Lake Placid Florida has a major objective to optimize its long term sustainability in both ecological and economic facets. MAERC/BIR combines a research facility with a commercial-scale, beef cattle enterprise (10,300 acres) to explore the role of long-term ecological and social dynamics within sub-tropical grazing systems (www.maerc.org). In order to maintain long term viability and sustainability, a balance between ranch profitability and reduction of non point source pollution effects needs to be established and studied. A possible solution to this challenge is to create a Decision Support System (DSS) for beef cattle enterprises. Such a DSS could serve to communicate simulation results and metrics effectively to the ranch operators, whose focus would be on profitability, as well as the researchers and conservationists, whose focus would be on limiting the effects of non point source pollution. Thus, the objective of this research project is to design and construct a decision support model of a beef cattle ranch system to simulate selected beef cattle and ranch management operations on a southern Florida beef cattle enterprise and to explore the management decisions with respect to water resource factors such as runoff and nutrient loading. The Questions and Decisions ? (QnD?) model system was created to provide an effective and efficient tool to integrate ecosystem, management, economic and socio-political factors into a user-friendly model/game framework. This model is a unique and new development since no other model before has modeled scenarios on a ranch-scale. The model is also good in that it is more than just a hydrological model but also a decision support tool for managers with a user interface that helps them in real-time decision making. The QnD model links spatial components within geographic information system (GIS) files to the abiotic (climatic) and biotic interactions that exist in an environmental system. QnD can be constructed with any combination of detailed technical data or estimated interactions of the ecological/management/social/economic forces influencing an ecosystem. The specific QnD version has been developed for the BIR (QnD:BIR) using the conceptual diagram which shows the integrated ecological and economic factors at the ranch-scale. QnD:BIR uses elements of the Standardized Performance Analysis (SPA) method applied to BIR to simulate elements of beef cattle production and economic dynamics. QnD:BIR uses simplified water and phosphorous dynamics at a monthly time step generated from the long term research from southern Florida beef and dairy cattle research. QnD:BIR utilizes existing geographic information systems (GIS) coverages and monitoring data available from the MAERC/BIR facility. QnD:BIR was tested on environmental data from BIR for the period of 2000 - 2003 for sixteen experimental pastures including both improved and native pastures. Specifically, QnD:BIR simulation results of monthly runoff, phosphorus load and forage production were compared with comparable field-scale data. Given the coarse monthly time step, simulations of these factors were generally acceptable for use in the whole ranch simulations. Given potential climate data for the area, specific scenarios were constructed to test different management scenarios in terms of P loading and cattle production metrics. The development of QnD: BIR provides a useful and modular system, capable of running various scenarios depending on the setup for simulating both environmental and enterprise functions, within an easy to use graphical interface with the ability to move cows and manage the enterprise hands-on. Further model development and simulation could be expanded to allow more detail in cattle response to temperature and surface water availability. Qnd: BIR is a simple model that uses empirical relations with acceptable levels of accuracy (and a Nash-Sutcliffe coefficient of at least 0.5 mostly). The model also takes into account rainfall, water table depth, temperature, and soil characteristics for its hydrology and phosphorus cycle. However, since the model uses empirical relations, it cannot be applied in conditions that differ vastly from the conditions present in BIR. Also, due to its simple nature it does not take into account factors such as light, for ET, or drainage within and across pastures in the ranch and into the canals. Considering the significant positive qualities and certain limitations of the model, it can be said that the model is to be used more as a guideline to point the manager in the right direction for decision making than as a tool to provide exact values or measures for runoff or phosphorus load in the long term.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Sudarshan Jagannathan.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Kiker, Gregory.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021840:00001


This item has the following downloads:


Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20101108_AAAACP INGEST_TIME 2010-11-09T01:45:24Z PACKAGE UFE0021840_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 1053954 DFID F20101108_AACCHL ORIGIN DEPOSITOR PATH jagannathan_s_Page_064.tif GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
5f2c5c846e994a3a911f250259ee8c79
SHA-1
ca29ebfc0d6a0ff94b9ba249851fe5acf3bc6ce2
6867 F20101108_AACCGW jagannathan_s_Page_043thm.jpg
eaf2bb7341529c41a6bcb722a71b92ec
96dfdeaf98799a748a60517d5edfc97cdc270807
25271604 F20101108_AACCHM jagannathan_s_Page_093.tif
188e49f36a1ce0017d6d394026fff366
9ac77f89c611e0d6043767ac1d09be6a229308c5
26809 F20101108_AACCGX jagannathan_s_Page_081.QC.jpg
7e3ce5b3788cca220e8665e80cf1c72e
5c81983442b6830d5cf0e0f009d769d07e3aa74b
798 F20101108_AACCIA jagannathan_s_Page_076.txt
1992819cd081e75a578d0464e4327b2a
6790f7665119c7d8959364a9ec4e689fb12c7d86
667 F20101108_AACCHN jagannathan_s_Page_086.txt
f5b9df765b907b2f0bd63c73f9a6af6b
d3f6685b85762a94f46eb35f0924b892740ade89
24431 F20101108_AACCGY jagannathan_s_Page_045.QC.jpg
2ac3b77b189df9469ea29cfc1f3d0058
d612cb179007da0ac0414e64c0eebfbb94fcfe22
20805 F20101108_AACCIB jagannathan_s_Page_009.QC.jpg
5e56aa50cb1a7f89e701ff49f4f8bf50
c14eef2227fbdf3669b3671d4b708484a2efd2ed
F20101108_AACCHO jagannathan_s_Page_120.tif
86ac2867c11f321c37e1c27bab2055f6
fd8a355ad8b0b892ff4bfbd14e0c0cebf54c012d
102882 F20101108_AACCGZ jagannathan_s_Page_044.jp2
856011c2cdabecdfb5fbb85ac23af069
529ad82cf84108c4f21af68dc6f256daec728c1e
6744 F20101108_AACCIC jagannathan_s_Page_082thm.jpg
537c8f55e450e5463f2bf6e8d7e1a766
6187df34f72ef981066059c942ea3e0616104307
55866 F20101108_AACCHP jagannathan_s_Page_026.pro
2990ceeb9818d6cb4c9f0bfe7f773b23
0864f3b529538b0ddc0b08b9aa2324ebd92e6826
F20101108_AACCID jagannathan_s_Page_077.tif
467d5a6316bd192276633a9abe892ef8
1652f60f8df3a8b37c3ec818ebf1b4456e871e4f
3973 F20101108_AACCHQ jagannathan_s_Page_100thm.jpg
218a91a778492833a8dcec071c3dea2b
87b0fd9305c167d69e1fe9154ef05f900f119823
6946 F20101108_AACCIE jagannathan_s_Page_026thm.jpg
e4d7e9a3b5aa790ffac032612f60e371
a9f5509b6d7bcaef29c4767aa42c45ca722b60b0
F20101108_AACCIF jagannathan_s_Page_003.tif
99f50312c22632ad3916dd4b8a568d40
5543026b13afea4a8739c9fc155de60152ec4866
6687 F20101108_AACCHR jagannathan_s_Page_030thm.jpg
da474bb46b39e38b221127607ebc4027
012c3528cc1de1ff58b6788ec6082fb39d146874
24388 F20101108_AACCIG jagannathan_s_Page_064.QC.jpg
9b752ec18d893f6df737c1ccd2696c50
85f1698beb238047c27add6eac6851def6bf9244
F20101108_AACCHS jagannathan_s_Page_080.tif
dbb74f4644e4e2fb125d8ea0d4c7aad1
37da482818896ea2a739a707230a7daa918f85e2
51008 F20101108_AACCIH jagannathan_s_Page_023.pro
5d19b84def16390b110272ba6927b34d
ddb8f0744165fbe7a10e6ab7a6f420aecbf46c89
48183 F20101108_AACCII jagannathan_s_Page_044.pro
c708104a14b76d5d3e331cabfc687b4e
61f8faa9e6a854224fb72a56ae80dca5f8b58983
2537 F20101108_AACCHT jagannathan_s_Page_118.txt
54a5d06ca72975591be7871ee867d4a5
4ff18f639ab1ee49d72c24ede6794d700ac3b814
6695 F20101108_AACCIJ jagannathan_s_Page_037thm.jpg
de7629c1b62c486824d7d47415733fce
f186b6ffdb3c7ef3bc10017e5c7b7dc5a01bbe88
F20101108_AACCHU jagannathan_s_Page_084.tif
65690988cb396ac757dfef04ff690e21
74e2504e296288d4cb517016b8644bc4158f90e0
15795 F20101108_AACCIK jagannathan_s_Page_016.pro
80f71b67812b145d695c65208dd49d34
7af1231ad10ac88244afcd799c8da36e4626a830
52908 F20101108_AACCHV jagannathan_s_Page_063.pro
0afb83e67f85eabe67233afb3662a8c5
39f7d996f166afda65b03a1748fbe70d85099f2d
13821 F20101108_AACCIL jagannathan_s_Page_110.pro
faffe5f2b2b03193a90050fa2794c89b
bdc00de755ba17b265c4f6d1a60b55eb82b21b7f
22792 F20101108_AACCHW jagannathan_s_Page_044.QC.jpg
bb0b3ac1750ab3a89e94aaf7c5dc296d
0d31aad54ce27006de4a93a97e0d0c240b7f0149
23050 F20101108_AACCJA jagannathan_s_Page_053.QC.jpg
d6f015c17c5f789ba82b48d864093139
b043b346b814ca4318a00b4c8bf54fc30bfb9236
6641 F20101108_AACCIM jagannathan_s_Page_040thm.jpg
b4d584feddb484f83f3bf1da476ff0e2
c451857883f35523f4500179bfb0534362092beb
528787 F20101108_AACCHX jagannathan_s_Page_086.jp2
45e8e694acb726a1052a12bf3978a1d1
c252a108ed7a66473f9d7b904b998013ae5cc50f
737 F20101108_AACCJB jagannathan_s_Page_085.txt
267bd182eb52561542ac2ac3f0e546ae
14a46a2abd20a01bab8fcf6f6b98609e2045fa0d
251009 F20101108_AACCIN jagannathan_s_Page_079.jp2
e225a967a6b23816e1474c08b8cbaa9b
2d7423c0735a34720fdd7dc1ffcb00e9b314a891
12885 F20101108_AACCHY jagannathan_s_Page_101.QC.jpg
33b1b486221fbb01ebd521884446dc45
16b9cd70bd475d41e61e382eda8392fe735b0d5d
47405 F20101108_AACCJC jagannathan_s_Page_031.pro
c8911a5ebdeab25f0b8dba73143f7768
40965be610925db41d307d292269bec841ba34e9
825 F20101108_AACCIO jagannathan_s_Page_089.txt
99195b04a784699db41cde4f7cc89e04
d3065f17e9fdd99787b9a03312accd1ce85de9f4
50309 F20101108_AACCHZ jagannathan_s_Page_021.pro
152fe420345ea71d2f2b5cf4911c3c99
e3f25c6ec0c202b417100e48f734e537f93ad35b
114805 F20101108_AACCJD jagannathan_s_Page_025.jp2
34ece479d18e873a6495a8a08e09bf02
417799129c438a396c3b32b0c9aa93173aeef6de
F20101108_AACCIP jagannathan_s_Page_040.tif
38a2e9af16229734ff238ebf8027db24
84ead667b2ed47f5925a52878c96a29cfc4ea4fa
F20101108_AACCJE jagannathan_s_Page_074.tif
d3841180b0cc0dd5d702daae83e9b507
00970a9773fa85cd789f06731801ccbc1556bb70
25178 F20101108_AACCIQ jagannathan_s_Page_021.QC.jpg
7fd26b8eb5aacd2a5f07b66587a93f0a
2cb024836abe2445ef2ff7c2648993a9ab6ac3a8
103641 F20101108_AACCJF jagannathan_s_Page_050.jp2
8df2bb456f0d8d35769ff09cd82ff2e8
b308f3ca5f9c4bf68259868faad0d58e0252e3a5
62923 F20101108_AACCIR jagannathan_s_Page_008.pro
aaa12a25c6f434a64a181b109731d372
582c883a3eec8064d93e7b96d0fafc69cb470bcd
328835 F20101108_AACCJG jagannathan_s_Page_033.jp2
646e48860ac6ebbf422c897b4eef7455
8476326017c0dea123eb0b12f16a5d2b23ab400b
3442 F20101108_AACCIS jagannathan_s_Page_077thm.jpg
411585dfc9022554f1bfb302a514b1ce
33a77df98d751c3a62c4ce60421a88cdc8190cb2
F20101108_AACCJH jagannathan_s_Page_115.tif
0387734399d122f014ca6d1e6675fd46
b74ba9fef11b3a5b87fd214697db4a31fe660617
14732 F20101108_AACCIT jagannathan_s_Page_073.pro
fe4f4b5aa62aedf3f5c17c037b76524a
e418fd7e4a3c549c98f2995b5545577e9393246a
24144 F20101108_AACCJI jagannathan_s_Page_080.QC.jpg
83234f5c4e941d641565a884c403bee1
192a52b1aa15a9723492582ec5765e4b10243d39
11093 F20101108_AACCJJ jagannathan_s_Page_035.pro
7a03f46c12c182a7a091f1ea2f4d5064
736074e9c90b1da494a1a6dc7c8fd42bb3b1a297
F20101108_AACCIU jagannathan_s_Page_109.tif
cf3859d030deb55a6818d6358ce98ff1
e14d716666e3611635af36ed745a84ad17e6caf2
113987 F20101108_AACCJK jagannathan_s_Page_043.jp2
af0cc6c81283f4e59337936c13a091a6
27147117c81d874832433d839170a6c6ddc6e48d
53110 F20101108_AACCIV jagannathan_s_Page_017.pro
1ea784e6c252dd95043b2762b9048a65
aa7d80e813e933d1b25db5c281518df0c9dc9088
63099 F20101108_AACCJL jagannathan_s_Page_013.jpg
34a69b7c5268d6c525c336fc9e5fc12c
03ff687e58c297fd900353138cea0ef12fe6e0b0
F20101108_AACCIW jagannathan_s_Page_050.tif
4300cb3c37998d1790c01156e3434705
b54536c898b7a0b8dc48aadbf1c8f0770b1fabcf
275 F20101108_AACCJM jagannathan_s_Page_034.txt
78b7d6a0d2f6b9ed4f8e215e10c35424
8bf5133d34d2d193243181be3e04807851fd1cd1
76053 F20101108_AACCIX jagannathan_s_Page_027.jpg
e7753599b8b4a45a07b230e65d6d54b8
1d126c2b72013be64935f1dca42453067537c8da
2019 F20101108_AACCKA jagannathan_s_Page_023.txt
1d6f558be4ac1c83a0eb5ec41a735862
03d73673a73817c3c2597afa391dfaec78da8931
3188 F20101108_AACCJN jagannathan_s_Page_005.txt
be30fcc8436c56776aaf4322ed7f072b
9436751534d528af237775bc0c094628ccd29cd6
1792 F20101108_AACCIY jagannathan_s_Page_003thm.jpg
63d9e2b715ea645f51d102e6e8c16ee1
8ff794522bbb93dcae18ea059b001884fc492cc5
F20101108_AACCKB jagannathan_s_Page_069.tif
63084bf57c952746548c6af5f27e898e
5a823a1a682ec0087e4fa07f4e7edc1b043c8647
26775 F20101108_AACCJO jagannathan_s_Page_118.QC.jpg
34ae295dd4d98199a5b2c48ac47d041b
5cc2b49ee69e3d9cd2badaf84597ea2c97971aa7
92739 F20101108_AACCIZ jagannathan_s_Page_013.jp2
1f0f7ac1d37ea36115408fc94efc8c4a
16a12798c512c777e59e3e49a6f8aec9c6f219c2
21982 F20101108_AACCKC jagannathan_s_Page_014.QC.jpg
3736e04908234ccdea8033accfdc5557
994f19a385761dd55ccc180fa7e63b10a49178b1
3747 F20101108_AACCJP jagannathan_s_Page_073thm.jpg
ff5be141995f57ff313553b6ca1dadd9
af81fe4e2d14395b79f08fbec25ce411511d2b63
23423 F20101108_AACCKD jagannathan_s_Page_061.QC.jpg
ac18943e7a2d9202b51e2157130303e1
6d00f8f332c743d8900ff24b0646ae4e0aa4f719
967 F20101108_AACCJQ jagannathan_s_Page_094.txt
6fab0b0bb8e5ad58199d3de06ed806a1
568f44655da14b9d7d864221ff6003f7bb3f4fd6
544542 F20101108_AACCKE jagannathan_s_Page_106.jp2
1df84a629e068e9c3d14d50109315956
0398083f5588c0411407752d3b68772ddde24ecd
13830 F20101108_AACCJR jagannathan_s_Page_099.QC.jpg
d190900dd686759e8cfba156457c9e08
6a61da163c186eeb74ee4a4a488574fe5cc7266e
F20101108_AACCKF jagannathan_s_Page_041.tif
0543f8c70bb47ed44ed375a424851aaf
ff53b7f7fcd8ee194fab22fd7423a235450b405d
499 F20101108_AACCJS jagannathan_s_Page_078.txt
b7154066bae2c86826dde428a09184dd
38e0e89ec58debfa4a154be23b479b291748b983
17730 F20101108_AACCKG jagannathan_s_Page_102.pro
872a95ed25d6ce96579c3f074cce2397
308fce1fdef12d7ec65c25a8a5ea15c727153dbf
1995 F20101108_AACCJT jagannathan_s_Page_053.txt
7db6b8672d5380bcc04dbd8cdf9ab282
7423e092c058ff9dc2d2eb70244b7fa1707fb245
184371 F20101108_AACCKH UFE0021840_00001.xml FULL
b1e46ca90ce7b39ce0b7d9c3e233fe07
43abcc21afbdb578ff575ee1a86c0fa071d2fb7c
105994 F20101108_AACCJU jagannathan_s_Page_065.jp2
74d179e49645e8fc382515a480174dad
a1d179d4bcbfcea1169bc5848e62274d64274b3d
F20101108_AACCJV jagannathan_s_Page_100.tif
cbd7edcec5ad31d7681d418d638a381e
f7a9b11fb6661930174ddf82475d25849383f22e
25835 F20101108_AACCKK jagannathan_s_Page_001.jpg
223b4129d8bf39c5bf9902dbedcd4289
5b3db08e98e94f032e975febe7c167e5831acee1
74507 F20101108_AACCJW jagannathan_s_Page_060.jpg
2c7ce5640a7f169f08527274626fea11
d922cfba3a7bb716922721c895aa97874fe30c03
10161 F20101108_AACCKL jagannathan_s_Page_002.jpg
cd1f7c9b1ac5a946bc172cad9fdeed71
8b3689f21a147b26b5753f12fa2f7a98d5f3d451
3986 F20101108_AACCJX jagannathan_s_Page_070thm.jpg
aba7776f1717aeaf09e966df44e32d21
e9a58bd6674f235e56d6705f69bf54a64d8e1809
78818 F20101108_AACCLA jagannathan_s_Page_024.jpg
be6ffa8e12bad18ff861604227a9a2a1
9f5c429158c1f14b1f2b2a1e8439f507c468e617
17224 F20101108_AACCKM jagannathan_s_Page_003.jpg
d2fa212e6467c54d02354f2d3e20efdc
6689dc7c8f9195af6dcbbac9e379f599f3b866e3
6596 F20101108_AACCJY jagannathan_s_Page_038thm.jpg
cfc969f766e64cb95150833e6cf40137
c85fd0600f4cb4edaa6aac7203458b9bc6767c96
77922 F20101108_AACCLB jagannathan_s_Page_026.jpg
66b4c06c38c5682d3eee020047442af4
126d8e1fea792f4ab756130c81b99a7d478c9245
63016 F20101108_AACCKN jagannathan_s_Page_004.jpg
945c19caa05cfa6f6311dd8fe1844c32
40f2efe7dda6050fc54bb8b3c33939b93b6717fd
115554 F20101108_AACCJZ jagannathan_s_Page_027.jp2
69bab053ade2de75dcd52fd11b181713
7edaf42491e9c3d7a2f18b9358a4b12db8085174
73192 F20101108_AACCLC jagannathan_s_Page_029.jpg
0b12495787f86f4bd6fb54b181bdf003
f7902748b44184bf4288552d280ff947d458bf6b
82881 F20101108_AACCKO jagannathan_s_Page_005.jpg
576513f3b3e6d9e0089baf545723ad84
80bd0db25c63f9c0123354eb40069361095fa764
76143 F20101108_AACCLD jagannathan_s_Page_030.jpg
b6c48de0afeb9a19a7258f9d7dcaa900
b614a84fad434493b4e77d6a2ad72925dac91128
39204 F20101108_AACCKP jagannathan_s_Page_007.jpg
46c1a209f0ab1d9cf620c26872526c5f
5fe573f6f58b3eb6bdc488dd7f022fdbd6da8bfe
32547 F20101108_AACCLE jagannathan_s_Page_033.jpg
7c515272559fe79596cde2c2edf346f7
c184c6173bc765ffea78e065049adb762c6fb898
80014 F20101108_AACCKQ jagannathan_s_Page_008.jpg
dbbccc5a72cce482e602e4ef37b55cf0
1ff6275d1b4885b8760c21effdbccf1081c7f74d
46982 F20101108_AACCLF jagannathan_s_Page_034.jpg
4058abfe701c2e6ffabcab748892ea23
cde7bb2870676897b40e53b3927f2b51d8b02a8a
70489 F20101108_AACCKR jagannathan_s_Page_009.jpg
b2c68e8621b23aeb9564a1101c91b8ce
62fc03cb3c0f6728aabeacdbae694561b7e798c6
78237 F20101108_AACCLG jagannathan_s_Page_036.jpg
ec7d350e6d6bac92529854535307c751
c7330a03d8232012c60561e099276a23d8582a48
49344 F20101108_AACCKS jagannathan_s_Page_010.jpg
c76ce2508232b0daabcde1d2ac3c8370
bb8249a54b54451df643853814808b859d03e601
70964 F20101108_AACCLH jagannathan_s_Page_037.jpg
878261025fcd00bacccc7343f9b1f6a7
8a15a5582e4beb7e74251ca86e3dd2d226b40b6c
78270 F20101108_AACCKT jagannathan_s_Page_012.jpg
d9624fd5f8e26d0ec0bda8fbf9e717ff
b3f96c6b1cac3086074d1cb8ed0a5d60cffc7154
77267 F20101108_AACCLI jagannathan_s_Page_039.jpg
01929eb1acd758e049c5ad96e5502196
11b30e5bd0a18bc6863d2b565e6aea3ea3aa274c
69743 F20101108_AACCKU jagannathan_s_Page_014.jpg
0173459026dbd862fb421e8a2be91d44
d32292a47cacedf84706613b00d2fa8cadb57cd8
63241 F20101108_AACCLJ jagannathan_s_Page_041.jpg
95b81abc477bef6e2fc61cc028378d13
7595a0f2093b1ba252ee52eba0f420ca2efe780c
76804 F20101108_AACCKV jagannathan_s_Page_017.jpg
5a344f955043d648d37fa64c90175f08
97104ddbeec5ffb435786a470bc1b98aeaf46fa0
73875 F20101108_AACCLK jagannathan_s_Page_042.jpg
180d4c2781fd8bfd8ce1fe0549f21b83
d81b8edb2ac09b1dc0fb4b1dffaaab354edc093a
68231 F20101108_AACCLL jagannathan_s_Page_044.jpg
04a8092c7311a869f6704e3381ee091b
dc0bc2cce8a5d8cd7f93c16a5c0b6e13d62c0e8d
81268 F20101108_AACCKW jagannathan_s_Page_018.jpg
9ffe3cde22bfb54551ad951512f51871
02593c632d581e2b4e49b0449a3ea81084f475ac
72576 F20101108_AACCMA jagannathan_s_Page_064.jpg
928bd302612b5039a8c5a3c2a3d7ad98
7ab789e69b8bbf190fadacff2d57951a23fa3fdb
76421 F20101108_AACCLM jagannathan_s_Page_046.jpg
b8c953972320829b0d37009e65ae27d4
7b3825f37f6e60eaa33e40e7af119767236ae0bb
73251 F20101108_AACCKX jagannathan_s_Page_019.jpg
3ace5bdb8e54004c9e9e5c09a61bc9b4
3f1f669a50c779c13145091a57b46ee0a6b51e5b
71450 F20101108_AACCMB jagannathan_s_Page_065.jpg
5a951ff0ec93a731e12f89d3f2293803
f463d9f765d381e325ecace0bf3f965b80fed2e7
77359 F20101108_AACCLN jagannathan_s_Page_047.jpg
2cc5be730ec67bf49937740481e48ed9
35d21ca1f1de8961161a18601588eba56ebc2b87
77962 F20101108_AACCKY jagannathan_s_Page_020.jpg
274f13202b25d6ff863d2116d3e53bd6
755af9632170605d46d4da9bca6ec3b1ae956ffe
56439 F20101108_AACCMC jagannathan_s_Page_067.jpg
2612e3b146d5977a3013f2d940ea2ae4
b82a1b79a07ccb4455ab6becfb5b760a4815cb71
73439 F20101108_AACCLO jagannathan_s_Page_048.jpg
705fb227312f1ef867d3283a88ec9934
dd84bb1d6a775c891676b8420a426d0813fa073c
76942 F20101108_AACCKZ jagannathan_s_Page_021.jpg
5406a64f56caaa5aa6d5f472c94d41db
b1cdc638835a6f166965b9444522ec5b656309a3
56178 F20101108_AACCMD jagannathan_s_Page_068.jpg
f771b5e89c5e4fbc0a9e02fce03787f4
0ed61b56ea8b584f26bfdffd4907e4fe5d85d2ae
59669 F20101108_AACCLP jagannathan_s_Page_051.jpg
3ea208ac4b4a92eb22360c6ffd276ec2
01f6f9e28fab366d806d7cc827437b253cc6ba91
42408 F20101108_AACCME jagannathan_s_Page_071.jpg
e1c3ef79812b9fc4f51108d20d430002
36c6a1749315aba3d02521f13cc77eba19f87dc9
30297 F20101108_AACCLQ jagannathan_s_Page_052.jpg
387340ee42459bdc1d293d3b817d847a
4b62c768b4190dc2f2e44383771564451d75551c
41906 F20101108_AACCMF jagannathan_s_Page_072.jpg
2704367c350316d9a2a6aeb52fb80810
bbbbc93f7cb7f23b719ef66d37c786213d3ac873
69964 F20101108_AACCLR jagannathan_s_Page_053.jpg
1ba522dc431815783a119c04595d7f57
29465bc1f6841f963d9e3d4ad9caef6ce0967a45
40628 F20101108_AACCMG jagannathan_s_Page_073.jpg
4ccabbd0467ffb1c6eae68b29100bec2
56c71bdd84777e7dc15a2e57287da07491da9822
72565 F20101108_AACCLS jagannathan_s_Page_054.jpg
3aa1f2ce0e0c0f171af9c346abefcf15
bca78da288ddb76d8bec94ba225cd2643360b44c
42200 F20101108_AACCMH jagannathan_s_Page_075.jpg
7c94c7ade927729a9b732a9721ae28f6
0b56a827002f3b0c0f2d230b287f42385f1707f5
71262 F20101108_AACCLT jagannathan_s_Page_055.jpg
94867642c5b05c82f5c638bd94d6d856
96f6e738887561f9709e2ae097d168b808fa03d0
38995 F20101108_AACCMI jagannathan_s_Page_077.jpg
f2b74be4ce16792671156d9e6f1f0bce
4761d7684ef653da3fb7914627fe46bacd609878
68030 F20101108_AACCLU jagannathan_s_Page_056.jpg
d6419b7f1017ebb2c3867f318c5a2b10
0a0fc3bc4abced50fb6d662ce34c5400199e0138
37001 F20101108_AACCMJ jagannathan_s_Page_078.jpg
93c0cd242477e9d19aa82cd01807ba03
cb7c4f022180d2a6afb5ae72453a625b3dea444d
77196 F20101108_AACCLV jagannathan_s_Page_057.jpg
bb647d2349efe22d72747333e799e05c
df18a9ce1246a4556339bc28ee570ff46881a899
72619 F20101108_AACCMK jagannathan_s_Page_080.jpg
d63acb1466563d70f18824075143056d
ac253440210bfe6a70cc338f1e3a13d771fbbc8a
74555 F20101108_AACCLW jagannathan_s_Page_058.jpg
4ea1573b91e77d8a5d010e22dd35622b
e4713ae6b5d4d983acf4dd120cd0595c3bc6bb93
74983 F20101108_AACCML jagannathan_s_Page_082.jpg
64ec872006ed372f3b0192d0b6ede993
bddc63b448b23d2032936abfbdc77a699db76155
42486 F20101108_AACCMM jagannathan_s_Page_084.jpg
8c1ec93f99aad95ed72327e628f24ef5
51debb1caa1427fd068d01161bfcb53fd5b9659f
69546 F20101108_AACCLX jagannathan_s_Page_059.jpg
71214fddbd26ff24b1c4f807796758be
b913b1757120191c86768d94c5dc75b7d272b9ec
39994 F20101108_AACCNA jagannathan_s_Page_101.jpg
05fdc637bd4186e468c97bf915f71760
fbfe6609417f3b029aa22969b805883bdfbe8e27
41894 F20101108_AACCMN jagannathan_s_Page_085.jpg
0f15533b1f563a212ecf6205ecdcfd15
ed0afe10c6e5f1f3e0ff42ab0eb5df7e079b3beb
71353 F20101108_AACCLY jagannathan_s_Page_061.jpg
74eb5b3834e082a997d5ef940c7d4302
1f0ebd8e209e99786aa8191ff71f2f0fa38be570
42873 F20101108_AACCNB jagannathan_s_Page_103.jpg
8b4912c5fa68a19df1504aaa78917787
98a6bbb0f526cb406c3551507810bd771a1b90f1
76005 F20101108_AACCLZ jagannathan_s_Page_063.jpg
e33ea56f7ae57dcd9e9728ecff9a7aa8
b986d9c2674068f9d29cdbefd01cc9d9cd50f6f1
42047 F20101108_AACCNC jagannathan_s_Page_104.jpg
e5eb324acbf0c7ac0714b8aaba1e1a0f
f01480b0c1f7008afdc7193a04a8eefd413d6bfe
40707 F20101108_AACCMO jagannathan_s_Page_086.jpg
d1b566a2cd7422765b63cb5a92f17082
69dca5d33c8b4945b781afb1eee89ba653f77e49
43209 F20101108_AACCND jagannathan_s_Page_106.jpg
2b88229504e60db1713134a54b8069ca
ccfcf157d49c696b30643778ca5f2f3841786bee
42888 F20101108_AACCMP jagannathan_s_Page_087.jpg
b1345f6cad9046a709db8310076d2141
2561ebe905b68248bea207bcf5374b5bfe16b8ff
42214 F20101108_AACCNE jagannathan_s_Page_107.jpg
0d30c449836bc116ec4d94c63c9df1ab
747bc0384dea3cbab13abaa05761ee085799d7b9
41091 F20101108_AACCMQ jagannathan_s_Page_088.jpg
f1436d2e656e22c55e8ee0c03311abaf
3779575715334e527fc62254acd3a5e09c39fd92
42157 F20101108_AACCNF jagannathan_s_Page_108.jpg
49568c2c0dae459bd3b399645882508e
681e08be50a2435a41d77542e83a036c0d0f531f
41442 F20101108_AACCMR jagannathan_s_Page_089.jpg
af065e1ccd2aef2eb9af8423a5bff079
d1b56b59ba875069839f55093d4efbb459b62c9f
41960 F20101108_AACCNG jagannathan_s_Page_109.jpg
c7fc71d23384e169592b485577e74557
90dc287e8fb6456c9239c6eca5be0c1dad33c8fe
40816 F20101108_AACCMS jagannathan_s_Page_090.jpg
7322bb19fc2e4658a82ae6a2a75fd7f9
ab1affd3fbf033d4fd20fab1a8a87c2001e99460
43426 F20101108_AACCNH jagannathan_s_Page_111.jpg
1a4d5dfc3fae9809771c060c1622f6ab
170295cdc1b714d7e5c1c4cc4f34f91d6203f3a1
41810 F20101108_AACCMT jagannathan_s_Page_093.jpg
e63776db93fa5d361956f61bd557ba65
c74ac492da9160338370930446a706dc107a9944
43521 F20101108_AACCNI jagannathan_s_Page_112.jpg
9a94ece42315cf055fea78c7a193c01d
93ebfad978a24926934a3a3d17c59f47002359ea
42970 F20101108_AACCMU jagannathan_s_Page_095.jpg
ace0dc724edef1a59c3d4d07e8b2fe1c
9ce96c286724d617bae177c4684a9aa17493de59
40886 F20101108_AACCNJ jagannathan_s_Page_113.jpg
d8719b852793f9af971e3ba8c2caf160
d1e53c0d621b5fd6195f866bcff17fb3cc64bd38
39783 F20101108_AACCMV jagannathan_s_Page_096.jpg
fc61a39e59b6de8e612d1b959cbd985a
ecc81e873bde0ccc60cabc5e4b06b8f2b96af402
41360 F20101108_AACCNK jagannathan_s_Page_114.jpg
240ea301e1ed98ab1ffe2fd290c74148
b477ce3e394b68f085eecc9e4ecff9a38fadeab0
39558 F20101108_AACCMW jagannathan_s_Page_097.jpg
a8d02daa78eadfc395c84dcc6f8c7669
236cec607e6e22df61f8ff1e041d75b5c7feee4e
91844 F20101108_AACCNL jagannathan_s_Page_115.jpg
bc1812c21cd473977d0cc15797436fe9
9443459df2d180721a91c665d73ce1385fac12c1
40428 F20101108_AACCMX jagannathan_s_Page_098.jpg
0f576c2fe288650792539e9350cd468d
9125fef8ddeeb209bafa3a0376863c5f3f7a613f
113485 F20101108_AACCOA jagannathan_s_Page_017.jp2
dcd4bd6d8aa53f0a7636b9f5ef2e95ab
675cbba90a1d916e09f43e66ad1844687208df58
96901 F20101108_AACCNM jagannathan_s_Page_116.jpg
b68bf2515773849cd07540ccafa66114
02db620e72511b57b5b624bc6f36bb50f48d3c6b
121265 F20101108_AACCOB jagannathan_s_Page_018.jp2
1c4355733c025cf8d41fae39fa900136
18c9d3919a5bddc06b895c5522da3a525caa46a5
93551 F20101108_AACCNN jagannathan_s_Page_117.jpg
08b8c2484d6572b725bccdd782785ba3
3f2095dfec9a7959147620638573a148f8b8763e
45549 F20101108_AACCMY jagannathan_s_Page_099.jpg
903ba0970c4c4d4302fae32f30782d8b
97a6bde8a558ffb0780877b3baa61dc192f6a466
111418 F20101108_AACCOC jagannathan_s_Page_019.jp2
92b4cfc6f42b1434f9a58dcd2e327779
476e910b4da9c506f3ed3d516264524078d38d3e
71993 F20101108_AACCNO jagannathan_s_Page_119.jpg
28da3274d56c56270bb7d2bc271a6660
0716b26651b29e83b9fc3120f0d039b0c0d954b5
42297 F20101108_AACCMZ jagannathan_s_Page_100.jpg
d36cb33c213ef75c39e2b777b476b3c9
9f169efcfd2a247c7c3db8bea5f7997e987ca42c
117478 F20101108_AACCOD jagannathan_s_Page_020.jp2
d75c0ec51d5ead3ed5e2d6f02b1a2c00
0ad667fc429773c1d00a5b5d8cb0bd51d833a7f3
27057 F20101108_AACCNP jagannathan_s_Page_001.jp2
8b0795edbe23f03802704984ced7c0d2
6200f902547db0b82a282246039d0940ff57958b
115889 F20101108_AACCOE jagannathan_s_Page_021.jp2
5f9b426fe0b1fc37517e7d6808236f6c
369f0d60daae68eeac81f6b46edcb83183bc171a
5843 F20101108_AACCNQ jagannathan_s_Page_002.jp2
30594a15bce3f5b9b06ddb0b5f292864
e400ae01fa146df05a923fd2a4f5040f2d1cda05
112301 F20101108_AACCOF jagannathan_s_Page_023.jp2
ae63912cc6831a309c6e9c7bd7351915
44c42caa1669bc06a5fd5f8fec773ba38d4c0686
17033 F20101108_AACCNR jagannathan_s_Page_003.jp2
81b474f6ab18c4763aa6e3dad02853cd
5a1d8f9c1a7427e04c0c1644bb3e40db9e9823dd
120073 F20101108_AACCOG jagannathan_s_Page_026.jp2
b5826f3e08708b9474508fd3f09d7c88
4b78b4cf151f16a1492b7235a254baae4edeb2c4
91269 F20101108_AACCNS jagannathan_s_Page_004.jp2
9a84b64c44d63f9b481ea028bf58ed6c
067af12e6df197f5fbe95389910decc9962e5ee8
119857 F20101108_AACCOH jagannathan_s_Page_028.jp2
dc273146cbec40f43e3a449f2e0bbb69
eeff925d79d200ee882472ddf8a66c9a3c58a458
1051967 F20101108_AACCNT jagannathan_s_Page_006.jp2
15b455a7459da90e5a5357fd2c9f2d3c
9b5554afd3013921882ba2efcb4d1e5ff10acdec
110918 F20101108_AACCOI jagannathan_s_Page_029.jp2
e7254c1006f5b4c5e7e48667489cb48c
fbb08ae3bb6e1211838fbf31969a4bf8ab8a5b94
934707 F20101108_AACCNU jagannathan_s_Page_007.jp2
c073dee9139137c7c48a1f3a0246d96c
c57f1b2531367e8d2d95114617b5eb2598c81c23
114168 F20101108_AACCOJ jagannathan_s_Page_030.jp2
29fedab2c94ece8fee248510fde3adf7
236c43d5c69f073bd1a3709279e2306ded0cb6b8
1051985 F20101108_AACCNV jagannathan_s_Page_008.jp2
de24ad4ec666f8c5faff023475bafb67
8902f65feedc83be0b6fdd824d8917c097f72715
105207 F20101108_AACCOK jagannathan_s_Page_031.jp2
073def5b7a27f300cea2e1396f311b9e
2f5aa23de62136a9ab680509555f16178f6b46ac
1051984 F20101108_AACCNW jagannathan_s_Page_009.jp2
e5572378c93c9067a03a50b7bc620037
07e36b39ed3c7708a2564583859ae1fc10d91a66
571275 F20101108_AACCOL jagannathan_s_Page_034.jp2
be239a6e33d3c3c15ed8085018e9100c
c5caed0361d91947e18cb22ade010f4d18201328
1051978 F20101108_AACCNX jagannathan_s_Page_010.jp2
d9eeb23ddbad6317cd6617e3542ffb4b
ec03f6813cbe150d5d3a32bc5c54690d0725b1c5
253314 F20101108_AACCOM jagannathan_s_Page_035.jp2
99f46008ac6b34e2cc073fd26d7b43fa
aacbb13f4e13573cb6c27c298390acf7d92b2720
103591 F20101108_AACCNY jagannathan_s_Page_011.jp2
27a772422edefb3761082d2d751e0fb3
ed7daa9baabfda95d04d2ad373a87759ad722830
112569 F20101108_AACCPA jagannathan_s_Page_058.jp2
1ddf5e1d48ea96d5f3c350ecd880eef7
0449937fcfa4b8c27c4e2e48f8e49d09950c1222
114775 F20101108_AACCON jagannathan_s_Page_036.jp2
8e92dd562112a987b1b1a32e94143375
c74154bb4934049ed3b0023a0a1e8fa8e98a8e2f
103737 F20101108_AACCPB jagannathan_s_Page_059.jp2
f61b432c3e82418bd4fd55ff2a4d6421
ad045949a3d571bde408929640ffb62ba9da6350
106114 F20101108_AACCOO jagannathan_s_Page_038.jp2
3ee46196b913f4d4ce43c87e5fbecc06
8cc2a9a15928861d2a8b0e8201ab5c114a36281c
104016 F20101108_AACCNZ jagannathan_s_Page_014.jp2
1b64eb10f3cf67f192614d0987509414
a3e4600ef11d1457221920f80e7da5073da80eb1
105069 F20101108_AACCPC jagannathan_s_Page_061.jp2
917eea600134225f1a5779ab0e46fab1
2e03c192092abe75c6f6448fb0153aea661903b4
115791 F20101108_AACCOP jagannathan_s_Page_039.jp2
ecd6febf3c6def962623d2ffea4d4fb7
f2a798f6f1064b5cb48dbc89821c94a7c7bda4bc
116866 F20101108_AACCPD jagannathan_s_Page_062.jp2
738ffce1cb18333d33298a164832fad3
36732dfe7bce4472e53697eb245f39c4542b27d2
109253 F20101108_AACCOQ jagannathan_s_Page_040.jp2
2df5e7adb2c7c6e38e281a31d0f9b9c4
2c7ae08776157a5c676b82a1882c30220062ea73
114338 F20101108_AACCPE jagannathan_s_Page_063.jp2
70408dba886c366c3ea5b3955d1c2f5e
10b94b2a730e8b9ea49bf39006e6c173df68eeab
95623 F20101108_AACCOR jagannathan_s_Page_041.jp2
d60aa8b2553adaa8bc96835bd293d172
636ec542e0c488ab05341ad64c0514e42dc2ff9e
110188 F20101108_AACCPF jagannathan_s_Page_064.jp2
3ac8c69fa13d9609649e8380536dab45
13ca692fbe768cc42a0b8d552abb958ab06ba2dd
110405 F20101108_AACCOS jagannathan_s_Page_042.jp2
7e2344adc9927f0f935f4635b5aaaa42
4ebfb1b2fa7ca6387da4b5cebb8ca8a005d6fc7a
90435 F20101108_AACCPG jagannathan_s_Page_066.jp2
a884ad82aa4fe2f157ff7a86feaf7a83
148ca0f8a69a44c2a76b843f4644be3a809f6aba
115570 F20101108_AACCOT jagannathan_s_Page_046.jp2
e74013c47795f123f37d7e1e1b4caf52
1cde7c94ee4cf510b45371885e4e626e85622cde
59870 F20101108_AACCPH jagannathan_s_Page_067.jp2
7839f2edcf5ecc2cb928453d1ce512c2
7c04ee3fec983fee78fc2b7ee607ac8c5edd1dc1
114639 F20101108_AACCOU jagannathan_s_Page_047.jp2
25024ce88711a98d53d73ad1dc4e8341
ba1047e9c92e4297b27473ce491db44a67670c03
58319 F20101108_AACCPI jagannathan_s_Page_068.jp2
237c9090201d8db51e0c78f9966e29b5
34ca98a2bea80bb7ff7ee097930dd2faed4564c1
87546 F20101108_AACCOV jagannathan_s_Page_051.jp2
aa98819bfe14686269fc9b5132c6fcfa
3ce84b3681d4c4ce229a822017942507561e888b
583513 F20101108_AACCPJ jagannathan_s_Page_069.jp2
6533f361ded21a346147c659d52fc88e
f8ba82d554c5a4ff291b3e7fda4c8bffe8e5844f
38567 F20101108_AACCOW jagannathan_s_Page_052.jp2
49fefe78a7844c9de76a3dd5e3ae4360
35a30eba9f658a393c24aa059f621f95808b312b
525711 F20101108_AACCPK jagannathan_s_Page_070.jp2
32af0dced9f29742878140d03f596402
4df94f9343478e5c04e98dfbb3882246572c58db
104096 F20101108_AACCOX jagannathan_s_Page_053.jp2
7ca00012812b6a2452a1c5c61f4a6648
89e7b7b784d8dfc49450ba2e0fda4b17133d1344
569198 F20101108_AACCPL jagannathan_s_Page_071.jp2
ccea40164c7cb2791d14fa62e2909d6c
c35365a6595662e1e54d06c76769c68865f6e97b
106698 F20101108_AACCOY jagannathan_s_Page_055.jp2
71816527285d9fb133132a4681cd67cc
ef65341a9fc6192c023a0909093a47786cb2c96c
511393 F20101108_AACCQA jagannathan_s_Page_091.jp2
5b259c423835df4bffc2052d13804029
5149c4470bc05c1abedbbffe605de2507aae4284
573936 F20101108_AACCPM jagannathan_s_Page_072.jp2
afd29741413a7e9fb7cef75975d9b1f0
96c14e9cda166083f38fcfd1162f5085b2627842
104207 F20101108_AACCOZ jagannathan_s_Page_056.jp2
db09bc63712ffd8d23316f3721cb4823
1dad3bd1ecb329b7df5768e2e12315e9a2e6979f
540073 F20101108_AACCQB jagannathan_s_Page_092.jp2
5b4d3740cd33c4614f7006948d178ddc
382ba3a9e135159bb020649b10d804a5e54e49e9
553064 F20101108_AACCPN jagannathan_s_Page_074.jp2
9bd6b3234db828c7330fb1c91d5911e7
17d09f6173dbb10ef6a82e2d86771256058f7216
563015 F20101108_AACCQC jagannathan_s_Page_095.jp2
b6bf28b5826daba4874340f3818d8486
b0e358c018896cdfefe422c999fd36eba9d69cb7
568669 F20101108_AACCPO jagannathan_s_Page_075.jp2
54576954f4cb964a6d0ee87c2fd9c56d
19c3a473524a89ccb49343ef2d18db708430f300
554503 F20101108_AACCQD jagannathan_s_Page_096.jp2
d23b940a941137b0a4ad9467de3ae41d
e202589f7224076f362cd85af864e1929fddcb9e
504920 F20101108_AACCPP jagannathan_s_Page_077.jp2
b8e0dea6012b5808b444de520f422162
4b903a663f56ee560d820818b669793c0928f9fa
518336 F20101108_AACCQE jagannathan_s_Page_097.jp2
7251b3c3a4c6b28edcd1470bd4d6674c
8d97502516b60a7a8daca889cdaf87b29de43c3b
438906 F20101108_AACCPQ jagannathan_s_Page_078.jp2
04048f6b25d6bd1adc88cdcd6c815b58
8bc3d2b00abbd821f41940513dff361ca910f205
540542 F20101108_AACCQF jagannathan_s_Page_098.jp2
66c08f785e14e13dbad95c1fa0f75343
638c0f7996e17c7d513895f5b07287fc41c95045
108341 F20101108_AACCPR jagannathan_s_Page_080.jp2
f2e8820305efd5afa8ded10a86524ea8
c0aa5f734f9a0ce3651382da1d76fab9404d1fa4
543376 F20101108_AACCQG jagannathan_s_Page_100.jp2
79db31a1d5043f20f0c5a9f0e881e7ed
831498849184ffa072b01c663aabe2f1dc65190a
120549 F20101108_AACCPS jagannathan_s_Page_081.jp2
941e57a9c4bcd716a9f81093c5443261
a46ffc26e592092deff47a8d060a244837f19428
514437 F20101108_AACCQH jagannathan_s_Page_101.jp2
0d0b9f978620a1d66266aeb9b85b7bd4
ba1298bf67710e2126a7c6b1f5c57827d9ce7e5a
110591 F20101108_AACCPT jagannathan_s_Page_082.jp2
a253a3667186fb1312e03e406991c039
13c6d738e8dde6c901789512b3f6abbcb4fac392
539287 F20101108_AACCQI jagannathan_s_Page_103.jp2
2b293876ca52defe7d0b3f6cdaacac2a
0302817c02844f8932226eaf94a8ec5db5430e22
592687 F20101108_AACCPU jagannathan_s_Page_083.jp2
9496050fa3d952dbe445d09d700787d4
10152356eaef6467f13c73c74c8458d96abb36cd
552953 F20101108_AACCQJ jagannathan_s_Page_105.jp2
90d45e6d928d58948a3df77cce9495fe
a4341367eafa50018fd8f30565055611e0e99156
568608 F20101108_AACCPV jagannathan_s_Page_084.jp2
09db86cfe775b32f6042ca4327776833
f1008995cd9572b7e0f07cefb645bd48177ea324
516262 F20101108_AACCQK jagannathan_s_Page_107.jp2
5925a24f3ae53d72e4372f78e6ed99f0
444e2f24d33e3476c1c6f1edd2ea27aee54c0e8d
551933 F20101108_AACCPW jagannathan_s_Page_085.jp2
c951536ae71dbcc26e5babbf94e00721
b2ff068303543bcbbbaca79a7d794bc148e4185e
530714 F20101108_AACCQL jagannathan_s_Page_108.jp2
22cc0e120b08344bddbcb476817da663
f8d495ba83ae8a0f82b67c2a9ea5b63a664bfe26
543053 F20101108_AACCPX jagannathan_s_Page_087.jp2
f5b851c5e35f5862f57d14f173e0c70e
49bed728c4a22d223618a0d70a22233110534fce
F20101108_AACCRA jagannathan_s_Page_007.tif
ad625cdb986e07f4f6a4047a162cd77f
2a28cbf22555b2f16156491123d18fdab206d70a
550948 F20101108_AACCQM jagannathan_s_Page_109.jp2
d6668fccbbe8df4a61b47b53d494d914
c9f61851dcfe27fedbf612f4249d05261b6f6559
545646 F20101108_AACCPY jagannathan_s_Page_089.jp2
87cc7aa9bbb023e269d87943452dc4b1
75771809e0f99c281e8d4a289a3099ea195b7107
F20101108_AACCRB jagannathan_s_Page_011.tif
7b95c4b9110d6c417336a02382bccd50
7741cd1737036dcfc44ee6e3fb4dc07644332c61
530091 F20101108_AACCQN jagannathan_s_Page_110.jp2
50febae014e43e8ee9f22ed918dcd3bf
0e95717b1c92c81ebb5ccf7d0b1671550408864b
513437 F20101108_AACCPZ jagannathan_s_Page_090.jp2
3604ffb0d710ee8016ecacf37704bc82
ee6f7c6b90424f2110dced3f1cdcdad2c10f4b3a
F20101108_AACCRC jagannathan_s_Page_012.tif
04bbd42709733acbe27fbcfc9774dc01
1ec655b7d6c9e5799cf6cba768ff1cf5b295831d
568284 F20101108_AACCQO jagannathan_s_Page_111.jp2
fffd3788dd3e070957eca7e7a4cccb53
f4f3f1e3381f46468b63e5b137240434ef0d5bd5
F20101108_AACCRD jagannathan_s_Page_013.tif
af7ed9e77720f303ab6978e82bc7d7a3
a33034ae3b8016b73491619a46a3732698cf5768
531516 F20101108_AACCQP jagannathan_s_Page_112.jp2
e2a644896365d738ea9a8241642506a0
44ec6810ca7229fc2abbfd2b72923be43b3e4738
F20101108_AACCRE jagannathan_s_Page_014.tif
1904800ca8bc37337a5b64eb866ba990
58e33b7626984b4456cf24df844d93bb27df858b
538098 F20101108_AACCQQ jagannathan_s_Page_113.jp2
3d4daaea143f54ba8625625a530126b2
1e3303519c3bab71d5b62a2c70c55c6fe4dacb72
F20101108_AACCRF jagannathan_s_Page_015.tif
fc8ced60454f97cd68eb80f1a7d39a2a
0b5c47bf1d629890b4091a617646bca1f60043c0
539179 F20101108_AACCQR jagannathan_s_Page_114.jp2
8e40b782bb53b7b67b138b90ec2be1b2
cb7bf014093779f2d405c26950c88261b357a832
F20101108_AACCRG jagannathan_s_Page_016.tif
17e4308ce62262a9ae941a2fbe4a40d3
6d958c07eee4167c003da7578993e879b1302f70
135848 F20101108_AACCQS jagannathan_s_Page_116.jp2
01bec52f9c139c13ea2e68f113738dfd
f2172b17473aaf60a019d6ee7d0172d721803b08
F20101108_AACCRH jagannathan_s_Page_018.tif
8fe39ba516b80f5d102dc0b670eef139
53f233e4a630e4055c30714bdabd277994727e5e
137215 F20101108_AACCQT jagannathan_s_Page_117.jp2
f702b4d54d02f42aae5fa2f3f1fd387d
9cdcc38084d8c2495f618a822842bc493beb7dcd
F20101108_AACCRI jagannathan_s_Page_019.tif
fb359c628d2fb3614b9577df6dd33085
70deb063135ab69ddc05f2c355cc391478864c9d
112853 F20101108_AACCQU jagannathan_s_Page_119.jp2
c4c78665fd7454a1bc6eb5f1fe1f1418
d475baa119a03bab8a5c5be16420ac1c3afc8bed
F20101108_AACCRJ jagannathan_s_Page_020.tif
d19b443b2477b3116cbba83d7e4fc8c5
43289a68c56a964f20019f5b60969ff997420f50
29984 F20101108_AACCQV jagannathan_s_Page_120.jp2
63628a2482603e35e0f72bd03e9f57a9
8481513a19ca1da8bf3a2b8131fdb5f9db727f97
F20101108_AACCRK jagannathan_s_Page_021.tif
41abe996ad5b1e155dfa05b16350bf83
4b97c9558427d8bded5ff46f9c640b7677677861
F20101108_AACCQW jagannathan_s_Page_001.tif
6dda2901f6fdc7997a28810123c49d16
7c59fd0453b2f39a826d249cf07bf0fcb82ac544
F20101108_AACCQX jagannathan_s_Page_002.tif
123bee62f47aed82b295f9b2c60cdcbc
ab27b49cf1949c7b9f96233bde1267ce92410c82
F20101108_AACCRL jagannathan_s_Page_022.tif
58ded58ec415444cfcd36759b9b3711c
597ed3fddb62a485b223d2d4b702513d70999536
F20101108_AACCQY jagannathan_s_Page_004.tif
ba89ab5ca4e1b386810c9461ed0469cf
55b510675828f1e19e6181d6a7480f2719be557b
F20101108_AACCSA jagannathan_s_Page_046.tif
9f11e8852ffd51c899c38290883de95b
a04f123e6860b870f6aaeb78054912c69a69d8f1
F20101108_AACCRM jagannathan_s_Page_023.tif
2c00e2f22ddc02e371efe8a234c33cda
23ff02dbbdf45bff0e7e8a6d5835f34c8d925219
F20101108_AACCQZ jagannathan_s_Page_005.tif
4400db3a6ad3a5be78fba51a70d43eff
e51520d9e9281781ad1b7cd8c4fe9211aacdbc0f
F20101108_AACCSB jagannathan_s_Page_047.tif
2401271e57562cc7ea47d904f3417c68
ffd2ecb1c3e1b73d3f1add81fbf523f39d4b0348
F20101108_AACCRN jagannathan_s_Page_024.tif
4f2b0b2c2492b2d2d38a770eac2571a5
e90ae6087256bc3d826ef7d3d721cf8109a566b8
F20101108_AACCSC jagannathan_s_Page_051.tif
de543db42f110ed7a471ae1f49307a54
7574cf567d11f35a819039f5061f8caebd926abe
F20101108_AACCRO jagannathan_s_Page_025.tif
ee2782221aaaa5a743e3233d7583feae
b993f1b899e841e4c095cb645da7f9619c89b20f
F20101108_AACCSD jagannathan_s_Page_053.tif
aa352e63bebad2916f62c8d2fd2ebb94
b38fd8c164c0eda306301437ba8edb345655c6af
F20101108_AACCRP jagannathan_s_Page_027.tif
e064373ec181be8fa01bdc3e00044fad
95acae9ad1add0dfc172fb29a3e68417135400b7
F20101108_AACCSE jagannathan_s_Page_055.tif
506217e910595e56c51f0c7bb5d78653
b84b89d5854444f4b253d494611815ed48b8c827
F20101108_AACCRQ jagannathan_s_Page_029.tif
b7ac34396ec582e2ee457431a9dfa986
ae522f80e5638bba10c1cc4d7e5912a32652d6cd
F20101108_AACCSF jagannathan_s_Page_056.tif
7f59746fa4a69d7059b3bd3c549f3222
3b52bab43c79fd08afc0b624be62c80e10901ef4
F20101108_AACCRR jagannathan_s_Page_030.tif
907ca2d4a17e5c2dbab0e7ff6012d2c8
9a225c90479756366871e9bccf167907499d3ebc
F20101108_AACCSG jagannathan_s_Page_059.tif
0a389ba41a18400e208b0b5388ef380e
bf7c4bd4cd320286cfcb412bbf1232948d3622f7
F20101108_AACCRS jagannathan_s_Page_032.tif
4d48cb5e4e451107cc12de8c0f5da436
50b3f203fa4efabc00a30d0949fb77f7240da803
F20101108_AACCSH jagannathan_s_Page_060.tif
b12241747d221df5f03c4bf3ad071ad1
eeeb7774b02be05e256bea1a664509200dfbeea8
F20101108_AACCRT jagannathan_s_Page_034.tif
d13b1fb55935fcdf13b932eb256799b8
87690ba7bc615f5080996535d4be978321904a1b
F20101108_AACCSI jagannathan_s_Page_063.tif
caf01370982d139b7a47fb50d2918be0
98ba390d9d5943c2ad67bbfc4cb2675e95516a78
F20101108_AACCRU jagannathan_s_Page_035.tif
8c622b4fcca10fae498fd91b396dd184
5d574ba971de53edac2b1b67e1fbf52f703f19fb
F20101108_AACCSJ jagannathan_s_Page_065.tif
abba6e3c5137535ef6a6e19196bd9c53
0e332d4e89a4deab0fd75b022663132f2a05a12f
F20101108_AACCRV jagannathan_s_Page_036.tif
3e36ed4ba98a267cab49ae7bf9fd169c
782af2fb32eca4266bfde6c8cb5dc5d2087301fa
F20101108_AACCSK jagannathan_s_Page_066.tif
4fbb17b5914856a5a517ccb0330251d0
03e3e783832fada568516d2963e7453585de85e0
F20101108_AACCRW jagannathan_s_Page_037.tif
3633bc97122bc78ad39ea22fd49dcd31
de6c4e15eddc60285e12ff86ed0de318c5eae28f
F20101108_AACCSL jagannathan_s_Page_067.tif
3032b7dc6863eaef9236f0dfaf3ebe03
528a0a8e8316a1a4526f47042e2c80d86ee5b766
F20101108_AACCRX jagannathan_s_Page_038.tif
01e875663cf2574ddebc72cb8222e40a
70561e473e23662466b682e008d2a74e88384ff2
F20101108_AACCTA jagannathan_s_Page_091.tif
1124fef6410851773e0625a549797a82
cc3f447ce465565c5255119f845f0ce3917652e9
F20101108_AACCSM jagannathan_s_Page_068.tif
4a774004bc64ad8100f32c1d73e45685
2bcf2aa2d33651b2ae7a22ca75ca33dee80bc87d
F20101108_AACCRY jagannathan_s_Page_044.tif
206476dd567ed4bf04c08f0cd5747684
ef71d7260c16a72707f8759a2d12dd2363a9e24a
F20101108_AACCTB jagannathan_s_Page_092.tif
529e41bb5cd88e90c42877e66c7f83e8
c8367a85bada1fd3d72b0febed70da6bc8615db1
F20101108_AACCSN jagannathan_s_Page_070.tif
c80bd5e4e58159c52600fc0e4af12546
b7b53c96e6a81b87452430a055bf49a61de4632c
F20101108_AACCRZ jagannathan_s_Page_045.tif
b288ce5030b7466f907c5bbc2ea2e43e
0ac2eff982c1541efd71dcfef2c8903251ead01f
F20101108_AACCTC jagannathan_s_Page_094.tif
9e6834254e2d408301841419c43ba588
f106797ddaae5aac044414487e4777e9b796715a
F20101108_AACCSO jagannathan_s_Page_072.tif
f6b9c9781391a194a33813c9113a5037
0f8c5f73acd49f5563ab99f6c5139d1e621639fb
F20101108_AACCTD jagannathan_s_Page_095.tif
c89f7f2937221dd64765b9a68ba52aa7
ee5ad76e386aea7192f5e23500a96da2b6cc6f6c
F20101108_AACCSP jagannathan_s_Page_073.tif
569095acb5350a49496b9c159e5d0d7f
f562982efe80e6a17e76e86468044f07b15858a7
F20101108_AACCTE jagannathan_s_Page_096.tif
ca25a6a7393cce3471d657b7c2553573
9fec8adcc231a11f06e11c23cd4113bcc93213dc
F20101108_AACCSQ jagannathan_s_Page_075.tif
b44e6f12d67628ac413293341da9187d
a87663c971b379ddfa17fea631c7d6532dedc893
F20101108_AACCTF jagannathan_s_Page_097.tif
92ef6eb98b2316ef676a00472e74f67b
540bdd6a9072c49f5b638256579462784356466d
F20101108_AACCSR jagannathan_s_Page_076.tif
9c1dacc6b8fe9765cc4475760b8977c5
e3796e9ca445b9c62f2e4b543934fcb5fdf83e98
F20101108_AACCTG jagannathan_s_Page_098.tif
2bf0fe37574683fcb6af5105b3710940
f75981bbb110c061ceb63b455659b1136cfdadd6
F20101108_AACCSS jagannathan_s_Page_078.tif
e2524509efb21e6a445fdd4ce2f6aa90
e5c9be9448b72e343011a6c87fb271a4ad156216
F20101108_AACCTH jagannathan_s_Page_099.tif
ac103818564814bdb8c594749ee2e63b
55e6b8ed35641bad043f60e26f2b61b3832994ef
F20101108_AACCST jagannathan_s_Page_079.tif
c01defc2610100e9ca9719b0d495a56f
f3a25082dc79b223c0b199bf696151b9f0e3e6cc
F20101108_AACCTI jagannathan_s_Page_101.tif
6461678af0c911e8fdc2d460dbef85ff
b27a4a691ff522375476e6b145d0aa6eff3a99af
F20101108_AACCSU jagannathan_s_Page_081.tif
cca9277e606084956d62c772a34806e5
806a5b90e7408394140f1aa04b4ade299afae067
F20101108_AACCTJ jagannathan_s_Page_102.tif
669e6469413d828dc861516e727969a4
148d842b82b4fc699b703968988d8d3d208a0c94
F20101108_AACCSV jagannathan_s_Page_083.tif
8bb9e2ae080da07bdd3bc1666ed6c082
502a2635c64f3171c84eb01753ee939d9db3824a
F20101108_AACCTK jagannathan_s_Page_103.tif
98d9ba076f8abff07b79a0fcc183c26d
79b0e646b019a5ae38148eefaa54fbb419860250
F20101108_AACCSW jagannathan_s_Page_085.tif
72a2b4806f2c5b985e7312ddbe7be249
808931c50b5ce1c7081968e29db1b994db463132
F20101108_AACCTL jagannathan_s_Page_104.tif
aabe8a1b39a21d474f17915d0ee98ce7
83ec6ad08b60b702247891f8c7ebbe771dac93a5
F20101108_AACCSX jagannathan_s_Page_088.tif
0dc93cdec214639bb82f852bee35b96c
bdd851b2909ca60b490b671aad0a9c94a779836c
F20101108_AACCTM jagannathan_s_Page_105.tif
b643049cbe808d8e43a5d24f63aaed53
654bdc09b2884847077c3731df30f8399d3adf80
F20101108_AACCSY jagannathan_s_Page_089.tif
33b033d46bbb98cdfbfa9599e2f19d80
fe696655d3d3d1bc2e9d1b39d2387fdda3a0fe65
6923 F20101108_AACCUA jagannathan_s_Page_003.pro
a1dad17554b7021ee90d25e89392e26a
250c557d0e786bcb7eb23777e0791c7bf29f6ed0
F20101108_AACCTN jagannathan_s_Page_106.tif
86dba2894ad0c35ffb57455270448036
7a02e210b81f50eb53de81528d97a69d16af05c5
F20101108_AACCSZ jagannathan_s_Page_090.tif
7181615446577faac5c9749cda0239b0
fb0596fc76738012fc525a3a0da0b67a3a05a72c
59733 F20101108_AACCUB jagannathan_s_Page_006.pro
71c901d5a774deb5753e701dbcb785e0
8c9449619555656d57e85db59ec2755252fa1d83
F20101108_AACCTO jagannathan_s_Page_107.tif
8238d8a6105552ccd18fd54afccf1560
6e17d3658ccaf44ce1aaef19a4ec049837481a33
57946 F20101108_AACCUC jagannathan_s_Page_009.pro
afc79db6e1defccec0cb76c3a2906cfc
b607e836013343ab9f6f89024e5b67ee315a5f28
F20101108_AACCTP jagannathan_s_Page_108.tif
36492880546faa90aeb971287070bad1
04bd2043dadffb9fc42e5ae2de8dc99fa09ba72e
40536 F20101108_AACCUD jagannathan_s_Page_010.pro
addf758448e3b1112afe599f1dfd84e7
3437b31196c2ff326041fb802cbb3b16fc112923
F20101108_AACCTQ jagannathan_s_Page_110.tif
a690168d6b4db7558fcc5c4bdb6167d6
14009a5eb3c89ab4bea8e2d983abb30251cf97c1
46561 F20101108_AACCUE jagannathan_s_Page_011.pro
48806ccece6a2931c146f95ab2c2acea
9069fa47d486ac9c813896744052cf3d78995590
F20101108_AACCTR jagannathan_s_Page_111.tif
e69518634a2495176eff4a2b770d3b2f
bdae369ccd96ecdef1b72b25672b11c499a7858d
510 F20101108_AACDAA jagannathan_s_Page_120.txt
bbc7e16a9db5362a8810b80a01f79004
3f29929564e7dfe7dce1a665314985f0c7510aa0
53959 F20101108_AACCUF jagannathan_s_Page_012.pro
c9c6112586cef91590e9bf56f3df53b0
f66e1a46bbcf3e664c67caaca38d8186578d2241
F20101108_AACCTS jagannathan_s_Page_112.tif
a14d44efef1b535f0fbfdfe9a66fabd0
c0500ee11e0fd6b832efdfb2d0d457450ec51402
3993830 F20101108_AACDAB jagannathan_s.pdf
7cc04b1af0fbd60cd336901134302bf6
7024337c7b73ea48d0a1ac1f3b1b0bded05e5aa8
42758 F20101108_AACCUG jagannathan_s_Page_013.pro
9df7fda5626e19b59d779ea12339036e
77aa82b9df341c5ab9d688b2805e4356db3b3350
F20101108_AACCTT jagannathan_s_Page_113.tif
d4f255faaf475a89c966d7d6c18b145e
1e9bb31e40dc198411713b5444cfd771bc2c1184
2336 F20101108_AACDAC jagannathan_s_Page_001thm.jpg
9b37567464e121051f2f0467d88f8caa
6a886b05cea08f7102b8261c1ef90838b506886e
46775 F20101108_AACCUH jagannathan_s_Page_014.pro
81d5a0783ee48ea85acec4e398a09fd4
d3b3f5fced3f71d97aa47b762bf1d84d1fc098fb
F20101108_AACCTU jagannathan_s_Page_114.tif
6ec227b6800f23421085678167af56ca
885fe82dc1a1feaf1b2e46aba6be940163b82ceb
3131 F20101108_AACDAD jagannathan_s_Page_002.QC.jpg
7ed5c27261a5c4423fb5a17c5cf9a383
b0bde72c496ff403a48e797fd17613ab4714f6cb
55964 F20101108_AACCUI jagannathan_s_Page_015.pro
594647bba22cfdd1dbc7fbf1b3699376
8f680a43472e721214cbd08d573edc77f9f89cc6
F20101108_AACCTV jagannathan_s_Page_116.tif
f95e4a393343b05ddba6f906ce4ff63b
e746776dfc6a0fb7cc92576944f05e9f8a98c8b9
1340 F20101108_AACDAE jagannathan_s_Page_002thm.jpg
d368b0d31642f73a77c4c5c44a491799
c04d83f7e1800b30f43c669f9aa56fc38c7eabb1
56598 F20101108_AACCUJ jagannathan_s_Page_018.pro
680580b35c590e554080c0ef87d55947
2c9171399b861471451c8ea8825b003b1d11b806
F20101108_AACCTW jagannathan_s_Page_117.tif
e49430ad8cbe12925f9c7952efb74ffb
57a4b33bdabb64f3714385495fcd116ea93b4fcb
5245 F20101108_AACDAF jagannathan_s_Page_003.QC.jpg
756b87da1962aba9ff4cd55ffb14ca88
4d1221ee189a8702e94472b1dd0e2bc7ba1dd09c
51478 F20101108_AACCUK jagannathan_s_Page_019.pro
932f981be6499d04fb1d00dd63abcda5
b847e0d0fb5026fe5c502b0c75228483871ccd0e
F20101108_AACCTX jagannathan_s_Page_118.tif
cedda51fb0f4bf6bac05af21a2fccf7f
0ab7da5a9f371ddf76745aa09743f2ddd5fab543
5747 F20101108_AACDAG jagannathan_s_Page_004thm.jpg
2a0559e60ce50b1402dd4adb8ce6f6e3
f4b861447d8c64e8d3abd5062504e3e89cf4f9a9
47783 F20101108_AACCVA jagannathan_s_Page_049.pro
b4836e2270d238d3a4e1a217ca1a351a
4ebdda17b5f4cc8a16e674f1e7b146c3c433ac60
53966 F20101108_AACCUL jagannathan_s_Page_020.pro
9aefdf9c35840b4f696e248f4b66b2bc
2aa7dcfd5ec1988c806f16115a26419b9c817e74
F20101108_AACCTY jagannathan_s_Page_119.tif
1edc1357c29893ba36067cb0170953c0
fe487cd474685f0737e54a961e73039f61e1593c
5819 F20101108_AACDAH jagannathan_s_Page_005thm.jpg
623fcf6733a77bff919a03cb7e37aa92
79083579a369236cae3dc45ea9f2b3e2a94c89e8
52849 F20101108_AACCUM jagannathan_s_Page_022.pro
7a6110c9a194cf4a1808a6d72f667924
1c617130d6d02159785ef42282cc9f7a946adfdb
8567 F20101108_AACCTZ jagannathan_s_Page_001.pro
2fa6688e0ad3bcdbfa3bf03a2d769974
ca681329ffac27a3dc46d151681e702056bfe625
17974 F20101108_AACDAI jagannathan_s_Page_006.QC.jpg
e17e72ab96b321b911b07ed4d0d57701
9298ee1b5c526588e0bde90086e3e5ee2af47073
47814 F20101108_AACCVB jagannathan_s_Page_050.pro
62331777d2b32534d61b0a9d78f5ae8a
524147eb0544331129b3ee6ec828ed6cea7279bc
55644 F20101108_AACCUN jagannathan_s_Page_024.pro
9eb43f0498d36303e3d89eab9d79bb43
af0489339d2f4e2405ab3073e059a4908b9d1f41
4641 F20101108_AACDAJ jagannathan_s_Page_006thm.jpg
d1c20a28034a988787a854c17acde30f
fcbf9bdbcbddd94fcaf26aa8b9a9a9159f7cc0b2
51704 F20101108_AACCVC jagannathan_s_Page_054.pro
5c5ded9cd3d5f478b7daa7a4728ce65e
b5a0ae05ad8d91a1ff445d5258b5dde5b88b4002
55096 F20101108_AACCUO jagannathan_s_Page_027.pro
7d11633e68c06cfda5bdf7832f6f1099
4026a473c6f71eea14439b3342c578893338a84b
10653 F20101108_AACDAK jagannathan_s_Page_007.QC.jpg
1cb4caf3c185a65487e2d8b21629e2a1
957720480f91522172eac0ce8d9e4507be1defb7
49200 F20101108_AACCVD jagannathan_s_Page_055.pro
f1532415d2ba3a70bbdd62de8e196792
e6df9a6082a1d8cb7cbe51d143ecfe257c48eb72
55235 F20101108_AACCUP jagannathan_s_Page_028.pro
f26968880fa7a6a5d3beaefa0b0ecf41
f8aa6799468b5cfa23bb3ec1157621cfbb1cdf81
48012 F20101108_AACCVE jagannathan_s_Page_056.pro
ace3dbf3d73bb1651a57745f90736f4f
6adc262d12f1994d6fdaa46d2dc3e16b19b1fc6e
53325 F20101108_AACCUQ jagannathan_s_Page_030.pro
be70ca70814c0248f0325879dbd7226c
13b77ee89b5492f307cb83ed21eaad7fdc8a05b9
23625 F20101108_AACDAL jagannathan_s_Page_008.QC.jpg
194315f9b22ed0afcfe57c58847fa75f
ae5ff8d9de4c9523eeb64ec758b25f6224446c57
53703 F20101108_AACCVF jagannathan_s_Page_057.pro
ee21da8514164c192830e3864573ada9
db42fd43c38ff892aa143f7528fcb360dba89d52
5222 F20101108_AACCUR jagannathan_s_Page_033.pro
b8fa06b23398d344a3284c86877548eb
93015c4a8077f58735047c81f056cf378c6f1780
7195 F20101108_AACDBA jagannathan_s_Page_018thm.jpg
8a53313224189c5c82453eaf71cd5ec4
60bd0a1a0de57794f6ab899e390f47078aa1c7d8
5961 F20101108_AACDAM jagannathan_s_Page_008thm.jpg
1cf1a660e368ba9db1b29836607a6489
6b20ebd80145388bc25f2cb406e0cf92e7243ec9
51479 F20101108_AACCVG jagannathan_s_Page_058.pro
d90cdaf3300042836bff52e92cbc914a
d5a7a9c3f3d8dd00cb35bd0e86e3ef7b3107b527
6550 F20101108_AACCUS jagannathan_s_Page_034.pro
43da7c1e7337e3a680f345f451c93520
3641f25caaf0a50e78bf67dcff5a31a7049fdbff
24169 F20101108_AACDBB jagannathan_s_Page_019.QC.jpg
b50888152371e172cfca034d7225525b
2e4257103684bd254e9120557dab91235578955e
5899 F20101108_AACDAN jagannathan_s_Page_009thm.jpg
ee5be1644beed54ea08efeb0e2553e47
80cac184af37ef9fd2d33ab076b6e7fc8800380e
51651 F20101108_AACCVH jagannathan_s_Page_060.pro
d48dec1f2917eb5000b00278a040a10c
314f423382810727253fcd5e03e760487403b0f9
46955 F20101108_AACCUT jagannathan_s_Page_037.pro
ac504ed080a972f224cdb5feb198d492
ebede5e4b592e87607e39338d2b682603a636d94
6635 F20101108_AACDBC jagannathan_s_Page_019thm.jpg
9df06d4427ca4aa52b6620f5c6f0e5fd
969838d12862a8c7dd7beca12cfc5ae90c214efc
14047 F20101108_AACDAO jagannathan_s_Page_010.QC.jpg
37a631c312be8f1d9de24674d6a5a334
a3b5bd74783deacf9aae7ecdbb13d99f677cda4f
54200 F20101108_AACCVI jagannathan_s_Page_062.pro
96b0632fec4e1d6c097349ae0b93cd86
a65016ca13c66f176c7a280e074c9e3de0f741c8
49828 F20101108_AACCUU jagannathan_s_Page_040.pro
c2d56cdcef52072f68b84768844ea3c2
b7834398cf93f8b3b2c6a3457934c533a02364da
6829 F20101108_AACDBD jagannathan_s_Page_020thm.jpg
ccab70900806f69b9485998b91fca7ef
671bb8b6d89bbd2a39e2071044f38659f3d7cdb1
21936 F20101108_AACDAP jagannathan_s_Page_011.QC.jpg
be3b8c20b933b99740b0e9da0c70c105
b8b09abdf69fcad2a0f40f80382f6711f625d055
50545 F20101108_AACCVJ jagannathan_s_Page_064.pro
e2e1ef6c2af8922eda005317411b4c46
221589705c073401436b620694b6952908fde876
51364 F20101108_AACCUV jagannathan_s_Page_042.pro
de6accd0fc2b99f5a4731fd1c1c60a21
a120e46b023e588da62c39e69389ba32a7666fcf
6849 F20101108_AACDBE jagannathan_s_Page_021thm.jpg
06261e200da727bcb600e9fa36d92193
c99c949cd0d5aaa7eeb5454b897ae9056c62237f
6235 F20101108_AACDAQ jagannathan_s_Page_011thm.jpg
2763439ef447574ebdb5f338b224d7ca
61541b2c197ff4598bc21db2b65d9f6169a6b32e
49915 F20101108_AACCVK jagannathan_s_Page_065.pro
45d53b050f4c73bfdc9dbea2bce77eea
9af3d367f144819a951843b07571ca8d7f62825b
51742 F20101108_AACCUW jagannathan_s_Page_045.pro
04687a6caf0723c282a6021e874a3e7e
662d2fc18be2dcd46b76393d0d9cb8e90d156dd5
26658 F20101108_AACDBF jagannathan_s_Page_022.QC.jpg
6cdce1aa32f6a997058f8ac737278fa1
0aa052411096985b099aebc74a327e01e12c3b81
15533 F20101108_AACCWA jagannathan_s_Page_083.pro
cad8b04cfb232b4de495ab5c34325f13
73f1eb0b0305afc8998a3d112a55c7da0d25aad7
25627 F20101108_AACDAR jagannathan_s_Page_012.QC.jpg
34a8074b9c3916dc92e80187223d550a
4a3f6ed10450ccfacf9c09dc133fb40e7ed02bef
41455 F20101108_AACCVL jagannathan_s_Page_066.pro
126910287c9a361028d29feeb42c30ee
d92bd1a9992f997a14bfc80221bc9944d70c589d
53823 F20101108_AACCUX jagannathan_s_Page_046.pro
6f0b95713adaae8865ded943b5bdc0e1
da32a0fe5a5fe674aa6f18d481ad3a9c9c28bfdb
6841 F20101108_AACDBG jagannathan_s_Page_023thm.jpg
c0657781f5e006c056001f0cdc9312bc
347039b20570173cf2e068967292d5477815c918
11135 F20101108_AACCWB jagannathan_s_Page_084.pro
aeb37a05868922f2fe687fa1f8afd08d
51be3c6f6d14ef6197b4d3e5ecbd35dddab9cd2f
6881 F20101108_AACDAS jagannathan_s_Page_012thm.jpg
fc209e85731a62fa569d9b5f57d37d98
bcdd358fbffddece3e322b3c2b538bb5c646162f
24044 F20101108_AACCVM jagannathan_s_Page_067.pro
464ea35a80afcb2cbda57daf61ea565b
32ea3c5ec28306e226c9462316752e2fdef8e2a8
52798 F20101108_AACCUY jagannathan_s_Page_047.pro
5c2e9d2b8634c2e30d61de9baa5f6c30
5f66a5679af01557272b117340babeba73dab1c0
7070 F20101108_AACDBH jagannathan_s_Page_024thm.jpg
d0b0830a7f83e2c6a8f0a9e030b7c4ba
66a05765231a4c9f336db203e90014717b9c38eb
21238 F20101108_AACDAT jagannathan_s_Page_013.QC.jpg
1569c3da538dc42824ef0e7820f531af
fd55322d4ac2d5746c421c22c4cbabf99e25f6be
23027 F20101108_AACCVN jagannathan_s_Page_068.pro
28ac403cb23ba3c026ad8c13a4f95841
3b9ff0756660a47016d1acbf7c867cc9a4691cf7
50461 F20101108_AACCUZ jagannathan_s_Page_048.pro
e71e6e5be30ab1d5fd38473357ae2eff
27d40956816bc7b8b0cd05d50366d21964f9b103
25722 F20101108_AACDBI jagannathan_s_Page_026.QC.jpg
9b07be8f13b19e2dd2cc634d21e7135b
4a4d34852b6725e763690bd9756401df35b58431
13380 F20101108_AACCWC jagannathan_s_Page_085.pro
17c6230c06bd38e44565e2843332e27f
c0dcbd35e82a19531e4c597e50a2aeaeb4643f99
6081 F20101108_AACDAU jagannathan_s_Page_014thm.jpg
1a6cc3a7de1284fc1c6842aaf311e6b3
d3fc70c9c33ab3a5f9edbb8b5285b4b4da221fb7
14201 F20101108_AACCVO jagannathan_s_Page_069.pro
7a40d44b81376621b728ca9215e29b45
7b41920c3b78de463c512ed60b490ecc7839913c
24850 F20101108_AACDBJ jagannathan_s_Page_027.QC.jpg
b2a25edf4076afe41f181d57072d9953
e28d953fe2cbabd33053bf5f0f6ebd2a9b387c00
17129 F20101108_AACCWD jagannathan_s_Page_088.pro
7f38a54010ba18402aa23035c29f6c0c
2f629eacfe47910b582ae79bb347fbeda05a853d
25731 F20101108_AACDAV jagannathan_s_Page_015.QC.jpg
b59e665546076740becaa720a8c4a38c
1e7443054bb58f8d7af3c71dd96447a0d9ea24ae
16818 F20101108_AACCVP jagannathan_s_Page_070.pro
1665b3e4add87fddf082e648738f6474
c3c96494aaf11696c9fbe256a0c172ac14c04e0a
6750 F20101108_AACDBK jagannathan_s_Page_027thm.jpg
5cd2b22ca2477dbc31ed0c5343c1e058
892a35d6f8c05ece4ef802ff38e4cae3e311a966
16812 F20101108_AACCWE jagannathan_s_Page_089.pro
82d051763ba8e2e175c7b7d804cd744b
2908cbab52f175b84746c13341e3c387f4a44d53
6861 F20101108_AACDAW jagannathan_s_Page_015thm.jpg
e7c3865c5fb8670bcfd6c3e6f1cb8898
53ae00705bd0e09e9752b0debb0005c3d9b68f55
15988 F20101108_AACCVQ jagannathan_s_Page_071.pro
62465c45b31fcfb7499d79410e5e496d
e28110cda5e362eee34d746c7e96e8a2cfa938f4
7228 F20101108_AACDBL jagannathan_s_Page_028thm.jpg
adbdc81993d679e0cdbd824736989c33
52e35118fdddb0ed5daa8a2ca63f5d2b23124659
13590 F20101108_AACCWF jagannathan_s_Page_090.pro
2cc8693eb9fd06927456f8af1f76012e
1091163b53f6e5987c12c622afcb4b1389a05530
9695 F20101108_AACDAX jagannathan_s_Page_016.QC.jpg
a11d68979a5e8975b8dd4aae50f9bbb3
0d55aeeab30d6a619164bcd53213009a24aa3197
17153 F20101108_AACCVR jagannathan_s_Page_072.pro
ecbe6408f4eaa41e4b83d0c0eb533a4a
777a1e7976ce9f2886610c12aaf14cf0d09b7f16
20663 F20101108_AACDCA jagannathan_s_Page_041.QC.jpg
6dc1c8a2dda338bb0751e7823bda0258
4858cb05e5ccd65385149f10d94139a7afce1af1
8813 F20101108_AACCWG jagannathan_s_Page_095.pro
3aff921299d3ad8fdac0e3abbf78c4cf
5de2f66964f3acce8d1c4875409a19d854feb417
2930 F20101108_AACDAY jagannathan_s_Page_016thm.jpg
fa36abcf3b8591806ffbb6c5a9446eb1
5d7a286b113f046da3e85a655531e56d1f7cbf3d
12011 F20101108_AACCVS jagannathan_s_Page_074.pro
b267175ae0ca2db2b88a90dc5db41f94
a6805e49577192598af138d6a0335846e7749899
6266 F20101108_AACDCB jagannathan_s_Page_041thm.jpg
1c40bd91eb15d29c08159060c1251c0a
7f47a1e9614d1da8839405a65e73feec76c8ce49
23990 F20101108_AACDBM jagannathan_s_Page_029.QC.jpg
c571bca0ee5a4641968cbd48598c7cfd
5cdf8b2584026666e85d0f6f35024b32fed44624
11865 F20101108_AACCWH jagannathan_s_Page_096.pro
f188211a7115b42efa1d9bb630c45097
83541e109abb5896b8b03060ecb03699543a6615
26587 F20101108_AACDAZ jagannathan_s_Page_018.QC.jpg
4cf4afff8c7457f5f073144dac8ff41c
01bcbf233f686633c5fdb0dac85811fbfe644079
15212 F20101108_AACCVT jagannathan_s_Page_075.pro
1988adab054573f85add391aabae7db7
154190bb0fabd62833d7b2f85fc67adc629e44c8
6910 F20101108_AACDCC jagannathan_s_Page_042thm.jpg
fa99aa0338dea6f39d515c639af05ab3
3380c3ddd136dabd227deb5309243db6f1cb8d44
6738 F20101108_AACDBN jagannathan_s_Page_029thm.jpg
5e24fa97b34cb694c58deede2af2845f
1af3d3368747fc07c5626711aab17f21d6f6e106
14860 F20101108_AACCVU jagannathan_s_Page_076.pro
ec8ce1e3ad8e48124a06ea4c94b99433
d3b8f243d16b15c7d96c2bbcb9da2eb296137048
25261 F20101108_AACDCD jagannathan_s_Page_043.QC.jpg
624d542cde6617a7e42ad79c0a7f6f86
a875c27226d1ba98376c27fa022bbc83f809af0b
23406 F20101108_AACDBO jagannathan_s_Page_031.QC.jpg
b6f734d5e557b1df6b93780823aa7dbf
2d9f9c83a7fc5adc18555707c567a76e3c7c1e0e
15911 F20101108_AACCWI jagannathan_s_Page_097.pro
aec0237edb80e0896d0a89169b480d17
9196bc393edac997447f13dba6d4996d81c99e3c
7673 F20101108_AACCVV jagannathan_s_Page_078.pro
7ce5db49ef2e7aaa33fe77f3b4b65d09
02752fd7674254888ad9d6f20993b340e1a85438
25386 F20101108_AACDCE jagannathan_s_Page_046.QC.jpg
3e08c17a7f788b298c82affc4e71ae35
80f9a097f4910d71d823e7dcd5cc7fe0634bbb09
6437 F20101108_AACDBP jagannathan_s_Page_031thm.jpg
ab738829c8d82f11288e2ee1528995d8
7c62db369e20f81a0c2d8914dcd5fcd7aef0a182
12510 F20101108_AACCWJ jagannathan_s_Page_098.pro
4befdc6d180dc0bed26ef2b781b92293
8dc875d223c890f11aa63441edd919fa3e7e03fd
9125 F20101108_AACCVW jagannathan_s_Page_079.pro
9854a7a6774967dc36180d9aadb1b6d8
86a46bf3f574fcbedb74edf94ea5e6926f2f6092
25936 F20101108_AACDCF jagannathan_s_Page_047.QC.jpg
293ac9fa6959bd9fb76e151ed7822824
e6ddde7f0091eea644c49e1b776cd04cee8f9a09
16068 F20101108_AACDBQ jagannathan_s_Page_032.QC.jpg
94f11e1f55af603db74874ac5662f6d3
03fcfe83192a784e118a4c270413fdc23a4ebcc9
14376 F20101108_AACCWK jagannathan_s_Page_100.pro
b630f2935dab8328e3fdd4ae0c573685
965589f98d686b570d715cac826681a46d8a72ab
49593 F20101108_AACCVX jagannathan_s_Page_080.pro
f58e75c64181fcf7bb9d516802ed67f5
b03f2150a0899d75943586caf0be5f8ae5c5d2d2
6943 F20101108_AACDCG jagannathan_s_Page_047thm.jpg
3447c5b118c6ae6c7850350fb80a16ca
c0212f60999c42cd80716f8d2838b12992329e26
51238 F20101108_AACCXA jagannathan_s_Page_119.pro
43c2f955f18b3271ea7c8d4be2c635e1
367bd312b4448a347b2bcf94081406024530ca0e
5001 F20101108_AACDBR jagannathan_s_Page_032thm.jpg
d0b0112d19f6b205bde07e5ce171a228
f76c2d51b1b89ddbd6bc79708bbf8a735ca45acb
14582 F20101108_AACCWL jagannathan_s_Page_103.pro
87cf50b127a84108724c523385e533fd
e5d92ba3d4a52e9e6633c7dcd2a79ba8ca6bd523
56020 F20101108_AACCVY jagannathan_s_Page_081.pro
c6d0ce47142990f14d5804b5dfb8f2fc
95feddb72e23ead60c092992c94546a875ffd311
24216 F20101108_AACDCH jagannathan_s_Page_048.QC.jpg
78b67f9721421b20c314eadb22f161e3
6c713c95c45e08247f7f9cc9e2a9cbc34abaf12a
11752 F20101108_AACCXB jagannathan_s_Page_120.pro
802e988c6234b5708cd729f9d29f475c
a45126d3f6e9299685877cd8ec580acfdb8ad856
3628 F20101108_AACDBS jagannathan_s_Page_033thm.jpg
01e91ef39afa671ae5e0fb09d6590277
f525653cadb34aeb0d9f11eaf683e08ca7fc3a74
16968 F20101108_AACCWM jagannathan_s_Page_104.pro
2834ae13cb1405c0df9e9a6e83b0dec6
1864c4f9ab6f11f68d2807b369f848d129f5ee01
50256 F20101108_AACCVZ jagannathan_s_Page_082.pro
93a5109b1af9ddd3ed86e3f84b0c2ef9
2e09b608d634b3d002bf7076b9de6a09f4fd1751
6926 F20101108_AACDCI jagannathan_s_Page_048thm.jpg
fa1ccee91c956af7d4ffdc64bb0d1b95
c73c79a8a18fc11bb9c02a1d8d69f9830590958b
497 F20101108_AACCXC jagannathan_s_Page_001.txt
676fd7b04be5593b39a8eab78b7c846b
73cac5b1028a5bc6c62e7a0944b3539ba0a3edd1
4625 F20101108_AACDBT jagannathan_s_Page_034thm.jpg
cba119224fb455b52c31380a260f2b29
5f9749fe799524c003b5492e42f7700e6fed754d
16340 F20101108_AACCWN jagannathan_s_Page_105.pro
e0f4bd7f72f5dacfbbb4d3dd9eb8e683
486f8ca32dfc09bcf1e588955ea42fa5eec76403
6415 F20101108_AACDCJ jagannathan_s_Page_049thm.jpg
267c5db172d0d90a5d51d70c0bd6edaa
c78c739a5fd3a9e91ce9991339198478fbd04f8a
3065 F20101108_AACDBU jagannathan_s_Page_035thm.jpg
d2aab15c6598469b11917a906dc91e83
391552a7c45ab8a3da850600c679d409c594e170
17404 F20101108_AACCWO jagannathan_s_Page_106.pro
fb73daf76605b3162e14743abfd0fd41
82fbcf5853db1b417b772ecdfad12ee4d20fafce
22241 F20101108_AACDCK jagannathan_s_Page_050.QC.jpg
c9df52cfc459b53278966b484bb3720a
1da446aa3f43e8c1123ac663e0a218e125991404
93 F20101108_AACCXD jagannathan_s_Page_002.txt
1ddba49157378064b9e631ab2128c957
c5048993d780fd740c21898608533488f81c8fc4
25254 F20101108_AACDBV jagannathan_s_Page_036.QC.jpg
774a363f376865b6177b89ecdff1efea
47efc12a8b723e369c2f788b44379b84125c2fe7
16265 F20101108_AACCWP jagannathan_s_Page_107.pro
c5aeb1931bf3b01c855b017b424530a9
cd7f18e223ee7a9bc59d36f672c9223360a152ac
6416 F20101108_AACDCL jagannathan_s_Page_050thm.jpg
5a4d24211af35052ce1157891ccd725c
32d02b5692d87f87c31b775eb1a862e866978d34
2557 F20101108_AACCXE jagannathan_s_Page_006.txt
b4085e873dea0e7f0d82640db79eba1a
a608804552890f09979340dffcedbc6e2903620f
24233 F20101108_AACDBW jagannathan_s_Page_038.QC.jpg
3d5297b7a7b500fba918e06f530b56d1
e94439b77ecc02e77ff414f0b7de5b5e21430741
16796 F20101108_AACCWQ jagannathan_s_Page_108.pro
7c0ce433d55a3bfa3661270035eac83b
fdd83980a8527cac15d8129c5d787224bac22160
6506 F20101108_AACDDA jagannathan_s_Page_061thm.jpg
da73a636ccb07c2a3da37989decca0f7
cba8ba739eedaa0ac810cf68ff879acded6e6559
19464 F20101108_AACDCM jagannathan_s_Page_051.QC.jpg
2d15ea179dbfd9cefd93123bd0a938e6
30648c4abd04176c24f16e357e5bc7aed91151f1
830 F20101108_AACCXF jagannathan_s_Page_007.txt
1d18b52ff1e51f020635d9fb3f619277
a23112d66face5b1fd971419a51691dc670f84c7
25421 F20101108_AACDBX jagannathan_s_Page_039.QC.jpg
94f85a65e0bc2f8857aab77e84ef403c
2f912ce469e73c9d28980eb571ac8196e92e6c73
12528 F20101108_AACCWR jagannathan_s_Page_109.pro
6eeac4581222c66e5302c9b50ad918e5
cb532d0cce51716bf74c8a2729e7efcf7397b440
25476 F20101108_AACDDB jagannathan_s_Page_062.QC.jpg
ebae3af5dd38fc9968b405fcc05db457
c24ddd8c8ade325075446d21ffbc6f70d0ebb8ce
2510 F20101108_AACCXG jagannathan_s_Page_008.txt
4501b78ab0060c386975a9f23f5f3df9
c92ce197fb878963bf1f76eaeab4d72a3548ce06
6924 F20101108_AACDBY jagannathan_s_Page_039thm.jpg
ca63161bb0d8c80e308e82d234d2d2d7
8c25b26aacccba07eef16f29323427ecea748b40
18046 F20101108_AACCWS jagannathan_s_Page_111.pro
ab3eaf39204e57ca18354d71811d05f0
2974a872cf41925c426226357ce0cab6b4f5df9f
24864 F20101108_AACDDC jagannathan_s_Page_063.QC.jpg
5ca912b914bb6ef1cc94d9139db4753c
05346732fee4cdaa8dd39f5127c30ce55e6b7914
5751 F20101108_AACDCN jagannathan_s_Page_051thm.jpg
d054ed33aa2575e36927c5bb2dcf6b37
20a7b25b57f38ac8c1eecd580f449ca0302d8447
2370 F20101108_AACCXH jagannathan_s_Page_009.txt
d985771a23e55955091be07c93da9ae0
7d2264e840a34e24cd397b78265d52e20bb9c53b
24429 F20101108_AACDBZ jagannathan_s_Page_040.QC.jpg
5cf879ccf404a6ab9b818a94f09d1141
aff2d8eefc57330a7e6dde23066d3c59132db00c
11521 F20101108_AACCWT jagannathan_s_Page_112.pro
c154d757e55cde3b9eaf2d6667696cdb
8cdddb8efe2261c365bc3fe300f65054df603ef1
9707 F20101108_AACCAA jagannathan_s_Page_052.QC.jpg
83bf6a77267cb933e9bbb6fe86a4c4c5
ea3d2827e501945b84398e58bbcbec5c7305897a
6987 F20101108_AACDDD jagannathan_s_Page_063thm.jpg
c325bff89633f27020cb8eb4f45e295b
aa7972c831933a5df7ce058e6b7ffafe83aba7bd
3333 F20101108_AACDCO jagannathan_s_Page_052thm.jpg
17fe9a5a0196aff05c9e952f4db7d52c
9f9731187a65bfde13987e43415fd59c74607fab
1605 F20101108_AACCXI jagannathan_s_Page_010.txt
9d5149b28ecf1fd200e6c9c877c22e7c
d0934a35b9169cb300f908a84dd0d415ba46e881
12870 F20101108_AACCWU jagannathan_s_Page_113.pro
b14b2310f0ea5ff96ded3523f938869c
f59d7581592d9cb8b103ca48364d86797cb24a53
F20101108_AACCAB jagannathan_s_Page_087.tif
e501ae428498973cfe89c227354cbca2
0ef6b9e155b8f88d2fbdc781796be97131d70622
6922 F20101108_AACDDE jagannathan_s_Page_064thm.jpg
5a9918cbe0befd95e46492dd6586be9f
0351f916e71412271d18dae66283304d4c45a2e1
6549 F20101108_AACDCP jagannathan_s_Page_053thm.jpg
28ad00dc608862776b227a7ef64b114e
77ae89441a42928a541250f79944eb73a409300e
2022 F20101108_AACCXJ jagannathan_s_Page_011.txt
bd746ebb1900ed06ac7905a45e79b387
9c9f72b74842b59906c00676ec4d33f8e4787f70
16614 F20101108_AACCWV jagannathan_s_Page_114.pro
7a14af012e102ae86e774795c375e34c
17a545e4b52cdc1618e349e5001b27bb21b7dd10
3759 F20101108_AACCAC jagannathan_s_Page_106thm.jpg
ea80cd1e9d93c545e22861e505bd32e9
fd0498019c3a6062b76a76807eeb71ad114561c7
5807 F20101108_AACDDF jagannathan_s_Page_066thm.jpg
9287b08a5f7123b806fa534cb30798da
488b4e855c283ae4a4a93012ada30ab55e7a4dd0
23786 F20101108_AACDCQ jagannathan_s_Page_054.QC.jpg
92f7bbe81f6c44adc10c15dae87762ae
10b8e226a3b24367fecc3697e6c362ad5b2844a5
2123 F20101108_AACCXK jagannathan_s_Page_012.txt
328ad58cdb5d8f76480c734989e4c1a7
e716c1acf1f52f3ccec2ebddf79e77ff90cbce2b
61950 F20101108_AACCWW jagannathan_s_Page_115.pro
72bf27bd90c0bbceb3b9c2e152c8a014
476c506d1dfc30a32510698daa5b9f78dfe2eae9
68325 F20101108_AACCAD jagannathan_s_Page_050.jpg
10a6cf2c87542aab520421efdd0e0745
1cfeb5ce23f65b75510c40b47ba40d9f6454765d
1883 F20101108_AACCYA jagannathan_s_Page_037.txt
521e47e50f9933a623bda4fe68731d09
43d5d794f1bcc3a5446678d8be01c2ea35a69b70
6578 F20101108_AACDCR jagannathan_s_Page_054thm.jpg
1105af2dc0ecaa5d2343b5987743c897
aff6ea1ce792bd9d6dd77a68fb0fd7ce26922d8f
1952 F20101108_AACCXL jagannathan_s_Page_014.txt
a38006236c983fa6e7c918a2476a01c4
4207436e83bf940d80508b7f25279a87d82cf9bd
62682 F20101108_AACCWX jagannathan_s_Page_116.pro
8d999a73bf75d144b15405499dcea1c5
e3efed4213739014137b9dc29b59f52acf4af59d
554067 F20101108_AACCAE jagannathan_s_Page_099.jp2
b9b340a5a2f0b0772794207a734f86e7
b571b1c04f748a2b884211d8a696e34e63cd0f54
18839 F20101108_AACDDG jagannathan_s_Page_067.QC.jpg
884b576616eed15f5401e4e8bae0dddf
32befa3f97b08da1b95479b98918df9827659426
1990 F20101108_AACCYB jagannathan_s_Page_038.txt
cdd59d512e7799d4672e82a854bd4fd6
b4e30c7852ed826d098c1deed37f48e1c36907ac
23979 F20101108_AACDCS jagannathan_s_Page_055.QC.jpg
aef2a344077610a1b7d4be2dcfde0715
aeb19a31b16658ead1511858e1d4ae8900773a2e
628 F20101108_AACCXM jagannathan_s_Page_016.txt
8dfca0cbe04fc297ecc081a2b2b11e96
53eb40d90b3ca119582dd6390b2b75c82c2d279f
63828 F20101108_AACCWY jagannathan_s_Page_117.pro
5b2b4c528365e127bcafd01ee883a6ae
dec90d9305eb4a8db290afd12507eb03b4c000a9
52980 F20101108_AACCAF jagannathan_s_Page_036.pro
238b4adec8acd0516cd6de5a01de2253
4d1fbde7d6f4c4a1705e7d40c1929611a5d80d79
5307 F20101108_AACDDH jagannathan_s_Page_067thm.jpg
7edd1c2c7eefbe3fe2d5bf557d13b116
63af2eb82527e0f9d1a28531eef8c6f051834d65
2134 F20101108_AACCYC jagannathan_s_Page_039.txt
28e343c472b3503d517689cc8a51cd13
68a51b62258b0a97d38f019a51b6b1c0e3856aad
6529 F20101108_AACDCT jagannathan_s_Page_055thm.jpg
3e8f81c0da87fa51fd463c8a1b8a6761
3aee6331c2c7b1e520f9ba03c8a69f8960d6294f
2224 F20101108_AACCXN jagannathan_s_Page_018.txt
ded808ed33f4d136866b1dba9aaa733a
c28435aaa4cf859e4c405d763d49738cd3135e28
62800 F20101108_AACCWZ jagannathan_s_Page_118.pro
ee62b880e406f7240120c678fc6152df
ce7ad0026b1b0a44d28a38898f85bafae1ffd559
12569 F20101108_AACCAG jagannathan_s_Page_096.QC.jpg
b14d49214181f1e831d45d17edfb04cf
3c164843de7a8262d29f262de18468e81276e75a
19676 F20101108_AACDDI jagannathan_s_Page_068.QC.jpg
50bbecd11a2b16dd4c758576aec4eef7
55f7ae63ba2d0d235220738c6df259aeae4f15db
2001 F20101108_AACCYD jagannathan_s_Page_040.txt
66ca7db3c106b609cceca32fd1e90734
900d03025e42616da6c877120d3852e85ee0be95
22221 F20101108_AACDCU jagannathan_s_Page_056.QC.jpg
862000629215484a9728354d31b0d802
bc1888ec318c805871f7464cbeb810d7ac16fc2c
2130 F20101108_AACCXO jagannathan_s_Page_020.txt
5eaa08ad264fc59a03c4f408e2b4d9fd
5daa53429da3dae38c6cea9a6b88f339929b8499
43026 F20101108_AACCAH jagannathan_s_Page_105.jpg
ea98f6319b09960d05a4b43b7d170c4b
93d1f991774d84b3913d9e73962134c61427fa71
13794 F20101108_AACDDJ jagannathan_s_Page_069.QC.jpg
c642703053c98d13cf48218086c49545
ffccef093dc8b866c1218ebf6c8f45f23c2852bc
7035 F20101108_AACDCV jagannathan_s_Page_057thm.jpg
723779714753c57d37189ae4730516f0
a0168f7983fdf4dad2317a19ae01e006a48634b2
2109 F20101108_AACCXP jagannathan_s_Page_022.txt
fcaad391678062d37431225b76121bf1
6e1bdc4698b4ced9bb81c414f8a5b09d90335f7e
13755 F20101108_AACCAI jagannathan_s_Page_075.QC.jpg
9f91e18c82b2ec8ef27d4b8291aa1535
feb2729d8dd5edc1ffee01189707de9c8dad932c
4332 F20101108_AACDDK jagannathan_s_Page_069thm.jpg
c6ec865bfc8023c3e14ada742b1d4485
6fcf253c58e7c556edd8d0b01f6fedd93655f4f0
1902 F20101108_AACCYE jagannathan_s_Page_041.txt
6b76aa9c055d0422518d0142c2d8e201
a91accaab9111871c20732f68e7533f56311e8ae
F20101108_AACDCW jagannathan_s_Page_058thm.jpg
8f449eb50382237eeb9bdb3be6bd584a
87dc92139591bb5d2167e5ac63affec0bb919d9f
2079 F20101108_AACCXQ jagannathan_s_Page_025.txt
079c7f6a22f4570afa609f0966383207
634e018a72a0584247f9dc9aca95642b797a5eee
20297 F20101108_AACCAJ jagannathan_s_Page_007.pro
0a55c4f8a8f56851678b61223dc71ba6
801d04f442dce1ea5302efe8ad082f668fd542ef
13331 F20101108_AACDDL jagannathan_s_Page_070.QC.jpg
c235a5f5e4d0475806249e0902e20a87
f35288340939c4ce2396580af35953aa9fe1586e
13515 F20101108_AACDEA jagannathan_s_Page_083.QC.jpg
fee50af30dd30320f793d98f5c5a4a9c
a7dbf8605c596ba805db9349bace7a2e9de5e358
2051 F20101108_AACCYF jagannathan_s_Page_042.txt
5aec0de2bf090383348d3ba8dcf9bd5d
51754c962590ed6b9a9990050d3571885aa52322
22330 F20101108_AACDCX jagannathan_s_Page_059.QC.jpg
1a664387f69d624f6e493be72542c7d7
8743af93f2e9917321f06e574c61c4f05972a670
2222 F20101108_AACCXR jagannathan_s_Page_026.txt
278597cb426a0707764d30a2e86bda4e
2712969be8606af73e0b79690a1588a1b3f0442f
13321 F20101108_AACCAK jagannathan_s_Page_094.QC.jpg
d4b7163ac6f40943fa4da05868c51b6f
a60dd11d03a2cfe62ff47827dc1cd1e1440475b2
12860 F20101108_AACDDM jagannathan_s_Page_071.QC.jpg
41802978bb62212c1308bc57a4f973a1
01602b185015fba7103a09137f8fec621abbadc8
3897 F20101108_AACDEB jagannathan_s_Page_084thm.jpg
7d1c986490f36ab6e376a22d49c03afa
1bde837e5c200b44dcec67faa511dcb30decc3d4
2101 F20101108_AACCYG jagannathan_s_Page_043.txt
88a782c035c1625d521131eb76f8b8d4
6286554f18a49f875c628f0d06f82c71cd9d1566
6607 F20101108_AACDCY jagannathan_s_Page_059thm.jpg
41edc32b240abb51ae54e235d031c8ee
fdcbb52498ffb676fcb2552f722294dd2fb53cab
2196 F20101108_AACCXS jagannathan_s_Page_027.txt
906ce7d4ae0d36c071929be26dcb38e8
92bd7776ba6be5797eb093790939522d6d4441e4
12861 F20101108_AACCAL jagannathan_s_Page_092.pro
fa2ce7cd810e9943b1ac3b5022bfb4a1
6c20a61d35b417e5af221e34f5d8e602d926334a
3941 F20101108_AACDDN jagannathan_s_Page_071thm.jpg
939345e8a27775b0cc6428d78c37d2f6
0c70fb1b5154e7b1b517058ffe96aa79971f4976
13598 F20101108_AACDEC jagannathan_s_Page_085.QC.jpg
6e75fb8a3df439449650f8317298cf24
d44ef83841adefdf84879764727738894686b038
1908 F20101108_AACCYH jagannathan_s_Page_044.txt
f9ce7d09180bdff6307b670a46533d1b
d6a7df2d86026999fa0b4f4a6091b9140ce82530
24694 F20101108_AACDCZ jagannathan_s_Page_060.QC.jpg
406aa0b71e61b381771cd3a9472818d4
7199a078075b806ea1802db17fd2d876897a9967
2168 F20101108_AACCXT jagannathan_s_Page_028.txt
7191a58d2d66f964c7830f98d562f413
8d764bc4d5aface47c4c20a0a1c854478ff919ea
F20101108_AACCBA jagannathan_s_Page_049.tif
10b92c288c8f77cabb49c86265e03d45
0920a33ddae4e8cba486f18a93a3222d0ee11d97
4040 F20101108_AACDED jagannathan_s_Page_085thm.jpg
e29d4cad53d2ee94aedaf0d00f7d3ef7
03828e6339685b5ad51a19ce824990cac0c2470a
2038 F20101108_AACCYI jagannathan_s_Page_045.txt
40c63a24bbfb304a52b1044a7a2b389f
023c964ab99cd58bafefe5fe33789ddd2ac0a789
2009 F20101108_AACCXU jagannathan_s_Page_029.txt
b53eb678fc0b1f70d2d07e93bd727095
5b2bf728c79a18c0a9da6a9448b3729996448a21
104991 F20101108_AACCBB jagannathan_s_Page_049.jp2
dca5e0e691890286cdf4ffdfe3728a3c
27be515735dbb9ada85306514f167e1e09c59616
13163 F20101108_AACCAM jagannathan_s_Page_101.pro
7d375b5095c11b93a2f396dcaf6b012a
65dbc88408616614cf91a0e287c1ec5c79cbab7f
13448 F20101108_AACDDO jagannathan_s_Page_072.QC.jpg
fc3f509d9ba8cdeb39f00133faf6ce5b
5653fd1ffcc156074cacbd55cfec64e309ed03e8
12363 F20101108_AACDEE jagannathan_s_Page_086.QC.jpg
e6d3be83554154a3537806fd4704728a
179d1f4806ac6d48725cf44bc441b975035c6f61
2115 F20101108_AACCYJ jagannathan_s_Page_046.txt
4e5d3e063256bd858681861c1b74bd86
1d5d0f5a2689916cd22af4390a328622b1bd273e
2103 F20101108_AACCXV jagannathan_s_Page_030.txt
a7e96c0e7007bfc6a18ce386ffdabb89
727215329f1bf44eb160694f90a88829db712765
104626 F20101108_AACCBC jagannathan_s_Page_037.jp2
e21af1a8d1836305a75d8650c00a0636
d90248a0a3c0fd804d780931b0345e877c5fe392
1004 F20101108_AACCAN jagannathan_s_Page_068.txt
ca486f6ce0922bf8e9aa890c46ebae40
f07e3009269a547155ea9fa8357acb3882c1d272
3974 F20101108_AACDDP jagannathan_s_Page_072thm.jpg
738b9fb4747bc7c611d62cac6de51fab
02279c1a99816c3db159da4aa41ff2d4333f486f
3706 F20101108_AACDEF jagannathan_s_Page_086thm.jpg
da52830b8d4ea721149cfea0281a4ebc
2351a76b79879ebf1461eb2b899d9f8a04854de1
2082 F20101108_AACCYK jagannathan_s_Page_047.txt
e11f9f3e5fcef3f420c107b8be97d88e
1824e38c9d1a598419851ef9593b638bc1facfe6
1885 F20101108_AACCXW jagannathan_s_Page_031.txt
b27e8aedb7c6c562f48f8e5e77eab003
9ec15c23d46b2a2ccd967dc4b57734a8423bc296
15640 F20101108_AACCBD jagannathan_s_Page_034.QC.jpg
3cc09d1c358b3ba0fa9fa777015dff4c
fb2409749735897f8d370e9e299e85d18ed539c0
15061 F20101108_AACCAO jagannathan_s_Page_052.pro
8891ac56f91e33ec13763e54e395a404
316f9210b491c71ce4e950643d2bd1bd3a7eba25
12184 F20101108_AACDDQ jagannathan_s_Page_073.QC.jpg
9e2cd430e4031c1903cc1a6d68ff0052
e0bf2a03628f1cf6101d83c2101030f33fa4e3f0
13733 F20101108_AACDEG jagannathan_s_Page_087.QC.jpg
1475af9ea17d0f3d29502f27b4174a3d
ced4c583144477ce3c08e3e3c0b2384cad17c60a
1992 F20101108_AACCYL jagannathan_s_Page_048.txt
28cfb20555fbc173604e998e4d401301
d463804234024640066605dd1d1a0531174f0e18
433 F20101108_AACCXX jagannathan_s_Page_032.txt
e2a7160290a33c776ce03b203c54dea9
ab48484b8aee536cf64050f5f415e0d4b80a98a7
6414 F20101108_AACCBE jagannathan_s_Page_065thm.jpg
4cebb6d804f6d4fd3a79f89df8e11673
961af44149ddfec1a9cab329608915abe4272b00
543948 F20101108_AACCAP jagannathan_s_Page_104.jp2
a6ff0245a9250c0157ee5b14919d55e4
1093aa624d7be7a4ed82060c11f1ebee6c0e866c
938 F20101108_AACCZA jagannathan_s_Page_075.txt
16e357030ed7cd5c417e2aa1556f1d39
18d0c70fa2cad710efe2d324413f5fdc50b26c80
3980 F20101108_AACDDR jagannathan_s_Page_074thm.jpg
fc2f380320468d18ad9cac1ed4fc36c7
c8f1fed1914e8689fc60d7c92e0b762115b7783c
3972 F20101108_AACDEH jagannathan_s_Page_087thm.jpg
998833cdb87628d3f3b2f9fb943e2c23
5aecf5c3d97cf0ce8f77158f538eb510a6af6f99
1919 F20101108_AACCYM jagannathan_s_Page_049.txt
304236e47ae9ea566398f78390c19350
1267a61680be9c26a4971955761d5afb7c2328fa
209 F20101108_AACCXY jagannathan_s_Page_033.txt
d926d1b9ed7e01e3868363bc47772c79
a6ebdc82c81b3be44f364aafdc57e701cb888ae9
537275 F20101108_AACCBF jagannathan_s_Page_088.jp2
914d807efe81dc9c86f4e1e59589e951
97ecb7dd0279f0c4f0d2e847cae69d9964ae5d6d
50938 F20101108_AACCAQ jagannathan_s_Page_029.pro
aa8703feb487f3b65721ad4223a25305
308790b309c962c410354d60b603bfc15ccc722a
733 F20101108_AACCZB jagannathan_s_Page_077.txt
91330f368824eea012fd3d49c1cb1587
20e900fab3ee55ed05c3aaaf1aa167d644b19f7f
13175 F20101108_AACDDS jagannathan_s_Page_076.QC.jpg
1755d64ff9e23542870b1a9d0ffbdbee
815b0f7ad96dfee17f93f88ac7f1a74d4911bd6b
13010 F20101108_AACDEI jagannathan_s_Page_088.QC.jpg
196eb31199dd1ea7e88d72d5f573fb53
ce1658e2eae60aa49ac4667336787a769ca5ca6d
2017 F20101108_AACCYN jagannathan_s_Page_055.txt
1e0383d0af80480a2941d17bed247eca
eafad03157cb50e6fc5fd0e30dc1ccb506b63535
490 F20101108_AACCXZ jagannathan_s_Page_035.txt
a6798429a32979b6eea84c1991217c6a
f98b498da63cc97dce35842d1bc5406c8f3d0c34
42433 F20101108_AACCBG jagannathan_s_Page_074.jpg
ff1d2ffc1d384a07cfa9fa72c6f0d355
d44e51b85ef5d156bab5118401ba76d3027d4653
24579 F20101108_AACCAR jagannathan_s_Page_058.QC.jpg
0878458b5206e2ea671411d993cec10a
3acd833213a89f038b823be09b7aac116d360e62
437 F20101108_AACCZC jagannathan_s_Page_079.txt
4726ab3d734ac6465ac069a2943e6365
254cad0c63e4fc610d0add6a6c9eec8d6c7e26cb
11691 F20101108_AACDDT jagannathan_s_Page_077.QC.jpg
7b2913359859c4e19f533ae411ad37a4
89fcebd68546d151aaea875d2fb0d43f44f52203
3883 F20101108_AACDEJ jagannathan_s_Page_088thm.jpg
474a2835b332ca5c8c81efb6ac445f0f
21a21ffbd1190ed3d23931d3ed188d79ccac55db
1938 F20101108_AACCYO jagannathan_s_Page_056.txt
438e18b6bddfb8744cabce21bd40c8ed
7eb61d39620fad3b1678b1aa6fef17ca06ae4cf8
44151 F20101108_AACCBH jagannathan_s_Page_041.pro
278a737df3d0f68fb253cfddb92bed1d
f7337c7fa59acb0746775492cce0831082d57a90
92591 F20101108_AACCAS jagannathan_s_Page_118.jpg
835042255833fada3142d16f2f8041e2
8685e2b286b60db4b44a1d25f3f7bf424bbbbc54
2064 F20101108_AACCZD jagannathan_s_Page_080.txt
029c09e8264e1fec6ddd82f1823fea16
ea7e941e5a687b0add7f73357b0ce9f805d6604d
11917 F20101108_AACDDU jagannathan_s_Page_078.QC.jpg
51d7cab53255ba2d09d9bf0e273dded3
aa500fc7259eb576b6f662b277cce2db97eeb764
3766 F20101108_AACDEK jagannathan_s_Page_089thm.jpg
cc8e30a822b23e4d97063f7a3f514f77
9ec652cc4fc7f122fec82af8e8c3f049b3829f1a
2042 F20101108_AACCYP jagannathan_s_Page_058.txt
fdce8d7f8c9ee78f48867f9bb7d6a28c
ddf070b99addc8dda127e36ae3a56eb96c0f3620
39665 F20101108_AACCBI jagannathan_s_Page_092.jpg
59df193f4f2bf6c49879e3fac5e6af05
7283ebb708f171a038dc3dc3bbafedebb88a9e57
980 F20101108_AACCAT jagannathan_s_Page_072.txt
8563cb1b1fc02de693af2a550d7368dd
8e4ef655d9345d4bdcd0b8a63002236c174229e9
2232 F20101108_AACCZE jagannathan_s_Page_081.txt
b2d20ef7a80a5d015cb870227e2ff1bc
7dfaa04f08fa714580075405c22950c3abff59dd
8114 F20101108_AACDDV jagannathan_s_Page_079.QC.jpg
c6bb0c34070c630b7df1a1a478c4b9f3
2a66207921df86ccf73a03612af7f602ca1dfc1d
12788 F20101108_AACDEL jagannathan_s_Page_091.QC.jpg
9e624a8fb34960a933768f24492717c0
86bb86e3b04e68bcdcabf1d3e9d702e8f28fa897
1851 F20101108_AACCYQ jagannathan_s_Page_059.txt
cbde416ffbb86db315e4057bd064edf2
aafa8469d13b43389b55f33e045206c10e1730e0
6689 F20101108_AACCBJ jagannathan_s_Page_025thm.jpg
2d27a217ee8408bf8468be5dc23b7328
9fe75a766b30eea15a6e220ecf05bf366762c7f4
72867 F20101108_AACCAU jagannathan_s_Page_045.jpg
e345996e7c0f191c2e4a3f68f18f239a
f918e31b97b78080eddd903f0e23469e4ff31537
2383 F20101108_AACDDW jagannathan_s_Page_079thm.jpg
9c3675e46d740908c294380fe6ec39e0
b35407e9d4e1080adc79ff3af411226493932341
3884 F20101108_AACDFA jagannathan_s_Page_102thm.jpg
37b828260767c01d4056f747d132a33f
e9641dd6db9ceb8f3292e514e40b2a212dd0752e
3796 F20101108_AACDEM jagannathan_s_Page_091thm.jpg
c2d41f2090b81a7ddc2b9c224e652f6a
09f59e16497569300560d2dd21e38b1b41871be4
6428 F20101108_AACDDX jagannathan_s_Page_080thm.jpg
15e093575b1a33a7d4d42df70ccbf89d
6418f027a994facbf292798f9c42c77484c7938e
F20101108_AACCYR jagannathan_s_Page_061.txt
822264bc90ca09f750704e68ebc8fb39
9d148562d53e07901f4f1722270d1e7d51266e20
1017 F20101108_AACCBK jagannathan_s_Page_002.pro
658a632c96dfa18b71225ff9530145ff
d571b93b2473ca7b36e4bca364b10ee4d97c9f72
20715 F20101108_AACCAV jagannathan_s_Page_066.QC.jpg
a040bdbc84f8e8efdc0c71901614bdb8
0924573d4317695f8dd38cfc5db01d9d1b02df63
F20101108_AACCZF jagannathan_s_Page_082.txt
b7427a28f3da4923c062f42098653d61
af2b24cabff5c445e0eb60a0e6154b592c0d07fa
12610 F20101108_AACDFB jagannathan_s_Page_103.QC.jpg
818dc304c962484790bf7d95d1efc421
1ace540cf3ded916381e6d59df9b8b3516d5a122
3834 F20101108_AACDEN jagannathan_s_Page_092thm.jpg
db59aef119f5b75fb0885719c33d8278
b79a11aaf5daaa5f3b59a84534ab4035eecf3b4c
7027 F20101108_AACDDY jagannathan_s_Page_081thm.jpg
8f6f1487bef84c1b412826c4970667ee
1b1717475a1525d4bf4cbab7f3c0aac09c7c3437
2145 F20101108_AACCYS jagannathan_s_Page_062.txt
db5dbff74b3630083dcd02389e925484
f16c107dc3c6116aa1691588d97e099166a66a72
68046 F20101108_AACCBL jagannathan_s_Page_006.jpg
a631a8fcf47871479e0389ecf11c576b
83d6833b8dd0b7a0108d18b87e2788fd2cfd3ada
60075 F20101108_AACCAW jagannathan_s_Page_066.jpg
88d5e68c88e9abfd706e6788af2f2494
1391e9658c79cbdbc84558af606e19756cb6ec14
836 F20101108_AACCZG jagannathan_s_Page_083.txt
3cc4ff4cd012d9db3f1805e7a8a4b9d4
6e851113b4f49d5e8ca54bf4c40d2e6bf0e5aea3
3783 F20101108_AACDFC jagannathan_s_Page_103thm.jpg
12848487d2655b7c359cd6ded25483b1
bdfaca6b9378196a027f9a6f509186e3d74e05f7
13825 F20101108_AACDEO jagannathan_s_Page_093.QC.jpg
8f9476e7d5385fb722e5dfd0b06fa81c
3efa84f22d60649288fcfb640ba09cc42f76f04a
24768 F20101108_AACDDZ jagannathan_s_Page_082.QC.jpg
162706c60623ecfa4f46a88eb08d1a91
07f730686f33f0d9f90618598cf328470decff3e
2083 F20101108_AACCYT jagannathan_s_Page_063.txt
0eaf21df6468de44eac49af100b70aa3
1b43e2b7d9a0eddef66968139f21e1eb5bdbfd74
550308 F20101108_AACCCA jagannathan_s_Page_093.jp2
c83e81a900ee9602d601f7788928b1db
5cd1d5570442d0e6712e09e9fb0a938a1ceffccc
71038 F20101108_AACCBM jagannathan_s_Page_011.jpg
8ee7d21d83be3a930e526985ac6d53c9
151f0f799d5929fd86a180018d7cb623ba406a32
F20101108_AACCAX jagannathan_s_Page_042.tif
cc92157d8cf9e0c4c3b5adb28c9c6a25
43bf1146518f7bf3b1b080ce63f620b772878464
566 F20101108_AACCZH jagannathan_s_Page_084.txt
3fb8bbfeeb4838cab5453ee962c38798
44c2520f05a74c163b9a8c38a9b18847ad548e5c
13022 F20101108_AACDFD jagannathan_s_Page_104.QC.jpg
e38e1fbded34f309419468116df99526
edbbf1b18638e7f611af93aa2a58a37497a8a9d1
2050 F20101108_AACCYU jagannathan_s_Page_064.txt
8bbc88b243d83a17fd01f9f6e0ee4596
194337937c8f792da92a4471695f7da02dea3428
6942 F20101108_AACCCB jagannathan_s_Page_022thm.jpg
e9ad2b934a723b3fcfee4eeb414c1c2d
7cf7632b476d81d0b15f3aea11e1e0a6ce76b838
2190 F20101108_AACCAY jagannathan_s_Page_024.txt
4c02082a8665fa5c6307722b13630950
f4cb7cf586ebfc041e3cd4fc3e83988cfd74de28
930 F20101108_AACCZI jagannathan_s_Page_087.txt
bc6f0868b1b8686f2ec891d42b4c5e45
b0dcc45069db09fcf9152d6b9d8d23f526cfefca
3999 F20101108_AACDFE jagannathan_s_Page_104thm.jpg
874ce9b0b99cee6ffa29c0403aee4cbb
8ea5c3e2f5386d237d22448f9667d51f4fce7250
4060 F20101108_AACDEP jagannathan_s_Page_093thm.jpg
6de607ae9417de6cff0112355b7219c8
d9db476282f1817f7f72777ebbc64cc64be3db8e
1973 F20101108_AACCYV jagannathan_s_Page_065.txt
d4b7732f7a3fe7dbc0a311e6f0bc2b4a
a3334e3c970187cea0e8f6c06956f4d0daf08629
941 F20101108_AACCCC jagannathan_s_Page_102.txt
ea119352891957a75c1e5f85c56d4829
48d69acb803dd22d4d7f9be1c272748d694ce713
3239 F20101108_AACCBN jagannathan_s_Page_007thm.jpg
152f0f40bece5d0bdd9ea473e2059587
e53baa462b15197600d388a07866ce9a72daab85
23647 F20101108_AACCAZ jagannathan_s_Page_065.QC.jpg
843c982fcf4bcebb46386033176412d2
b5ca4cd07652cdddfc457fb95f9fb2abfa54252c
717 F20101108_AACCZJ jagannathan_s_Page_090.txt
9ef22d129fcf4ac03716d5fbda092f2a
d0d3faaa8e38738e1cf5b96f6206a6b33eef98b8
14078 F20101108_AACDFF jagannathan_s_Page_105.QC.jpg
7ebc2e25295aaddab6a585602b08bc4f
183175eb6de8e46fe080c9c297cec14f574f28e9
3862 F20101108_AACDEQ jagannathan_s_Page_094thm.jpg
e3912fcf46f70d3edae7ddb438f488b9
7675ad94eaeabeea28169d42be32e4e047fc1a0a
1042 F20101108_AACCYW jagannathan_s_Page_067.txt
88e1b45a53da65671555c3261b1107aa
5eb018e79cad6fbe5aa263312aa085c9f60033f8
F20101108_AACCCD jagannathan_s_Page_054.tif
4f8a657b9cd7a900e40d251951212fbd
03a928dcbb19ca51f9013244f5d0ccce2ddf94ee
5521 F20101108_AACCBO jagannathan_s_Page_068thm.jpg
d9f726f860f466bea985fcfdfd5e849a
2200936cde9d8b9d93077d71f0550b4664b55edc
785 F20101108_AACCZK jagannathan_s_Page_092.txt
9270ba904a007af852c13d61d88ae838
e3a1b0bbc6ac66beb96fbc6526515c555ed69a1d
3978 F20101108_AACDFG jagannathan_s_Page_105thm.jpg
d31940c8ca5f53c487fdb7ee716985f4
b3f2b3666123090d7cb7bc1dae081e1179970f4d
12891 F20101108_AACDER jagannathan_s_Page_095.QC.jpg
848c5cc34206523b09f6e120d2c902c9
2c359bf7bfb06a53fe745a55e3a38ef1e615a12d
887 F20101108_AACCYX jagannathan_s_Page_069.txt
c4d6d09d8f3ee761508a6d2e5cc97d0c
f0b6ee3c99daa183fbeea5d9cd09ab994b6676c5
2119 F20101108_AACCCE jagannathan_s_Page_057.txt
bf39540b4f34b5dab1f2af26a8536281
7ade4b4e1f35438f08a12c984f54e16e29e3a456
6559 F20101108_AACCBP jagannathan_s_Page_056thm.jpg
fbeabc8cbce98694a7a8b879a517144c
40a63e2e67d437b4a1e048b362fa1ca23b6fc29d
734 F20101108_AACCZL jagannathan_s_Page_093.txt
55eb79568d9764a85c4fec8087e41578
f16566509333ce273fc9535eef76b42a23605df8
12945 F20101108_AACDFH jagannathan_s_Page_107.QC.jpg
f58f4e10d9c56802770b92448119638e
1d9e450d5682e5a7f13dfc9e497834e2993711db
3654 F20101108_AACDES jagannathan_s_Page_096thm.jpg
b147b182c8287271eec772da5a142ec4
91540d4ec47cd182b0388cdd8fc3a7cbb48d7aa3
886 F20101108_AACCYY jagannathan_s_Page_071.txt
79a0bc8b57d609cd3b85d84cb09ab3e4
7d05873f2346c0c85403ab8521182460850de7f3
812 F20101108_AACCCF jagannathan_s_Page_091.txt
6b287a19b2c15df90677c0636108e1a9
26f1349a1b78d6d7520d3c8a34ee0f0d09db1bcf
3635 F20101108_AACCBQ jagannathan_s_Page_078thm.jpg
cbace07e667ede86c3399fb01a39b4c6
f3cee3424b14b8bfb82a5d60f1af3cf9cb50d883
923 F20101108_AACCZM jagannathan_s_Page_097.txt
564b5ce6ab2d3235800812473c3f9eab
7ceb4afb708efee685f0174731b608607e834482
3749 F20101108_AACDFI jagannathan_s_Page_107thm.jpg
8a1a7f23ce23d858ae288148892a1c26
81d2f6f3b0bd5f5e8b214849c07ce3f69b4ff884
12296 F20101108_AACDET jagannathan_s_Page_097.QC.jpg
1979c319a22eb3193f4e9ea444ce6fda
a99ac7f44d581fe5cd4f8cd5be89ab9f0ec2e4db
736 F20101108_AACCYZ jagannathan_s_Page_074.txt
4e905b8ab2c644563056e14356b9e532
ae3a195ac9d41642362e31e9c44d3377414f5991
F20101108_AACCCG jagannathan_s_Page_082.tif
2d00499613ceb73096ae34debe8761d0
18356fc74fc5d592a38d0bacef966bc692bb7552
44095 F20101108_AACCBR jagannathan_s_Page_083.jpg
20391c19e202498736aa51616eb035f9
7f2e678097e562a67e7eae50bc6f7fe89acdd0fa
F20101108_AACCZN jagannathan_s_Page_098.txt
cf8e9ac143113eaf3cfeba3a523ac0a6
3a29c5d0231664d7bb5fe9e2a2d52cae00a3a8d5
12707 F20101108_AACDFJ jagannathan_s_Page_108.QC.jpg
1543f66ae15081e3ff73b4f9a57117fe
2a336963fbf4036551072fe0e35b94178a90998f
3711 F20101108_AACDEU jagannathan_s_Page_097thm.jpg
36c6002c5aa998f6851d287316610674
8f67ca4c32df802b3d82f08b46a4bde89733536b
40964 F20101108_AACCCH jagannathan_s_Page_076.jpg
1a3b511137f692c77f7237c925497864
e417a77fb571392781eb38ca65a0d32a285500a5
12475 F20101108_AACCBS jagannathan_s_Page_090.QC.jpg
3be9afcbb381a76265d396dd95850c79
d6f143ef914ff0af6af387a85bae900a773c9875
757 F20101108_AACCZO jagannathan_s_Page_100.txt
1a6c3815bf349fae3b2ebd22118e8171
e7c9cc45991bac8cd0881463ea9ff0545821de60
3594 F20101108_AACDFK jagannathan_s_Page_108thm.jpg
107b76c671e4211e1c973a3d47ba3a58
4162f985a5c2dc4c988da5441e2b5ff3d65671a8
12541 F20101108_AACDEV jagannathan_s_Page_098.QC.jpg
05078ee1a2f55e61bd7e57265aa99e97
d7278975dcbdb618dc97be28349d6b5d5e4e9ee3
49060 F20101108_AACCCI jagannathan_s_Page_038.pro
339f5bb4ee3cba169adbdd0a2c3ade70
eaef7c41049c369c52851ca631053899c092572e
39907 F20101108_AACCBT jagannathan_s_Page_091.jpg
cc8769baed60c9abad3c5a31ce477a8c
c362d08affaa6324a9544ede6baa8091270b536e
911 F20101108_AACCZP jagannathan_s_Page_104.txt
d3db7bf9cc3404cbd44484dc0711a18d
d9ec915c94b854be0fab2c5d1cb0e29f5ac4d46b
12840 F20101108_AACDFL jagannathan_s_Page_109.QC.jpg
abf56725c30a629df632d35b9e63acd2
079f7512b698118d846b12a89fffc0c39590546d
3764 F20101108_AACDEW jagannathan_s_Page_098thm.jpg
540bcf06fad7c7828bc1ddfcb4ebe07e
a89a260afac74e93971d96399f549937b6feb37b
55069 F20101108_AACCCJ jagannathan_s_Page_032.jpg
c45a06ed08d0e2c59bca1d48e5c130f5
0a7e6e15547317c09bb5d8cf42dc6ee707d7eeba
1685 F20101108_AACCBU jagannathan_s_Page_051.txt
9b079a85dd1b6864779571f3a0be26c7
5cb999e625f9ff97c573f8645439b7fd8d245e8b
842 F20101108_AACCZQ jagannathan_s_Page_105.txt
d1f8d4af701be4967025215e4a7ff50b
ebc00f2ad9173f1749f9aad6c9c807f967e88bfd
6013 F20101108_AACDGA jagannathan_s_Page_119thm.jpg
1c769dc48452452189ee13bcbc106576
e9db2205ed295424c3605c5e0965c3fb7fe2a325
3716 F20101108_AACDFM jagannathan_s_Page_109thm.jpg
54bd0b3071da8a83699f79123a7034b1
57278636a29e4db35e4caf61f7c5dfa0ef289830
3968 F20101108_AACDEX jagannathan_s_Page_099thm.jpg
bc241b7f468f23abdbd41f14895897b2
c0edbd0705579ff27f8c7b35ee55b22edd5f1628
F20101108_AACCCK jagannathan_s_Page_103.txt
70479b7cb68e0b7c730a26ac29a3f099
3140f16381270c50e68caae64316c9f574eff1ff
11427 F20101108_AACCBV jagannathan_s_Page_033.QC.jpg
bf79d4840845fc28125b754779635777
a7da64c8fd6c250602362e61bb336afc7451bfb7
861 F20101108_AACCZR jagannathan_s_Page_106.txt
25439e0ffc6467751f970772fb822d57
df678c07340998bed7cce7c67c464aea717f1384
8215 F20101108_AACDGB jagannathan_s_Page_120.QC.jpg
ec4ce309efbacd39ceb959b7eeb35db9
3f161b4ab3ca3793e879d69dc077b8226b3f90a4
12930 F20101108_AACDFN jagannathan_s_Page_110.QC.jpg
08bbbfebc16b00505966eae439b64464
4767118c6928efccd2bfc30bf5cf32d5443dcc9b
13177 F20101108_AACDEY jagannathan_s_Page_100.QC.jpg
5e1df7332314a3a239b11a3ad2f97bc4
14584374a3eefa72c56395e70335d15ff6bee7be
17265 F20101108_AACCCL jagannathan_s_Page_094.pro
8b477e16222066fedb9009d8e3ae8407
a89c4c75d719b724c306c7506baba35bcbb0f55f
25757 F20101108_AACCBW jagannathan_s_Page_020.QC.jpg
c5974c346f8e8938777a6134db14d98b
ca036938ca1850e05f202277c00f4b3077a6001b
900 F20101108_AACCZS jagannathan_s_Page_108.txt
a3827385ed001493ba5d341f85d7b6f6
72307d11a8b39edb47844a2d01d3c3581bab24ef
2508 F20101108_AACDGC jagannathan_s_Page_120thm.jpg
cf2e65470e5e583d903a8710168534cf
c2351888e237d03030fa8f8b96d5c775bea5eabe
13538 F20101108_AACDFO jagannathan_s_Page_111.QC.jpg
31e99dc78a43ae95822970bf36b30a56
1c48d9958360e97cf640f6411ab87b73eb021225
13837 F20101108_AACDEZ jagannathan_s_Page_102.QC.jpg
b3ebdf2ec28378ff642156ff35f66cba
5a2969bc07634de1271df2b4967561be155f8624
41216 F20101108_AACCCM jagannathan_s_Page_110.jpg
fcaeaa2ba67b7ece9dd8e11b5946661e
54e6ec2a71c2b15f8cba6aa1bce88bc5686d024b
6088 F20101108_AACCBX jagannathan_s_Page_044thm.jpg
af54cbc202744c825b91855dd7ca63a1
4e249761217282299ba344c4ec198b45ee6450cd
652 F20101108_AACCZT jagannathan_s_Page_109.txt
589e743a01ad57e8d623e0a39d7bde66
c9a3767a5effd2bc4515570ed50d91f47a06b0fb
25199 F20101108_AACCDA jagannathan_s_Page_030.QC.jpg
160168cdd624cd82f05e6e3a12daab47
1a99fccbf17657755e029cf7e012f982df9da97a
142038 F20101108_AACDGD UFE0021840_00001.mets
dee9f843eddf3769f23f05da1709ebe3
6b0f363966fb96cdceee259166dd3e7ce17f6a9d
3923 F20101108_AACDFP jagannathan_s_Page_111thm.jpg
649d27768eeec183c544fdf96f269ef3
97e067a36d28ea6ec2c067958d711569c8c82233
F20101108_AACCCN jagannathan_s_Page_009.tif
e8b7798642ed4869596339ca53f9e141
2c4e1ad27c0635a4c34dc6085f790e43b9ac222b
7201 F20101108_AACCBY jagannathan_s_Page_036thm.jpg
210befc6bb0481557892e4b017e9fd7a
f3fee62324a2666544a24e7557533d30070e0b10
675 F20101108_AACCZU jagannathan_s_Page_110.txt
edf049fb4bfd514f4fb10f026fd572b9
291538e008bc49dab04b4d9b646f1bfccd59164d
25456 F20101108_AACCDB jagannathan_s_Page_057.QC.jpg
78c27ca6621f0c7f2dadca9b90ed698f
fa27a3a2e0ee5a7c9e72487448b91c4677261a78
80236 F20101108_AACCBZ jagannathan_s_Page_015.jpg
687ff09dc659a3d05b754ba52d939e9a
f62a47a18582401fbb12af3279cb6a5fb3419a55
1046 F20101108_AACCZV jagannathan_s_Page_111.txt
63cd073545c63ad02d13e0912ea7e88b
a6bdcfe14dd40cbe4ad712a0347cd05f75e35b05
136086 F20101108_AACCDC jagannathan_s_Page_115.jp2
5beeb951bc3ef38538015cc47fb4cd77
87b5abc4deb4425e889b2e62688af6f10747f84c
4041 F20101108_AACDFQ jagannathan_s_Page_112thm.jpg
149fc16f0102c81de29691b62705e00f
9666e89c9b6db3283c5cb640b2b0ef2dd163c921
F20101108_AACCCO jagannathan_s_Page_071.tif
0d968e0118e67cae34e790bcfe638ef3
931c4fb19415acd08be1a76b6310ba599ece6ac8
587 F20101108_AACCZW jagannathan_s_Page_112.txt
81d1704abb0e769ddaf7ff7eb7d6e1c9
6d8f34e23b7e8e46667d79b1f29ae16f87cb46d5
330 F20101108_AACCDD jagannathan_s_Page_003.txt
43e4735923681bfdb54de698426eb835
74343f80a4889dfb4def51a5146e523aea449efd
12847 F20101108_AACDFR jagannathan_s_Page_113.QC.jpg
abe936d2c6b74b3f8d028a253e8f3862
945cfae600df830fa6827adb9ddc816d7a971773
117756 F20101108_AACCCP jagannathan_s_Page_024.jp2
0ab5181e6221f0359a7c634ae75a945b
d045ff94451a4f57d1ade54e77660b27478d2e73
884 F20101108_AACCZX jagannathan_s_Page_114.txt
da107e80b8eec53f7bac60501c643420
610a8d925da019074e9fc298e5ca91e57824cc3f
80626 F20101108_AACCDE jagannathan_s_Page_081.jpg
b3557e5a490cdd03f93706aa8a7328b8
99786f6644e448ec6d98ffdd92bc396c8cc0478f
3751 F20101108_AACDFS jagannathan_s_Page_113thm.jpg
ebf80021c327331d21682070d27705d7
5f1ae108273eda3893055e0063d5469e08970b02
13014 F20101108_AACCCQ jagannathan_s_Page_106.QC.jpg
d39fb8395ff37439591ecce02579e1cc
9730916515b2bafe7befd8aff9b76695a8c0fbf4
2533 F20101108_AACCZY jagannathan_s_Page_116.txt
23e43a90a1c3d7afaf48cc971aa2f373
c5aaa6902367225074f4832996f63d1efc1f70dc
960 F20101108_AACCDF jagannathan_s_Page_073.txt
208f1dc335917e385103cde215f6efed
a30b1ad6c7bcdb8c164a9b433ecc359f6ea396f5
27568 F20101108_AACDFT jagannathan_s_Page_115.QC.jpg
eb5fe2373ace8c7edb6cacc3e7652519
b37b2084d75af504ff214071f7a53aaff7d04c10
4093 F20101108_AACCCR jagannathan_s_Page_075thm.jpg
8c6e9dc92270fb8e5bbe617ea5708ba4
dace8fc1df40aacd3d0f00766bf6198b64a2cbb2
2587 F20101108_AACCZZ jagannathan_s_Page_117.txt
cf2c21c9db1e359211be5711c523ce6a
1d1cb98fc50589334e3c6cc4e70e9039a8f09393
F20101108_AACCDG jagannathan_s_Page_061.tif
7ee2aa6c7c5a46a6ccc848487f584f9f
146d8960113b7efe750b05bac2fc0b0cedd29918
7313 F20101108_AACDFU jagannathan_s_Page_115thm.jpg
944469477c8f4c7c619699226c437d4c
3d744686c25e887dc93ea09fa86c99097397879b
12808 F20101108_AACCCS jagannathan_s_Page_084.QC.jpg
c61e2b625ac7299f27ede6f6f0f6e4d1
5638a6af0e8b2b23c68561815f7ca1a06d0aaacd
25449 F20101108_AACCDH jagannathan_s_Page_024.QC.jpg
c90e35c644ec13d643e9e615f2f39484
7796540f3b514e2c7f0260abcd0a23c53e207a37
27564 F20101108_AACDFV jagannathan_s_Page_116.QC.jpg
8f8c931cf95e67fe442ee648f501967c
316317a4225be3a6a9f08733b9e0dd0440c89d75
43625 F20101108_AACCDI jagannathan_s_Page_069.jpg
e16dcc00cbef9751725503932d32a6e8
6ebe86c7554b13412ddaf912fb2622c101675d21
566262 F20101108_AACCCT jagannathan_s_Page_073.jp2
a24d712b28faaca9bbdeac4b95e26b6b
5311b97344de6f979c5f501fffb766dbd3eb5c55
7586 F20101108_AACDFW jagannathan_s_Page_116thm.jpg
d4cbbfcb24abf038cacbfe1c1b83be1e
ec7c620aa580f6a39f5169adb624b43dee04bb0a
F20101108_AACCDJ jagannathan_s_Page_086.tif
f8fb77a2e8cd13c44343cd2548c32c0e
ed0bc12049b9c5644e6edafd73b165329087a3af
865 F20101108_AACCCU jagannathan_s_Page_107.txt
15baaa0bd9028cf71a77c41bbda0516c
762b13cb9e70ae29c9c6601d27655cb313de0f60
7079 F20101108_AACDFX jagannathan_s_Page_117thm.jpg
6d3f555ab1fe16be45ddd2b2d9bd67f1
7b79aed144ddfb51787828f0caf3376c6e4ac29a
549904 F20101108_AACCDK jagannathan_s_Page_102.jp2
40cd6732c96c3b588eb329a63fef3fab
2c05435f7699c9ee534ac6f1c8af84689caf9d57
39843 F20101108_AACCCV jagannathan_s_Page_070.jpg
21f734e33417dec50d67b618888e9a62
1e7d3aefc4831090d6d50d8717a804ac3fbb9181
7099 F20101108_AACDFY jagannathan_s_Page_118thm.jpg
21da4a9fe8ea12b4c4fb341cf13d3e95
c243cdf86d60166373e1f39923861f7f2572f6f3
F20101108_AACCDL jagannathan_s_Page_043.tif
a2b850ae5f0484ca3ec0d00f1e6656ff
13ba91c6386eb9e86a1704e684cd6fb951aae4e0
1051950 F20101108_AACCCW jagannathan_s_Page_032.jp2
a030e76cbe5d7f07e8b7a1097a53c79e
061afa93d6df8407a7d94c8c5c93db4a16f622da
22446 F20101108_AACDFZ jagannathan_s_Page_119.QC.jpg
cb21578e7955bd872c3c251a696bc1a4
724c45f3667b635ca24adb0b9d573cdfeb65e91f
74830 F20101108_AACCEA jagannathan_s_Page_025.jpg
d7b6be6e4356796e6de47955ff6694f5
b194e6c48b89781f9962a23235d3c7435dead1bd
15128 F20101108_AACCDM jagannathan_s_Page_099.pro
7775b886e982d2ab0e038b25b796216e
f334a81b5a6f3275082a75cb8d09f8e9b913d36f
F20101108_AACCCX jagannathan_s_Page_057.tif
9e58be08dbade27704b860c11e3648f7
8f1e6c0d76809f9caa7b7b0cc796ebc94b176e03
F20101108_AACCEB jagannathan_s_Page_017.tif
d1b5d74730bdcfadaa89e1bb37b4e952
9f1f2383f9b224af0a25c1857b33a349e47bbfec
47929 F20101108_AACCDN jagannathan_s_Page_061.pro
6fec090dac4da61a8c0327587907ea7f
f394afdb564f39fb88902602dc4970307bcae3f6
52918 F20101108_AACCCY jagannathan_s_Page_043.pro
6cc0b4fcda910be5006fcb71661a1b1d
2273a942e1795fca4d9390ee9eb33d9690d74eec
2040 F20101108_AACCEC jagannathan_s_Page_054.txt
e135987dd628788f60522c27d65d1eb2
600716173667c20a59ef433ecc66cd7a483e5b4a
13153 F20101108_AACCDO jagannathan_s_Page_077.pro
c4b1d2a3685433f7dc282bbbf234ee7e
8294108383cb24c100d0431f0afd1dc747038835
24810 F20101108_AACCCZ jagannathan_s_Page_025.QC.jpg
32f2f323891039c2b9bcc129531305e1
3be410d871ddf80a95bf1570f59b20270f248307
108946 F20101108_AACCED jagannathan_s_Page_048.jp2
5484b29440695c0ef3feb42c235f166d
49f9df9578d74c523f7258ccfed0cc158ae27d9f
24346 F20101108_AACCEE jagannathan_s_Page_042.QC.jpg
b3d21b9046b02bb1d96280fcfd23042d
18bc75beec44d17cc3a429640646056da87afbbc
121371 F20101108_AACCDP jagannathan_s_Page_015.jp2
f45fc8c397420ee6f63c6fd8c38c8521
a068b7649b6edb39217e755c559597c73ede9960
6925 F20101108_AACCEF jagannathan_s_Page_062thm.jpg
42e79b7a4bf002740e6760fafe54ca15
ed182434a9accdba5106c919f0d39a1cb2fd2b15
8442 F20101108_AACCDQ jagannathan_s_Page_032.pro
7e7bc29f09a99a78ca2bf8f3953dab5c
affce17499356236fad8bc039616a894deb0277f
13762 F20101108_AACCEG jagannathan_s_Page_091.pro
634c7bbda3a13cbfbe1bb4f095acc47a
9526b2da73df27182c8a0e65bbdf7a16e3259199
725 F20101108_AACCDR jagannathan_s_Page_113.txt
5b5f78d8aa549f171aab70f6e32d99f5
6b0ab18d3a4a9c1de1bf29e57c32e1c5f5f52f81
F20101108_AACCEH jagannathan_s_Page_062.tif
198771dcc545a8a67af25e9449bd05e9
91a83adc3f59170dfccebaaf3635948022a0a380
28450 F20101108_AACCDS jagannathan_s_Page_016.jpg
2f72e2e5f99783ca9ac8e791b087b964
684403c8c36a11e43c8dd79e5a37a6f89d7d9ec5
7714 F20101108_AACCEI jagannathan_s_Page_001.QC.jpg
ab1590e7666407147878c1502b99f3f4
a9f768515c0b88983a72132414f9cae58d372f63
3795 F20101108_AACCDT jagannathan_s_Page_101thm.jpg
9ba56a264061add79d81fea0b4d0a950
da940f7d45189e74c570a0a8b54724ae87d1f0c9
119131 F20101108_AACCEJ jagannathan_s_Page_012.jp2
9c6f00bfefd866030b65263afe65ccf3
fd34d426c436552beccbca45b1d85b366fb4a6a5
13016 F20101108_AACCDU jagannathan_s_Page_114.QC.jpg
94b7713534279bdf0cc98d8a17799097
f983c7bc6c98ce421d42883cc317ed48a491d71a
70526 F20101108_AACCEK jagannathan_s_Page_031.jpg
fae4021093ed4f392ce4ad1f038ffd4b
dfb58b2bc159c17fefcb2acd06659aaafcc7f644
6941 F20101108_AACCDV jagannathan_s_Page_060thm.jpg
48004e40429f9b8229b9ad4926a48f80
575e2cb0f002fd4bfcafe2a751762f0216332a9e
113141 F20101108_AACCEL jagannathan_s_Page_060.jp2
521709d12f2183e38f80a272b130769c
993c46058e97c80ea89aba17e0e88653a4e13690
23638 F20101108_AACCDW jagannathan_s_Page_037.QC.jpg
28f9e02dbbb62aa99c6955277bcbf7a2
e456b43e00f5fa0371b7e650ea219d3a9d2b2da3
850 F20101108_AACCEM jagannathan_s_Page_099.txt
b82dea259bb31136d0acdde8fbc6178b
abc3d2658eea6d93f280a2a65fecf0686de527bb
1898 F20101108_AACCDX jagannathan_s_Page_050.txt
4771a252919903c936a62244b64e7ff7
3baa3458a0c34ff9d5e71cf6212aa0649953f073
F20101108_AACCFA jagannathan_s_Page_058.tif
8dd8fedceeb0f466984daa40d30be662
ae6b9c70a9cc2434430d14e1a32a094f527210db
20776 F20101108_AACCEN jagannathan_s_Page_004.QC.jpg
88995fe2ea4869e1ffefb14041a0f5eb
2d251d4389e3619474515cbc3bcdbe63be8305e5
5767 F20101108_AACCDY jagannathan_s_Page_013thm.jpg
0d4bc64cd8fbe3ecc524fd468fcd8546
af9727ca9fef190244a109d704bf8db1dc475ff5
3816 F20101108_AACCFB jagannathan_s_Page_090thm.jpg
7d71d0d7980dc64e497146ef799149b7
37ad1ff6b5e44557f719e35f9a8ffccf06a35df8
F20101108_AACCEO jagannathan_s_Page_039.tif
d808befde9527655dbc89d1840ea18f7
2a39e2fd2b190956b109fa620ae5f33b5e0fe18c
808 F20101108_AACCDZ jagannathan_s_Page_096.txt
3a701df2e3734ebea92b6988fb20442f
37db0fa1651439ad6b162daa80d0d9296b3a32d6
47989 F20101108_AACCFC jagannathan_s_Page_053.pro
a5f727803112dd7fb897f44cb1af2ab4
3909965c9133cb8eade8f4eec57a53f0d05f5880
68889 F20101108_AACCEP jagannathan_s_Page_049.jpg
f41eb24a36b832f7d769aece50dde6e4
81e39aae379070e3e918649e66860817356898ca
1692 F20101108_AACCFD jagannathan_s_Page_004.txt
7c8e8bb52e21e9836da157db9602c6e0
1980b5982d014aba7b5940a3ee188cd4f40027b0
13223 F20101108_AACCFE jagannathan_s_Page_089.QC.jpg
2154c6d5814c4b1c5a8bb44f69e1bac7
fcaae739d167b0a0688eedfc818cfe257a02385e
41483 F20101108_AACCEQ jagannathan_s_Page_004.pro
1bbb35ed6f072a0bb64e171225565242
8a962289a254877420e457607ab2a8fd7c22eaf0
545242 F20101108_AACCFF jagannathan_s_Page_076.jp2
7196d0073c829462aeb2ec3c537d1258
62ee2933c1f17b5012b2e20adc8fe34405dc2ad0
73463 F20101108_AACCER jagannathan_s_Page_040.jpg
a59f80b6963d3ebd298b63fefc8cdc1a
5ea58afa814c90e5fdd02083ef148f14371f4a4c
2504 F20101108_AACCFG jagannathan_s_Page_115.txt
4da88a8fb4140e635d60293c1c78a351
65d4a0fb26e73901dafdcd26f12eceb7941fbe86
135244 F20101108_AACCES jagannathan_s_Page_118.jp2
d5762d4e083873d7b64a10f1e3d3ac60
17e9b4aa5454499a96774c4829fb531925f7208c
54023 F20101108_AACCFH jagannathan_s_Page_039.pro
d9f4be163a475dcefdbbc42495de2225
2eea3cd5d846615f5bef2e9aee47d7ad0afb2b91
8593 F20101108_AACCET jagannathan_s_Page_035.QC.jpg
0e6e2c8ee5cf5808522fd42198c50aa5
47195db99101ceef20843cfd680cba0c10967465
21621 F20101108_AACCFI jagannathan_s_Page_005.QC.jpg
7946c0c91af1512337b5f167b3e969ea
b38a18ac9590bf45a5b11d11f8ea2b057963bc5c
F20101108_AACCEU jagannathan_s_Page_052.tif
741e76f0f54ad3d622d699c9719ba55e
4c8b6c351f114bf98b414af9a567ff1fb68eb142
869 F20101108_AACCFJ jagannathan_s_Page_088.txt
a964191b0e6c2f89e4c1402eef260fe1
d4b781ec98cf53fe967f10eff8758b3b0c127215
F20101108_AACCEV jagannathan_s_Page_074.QC.jpg
97dff3f41c9117d6f6d0a0333ba68f0c
843de74706817420d0caf6f1851a6b63bba0c5d9
3734 F20101108_AACCFK jagannathan_s_Page_114thm.jpg
9cd617dd79e7f897f71ff74aadc55a2c
8823167deae1638399fc3ca5be36ce0b01818f89
855 F20101108_AACCEW jagannathan_s_Page_070.txt
4bdc6b1fcfb56efd03dcf36f9d4c870b
1098f30db787b756d14be1a3fd7b976bc024f178
2092 F20101108_AACCFL jagannathan_s_Page_119.txt
859d2d34bbcb28d7e5d05c5147c73219
e5168778ba8edb417a14bdb4851b6ed23bdbd3cc
118108 F20101108_AACCEX jagannathan_s_Page_022.jp2
08afa9ba2430d96720265e89161f7f4c
20b382e9c7efc706877ec60fcf6624c74db13a5a
14738 F20101108_AACBZR jagannathan_s_Page_093.pro
e5489c7ace942f16b1bed04580dfc2c1
9acd6232a1c08cb5569790b4aaac9eb33f932fc0
76618 F20101108_AACCGA jagannathan_s_Page_043.jpg
048ec1817fb40b991a866a72e9608f91
943e41289c4d827177c807b532d4d16b26a5febf
526 F20101108_AACCFM jagannathan_s_Page_095.txt
51b01c958c8b381e61f804d869853dc1
2818cb9b59cc804a45f7ec6ba14d506b574a6554
75164 F20101108_AACCEY jagannathan_s_Page_023.jpg
ccd896a3023fab2e176d66a9ed4c913f
f30d88fe04fbdd7d4e18fa45100275c37f9683b0
38512 F20101108_AACBZS jagannathan_s_Page_051.pro
1a80f36ac599516df7105b4390907a73
fb297c768b52e4157fcf5cb367cd2b11d2b891ad
77621 F20101108_AACCGB jagannathan_s_Page_062.jpg
b15ecd69b9d08037d3ca6087e5b18529
0f1fa5214ad1e05e9b3ee0647c2d88b754ad4797
108407 F20101108_AACCFN jagannathan_s_Page_054.jp2
a2384516cde146c8c102eb3cdf39947e
8f83ebc36954f5e91f7735c6d0c4e2cf6b023dbc
4351 F20101108_AACCEZ jagannathan_s_Page_010thm.jpg
fbca7eea4e60cdabb669ee2cc62cbfde
6157ed7b39e1bb6426d5f13b91b60ed4166f85cb
6546 F20101108_AACBZT jagannathan_s_Page_045thm.jpg
ac0fc96af1bc607450ac3cf051d10e97
ab2468538a6513de135d30e7574f32c0ee28e2ea
24851 F20101108_AACCGC jagannathan_s_Page_017.QC.jpg
f8f1db5a8465acdb07ebac90d9232d8b
7a2b1c76b9ee391c17d110012665cbfaa3fc73c1
72012 F20101108_AACCFO jagannathan_s_Page_005.pro
8e642f7047dcf070c6ae90db58c54bf8
5efaa56a90379f14467db055d55fe5c1a48269d5
2163 F20101108_AACBZU jagannathan_s_Page_036.txt
55c6cf0582c79fcf9a9246c609f5a523
243877e2e149bb535184056a5c23c375cc775ddc
45392 F20101108_AACCGD jagannathan_s_Page_102.jpg
221c217e11d4360f8f647f23f8d25747
e316aad4df92bb1cf8ecc311e41dd85eae068179
37298 F20101108_AACCFP jagannathan_s_Page_016.jp2
3835776c7054594528d23a2d913a6e99
6c1229700614833eba05a0684a60e2efee953447
F20101108_AACBZV jagannathan_s_Page_031.tif
ec60d39dcb9a757e528624336a3a8eec
355a2ee6d9026a7431e9f834cf9da3b3f71291f2
18148 F20101108_AACCGE jagannathan_s_Page_087.pro
504a9bb13e3728813ae22beeb8b8b126
bab4072806f9541defbbb4404ce19447a67b9dbf
F20101108_AACCFQ jagannathan_s_Page_008.tif
062c26d3950bd5ce25e0b6e3d3dbba0d
e282d4cad62097bc802b7d2a4daa5618f7a97e50
F20101108_AACBZW jagannathan_s_Page_028.tif
04a2ce8377e4656a00294b6146a728ec
5a062772068cf7955567e4ff4d46ba38421dcdb3
52457 F20101108_AACCGF jagannathan_s_Page_025.pro
96221241507290761e291122453496a5
85860c39b8d922d16c1ab06fb70673d282b619f9
13121 F20101108_AACBZX jagannathan_s_Page_086.pro
b173d82d21acf3d9f64e1e52ee53481c
11a216b2b7e7887c272541a25b644b566f757f52
25166 F20101108_AACCGG jagannathan_s_Page_120.jpg
cd7e324ce686db7bf67a89388fc83b24
78c1b97b378288cb90eaa009e7c48dae09a25dac
27479 F20101108_AACCFR jagannathan_s_Page_117.QC.jpg
8f545b62d8b634e110820cd6f74f60ae
eaf554b8efd57fc286f8fe693690726a6e267fb4
70682 F20101108_AACBZY jagannathan_s_Page_038.jpg
1fc0373bdb818f75d3fbdafafb4cdc36
de92b8cccb7526fa9368df57522528698c8ba6a7
F20101108_AACCGH jagannathan_s_Page_033.tif
8a234bfa318c8d7574d29ecb0496d287
c957ac9b54688a8dd9cf57ba14e1641c59875c18
3948 F20101108_AACCFS jagannathan_s_Page_076thm.jpg
19772952b9923018dc768d4e2114c1e7
2867792146cdd53fb0873b8b68cadd691781200c
1051980 F20101108_AACBZZ jagannathan_s_Page_005.jp2
d899092d9ca93289e869d03ef0bdb00f
94ac1a6a4990a15d9f0017e3fdf0898ab52409c6
F20101108_AACCGI jagannathan_s_Page_083thm.jpg
f06384fe9e51343578d1674e184eeeee
cb37e4a6131c5c43367ff16f86e1f13ae302f0d1
41667 F20101108_AACCFT jagannathan_s_Page_094.jpg
56531e06bc40d1f1b616c542814cc482
0bac159bba395594ea2838d18e6f53bb66aae8f7
111359 F20101108_AACCGJ jagannathan_s_Page_045.jp2
966a4c2af435a1ee07e6fe203973bdff
9afebbbd9aac3e0bc0bfbcc8422303bcd96f567c
6899 F20101108_AACCFU jagannathan_s_Page_017thm.jpg
5c63cc0ba39ec19b576dfea14c7b276b
c5f8172b5e3b930d9f3bd35ff329dfdf2328d161
13281 F20101108_AACCGK jagannathan_s_Page_112.QC.jpg
08767c09dcd70b944e70a8c2d415628f
ef641f3dbee064a1cdfbd784517cb0f27a8eb626
2068 F20101108_AACCFV jagannathan_s_Page_019.txt
1807fbf0414e24ab5f497cf15041eaa4
598ffd27c994704d96d2a80482741125c090c2cc
564347 F20101108_AACCGL jagannathan_s_Page_094.jp2
a9cade9c7c44f894e79603d0918d4e54
22630b01f616e1f856fffd507fc36e11fa190639
1715 F20101108_AACCFW jagannathan_s_Page_013.txt
4d8ab1fc022632859e8f90a65b4c7bf3
5845fcc64daeeabae2b224dcc4918aeefe8cd8cb
23038 F20101108_AACCHA jagannathan_s_Page_049.QC.jpg
19dbd5959f450c4b54fed47f2349cf6c
c61c19f15dc086337de5756cb9453975ab0e8815
116844 F20101108_AACCGM jagannathan_s_Page_057.jp2
867a17d7c0ffd056d8ae7950e78b35b3
b776555166822b2a29d3736240b372ecab3f8b42
3723 F20101108_AACCFX jagannathan_s_Page_095thm.jpg
c682fe3ce0e3da9b4d2b90c5c2cb7c39
f8ccc180ee678a27888c1dcb1d4d92936547f802
24516 F20101108_AACCHB jagannathan_s_Page_079.jpg
55a41daa6cd076b30c3d0e8abb8c31c0
694fc256db1e09ba69792541a5704bb0519a9380
2248 F20101108_AACCGN jagannathan_s_Page_015.txt
6f964b9a0aa1ea0409ad26c6253cbe24
dee737022aee150f0d3133b88f42ea9a8deb0006
2186 F20101108_AACCFY jagannathan_s_Page_017.txt
bd0fcc31f195cf46da7bc492b5d3a950
cf27f9fbd6bb34ce8787db54c191e4fec5e2eaef
24350 F20101108_AACCHC jagannathan_s_Page_023.QC.jpg
a67a5afd5dda0e92645119facfe6dadc
42c2f24e12c2e977140a1af3135ef347f912cd30
739 F20101108_AACCGO jagannathan_s_Page_052.txt
88a738093d4160afcb43e37ecefc175a
1bc3e033e48536c18a170a4d97a02382540001af
46546 F20101108_AACCFZ jagannathan_s_Page_059.pro
4e8720b74e05e2d981311fad5b5fb638
462efeef34a993ef4ea05d80668c98658040270d
F20101108_AACCHD jagannathan_s_Page_022.jpg
c31d59384350ccdb5d15b99fb174cf56
289dab615e9795d28ce9ee6bb935ebe20c7d218e
F20101108_AACCGP jagannathan_s_Page_026.tif
441e7491e752625d49eae3bdf809fa4b
41d27848f1bce4aa9090ae06c8c4d74ced3fc401
12186 F20101108_AACCHE jagannathan_s_Page_092.QC.jpg
a2037b7b46941bd1b47f05be2119ee26
4e95381c3dcbc0c52d442ca758c2e62d931737cc
6936 F20101108_AACCGQ jagannathan_s_Page_046thm.jpg
1e729f9d34ac4024537b79747b4a7364
0379703909f7342946df6967938d4b705fb86ae2
F20101108_AACCHF jagannathan_s_Page_010.tif
054060abab8ec7e2f1d1337cd7390d19
f2fba45007ac166186e1f249df97ee220d50d572
F20101108_AACCGR jagannathan_s_Page_006.tif
6278d8ee7cba6e931ecd342d2b367a57
2aa083063c8e3a4eeb68172558f6e5d2925e287c
27404 F20101108_AACCHG jagannathan_s_Page_035.jpg
d3b1aa96b96d37ae84724c629d15918d
9f27ae75101ec4b7f2e8d1e8a02a53cb523b3664
79758 F20101108_AACCHH jagannathan_s_Page_028.jpg
c45fe656ef4ec17f0363f58413b3fbdc
fe55548c7c76e38a364f3ef84d6422f3da792bbd
F20101108_AACCGS jagannathan_s_Page_048.tif
162a315ddd7e4be514a6bc1963acda0e
407e2473b2ca5c7bcc4a342eb912f15d91162236
3888 F20101108_AACCHI jagannathan_s_Page_110thm.jpg
d8d38c16d7350e7c231c0942fdc1c710
bf3a76a62fe51c0ea0d1efb875629f4b4ec6035b
1666 F20101108_AACCGT jagannathan_s_Page_066.txt
b5948967c0b0eed92a40ed288e600eca
d886def2bbd8d3d086bcbd073874454209fa5699
2048 F20101108_AACCHJ jagannathan_s_Page_060.txt
534ae423d8f0bbebb9fe360455a3cccf
aeedc15ec5fc631b6b9e536eccffa838f722f259
674 F20101108_AACCGU jagannathan_s_Page_101.txt
68f6c94993434b55b5d7057b24af409a
4134d489a1be81a074fdc0b82bd8b095101d0311
26250 F20101108_AACCHK jagannathan_s_Page_028.QC.jpg
8b88e644d7e44f54ce570994d5a6cc94
95588748b6010f771f4b4193066c04f5cb2d2bf3
1985 F20101108_AACCGV jagannathan_s_Page_021.txt
eca440e0d91c2d70eb9ba310eaa143f5
6f45d01c3849c377027bc459d58468def5cab38c







DEVELOPMENT OF A MANAGEMENT FOCUSED DECISION SUPPORT TOOL FOR
OKEECHOBEE BASINT BEEF CATTLE AGROECOSYSTEMS




















By

SUDARSHAN JAGANNATHAN


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA

2007

































O 2007 Sudarshan Jagannathan






























To my father for all the support he has always shown me; to my mother for being a tremendous
influence in my life, I would like to dedicate my thesis and all my achievements at the University
of Florida. Her memory and blessings have brought me this far.









ACKNOWLEDGMENTS

I am greatly indebted to Dr. Gregory A. Kiker for his complete support of my research

interest in the field of Agricultural and Biological Engineering. I am sincerely thankful to him for

his enthusiastic and generous support which helped me learn the subj ect and his continued

encouragement of my model development. His genuine interest in his work and the people he

works, with have earned my highest regard. Without his support and tireless help, it would not

have been possible for me to have successfully completed my research. I would like to thank Dr.

Rafael Munoz Carpena and Dr. Clyde Kiker for serving on my committee and taking the time to

meet with me and share their valuable insights about my research. I would like to acknowledge

the immense value of the support and help of Dr. Chris J. Martinez, for the Buck Island Ranch

data sets he provided and for his valuable inputs in the model development process.

I would like to thank Mr. Gregory Hendricks for his time and for sharing his knowledge of

hydrology and ecology with me, and Ms. Thej aswini Somasundaram for her support and

encouragement throughout the course of this program. I would like to also sincerely thank the

faculty and staff of the department of Agriculture and Biological Engineering for their technical

and moral support. I would also like to express my thanks to all my friends and colleagues for

their constant encouragement of all my endeavors.

Most importantly, I would like to thank my father for his support and encouragement, and

the belief he has always had in me.












TABLE OF CONTENTS



page

ACKNOWLEDGMENTS .............. ...............4.....


LIST OF TABLES ................. ...............7.__. .....


LIST OF FIGURES .............. ...............8.....


AB S TRAC T ............._. .......... ..............._ 1 1..


CHAPTER


1 INTRODUCTION ................. ...............14.......... ......


2 LITERATURE REVIEW ................. ...............17................


Study Site: Buck Island Ranch ................ ...............17........... ...
Experimental Pastures ............... ...............18....
Overview of Past and Current Models ................. .. ............. ...............19....
A Brief Summary of Hydrological and Nutrient Models ................. .......................20
Overview of Forage Models............... ...............22.
Obj ect Oriented Systems Development ................. ....___ .....__ ...........2
Decision Support Systems .............. ............ ...........2
Introduction to Questions and Decisions QnD ....._.__._ ..... ... .__. ......._.........2

3 METHODOLOGY .............. ...............36....


Design of an Enterprise-Level Model for Buck Island Ranch QnD:BIR..............._._. .........36
QnD:BIR Hydrology ........._.___..... .___ ...............37....
QnD:BIR Nutrients .............. ...............39....
QnD:BIR -Forage Growth ................ ...............40........... ...
QnD:BIR Beef Cattle Management ................. ...............42...............
Breeding .............. ...............42....
Calving .............. ...............43....
W meaning and selling ................. ...............44................
Cow intake and waste............... ....... .............4

QnD:BIR Ranch Incomes and Expenditures ................. ...............45........... ..
Addition of New Features into the QnD Model .............. ...............47....
Model Calibration, Validation and Data Representation ................. ......... ................49
M odel Calibration............... ..............4
Model Validation............... ...............5
M odel Evaluation .............. ...............50....
Statistical Representation .............. ...............50....


4 MODEL TESTING, RESULTS AND DISCUS SION ................. .............................53












Model Inputs ................. ...............53.................
M odel Outputs .............. ...............55....
Model Calibration ................. ...............55.................

Hydrology ................. ...............55..._._._ ......
N utrients .............. ...............56....

Forage ............... ... ...............56.......... ......
Model Validation/Testing ................. ...............56.......... ......
Hydrology ................. ...............56.......... ......
Phosphorus Load .............. ...............60....
Forage Growth ................. ...............63.................
Summary of Model Testing ................. ......... .. ...............64.....
Enterprise Wide Simulations and Scenario Analysis .............. ...............64....
Scenario 1: Measured Rainfall .............. ...............65....
Scenario 2: Low Rainfall ................. ...............65...............
Scenario 3: High Rainfall .............. ...............66....
Results and Discussion ................. ...............66........... ....


5 CONCLUSION AND FUTURE WORK .............. ...............80....


Conclusions............... .... ..... ........8
Future Research Recommendations ................... ........... .. .......8

Integrate Future Climate Predictions and Analyzing Different Scenarios ................... ...81
Improvement of the Cattle Production Module ................. ...............................82
Integration of a More Complex Model into QnD ................. ..... ........... ................. ...82
Integration of a More Advanced GIS Application Programmable Interface for Java.....82

APPENDIX


A MODEL RE SULT S AND GRAPH S .............. ...............83....


LIST OF REFERENCES ................. ...............115................


BIOGRAPHICAL SKETCH ................. ...............120......... ......










LIST OF TABLES


Table page

A-1 List of values of Nash-Sutcliffe coefficient (Ceff), Normalized Mean Square Error
and Root Mean Square Error (in million liters) for runoff in summer pastures. ...............67

A-2 List of values of Nash-Sutcliffe coefficient (Ceff), Normalized Mean Square Error
and Root Mean Square Error (in million liters) for runoff in winter pastures. ..................67

A-3 List of values of Nash-Sutcliffe coefficient (Ceff), Normalized Mean Square Error
and Root Mean Square Error (in tons Kgs) for load in summer pastures. ................... ......68

A-4 List of values of Nash-Sutcliffe coefficient (Ceff), Normalized Mean Square Error
and Root Mean Square Error (in million liters) for runoff in winter pastures. ..................68










LIST OF FIGURES


Figure page

2-1 Buck Island ranch with its summer and winter experimental pastures..............._._...........32

2-2 A simplistic UML look at the different components of QnD model ............ .................33

2-3 Role of the coder or the code developer ................. ...............33........... .

2-4 The developer's role ................... .................. ................ ...............34

2-5 The interaction of the players is minimal with the j ava code or the XML files............_..34

2-6 The class diagram of a typical QnD system ................. ...............35......_.. .

4-2 Monthly runoff in the SummerS pasture. ................ ............... ......... ........ ..69

4-3 Cumulative runoff in the SummerS pasture ................. ...............69........... ..

4-4 A measured vs predicted scatter graph for the SummerS pasture runoffs ..........................69

4-5 Monthly runoff in the Summer4 pasture. ................ ............... ......... ........ ..70

4-6 Cumulative runoff in the Summer4 pasture ................. ...............70........... ..

4-7 A measured vs predicted scatter graph for the Summer4 pasture runoffs ..........................70

4-8 Monthly runoff in the Winter4 pasture ................ ...............71........... ..

4-9 Cumulative runoff in the Winter4 pasture ................. ............... ......... ....... ..71

4-10 A measured vs predicted scattergraph for the Winter4 pasture runoffs ................... ..........71

4-11 Monthly runoff in the Winter3 pasture ................ ...............72........... ..

4-12 Cumulative runoff in the Winter3 pasture ................. ............... ......... ....... ..72

4-13 A measured vs predicted scattergraph for the Winter3 pasture runoffs ................... ..........72

4-14 Monthly phosphorus load in the Summer8 pasture .............. ...............73....

4-15 Cumulative phosphorus load in the Summer8 pasture .............. ...............73....

4-16 A measured vs predicted scattergraph for the Summer8 pasture Ph loads ................... .....73

4-17 Monthly phosphorus load in the Summer3 pasture .............. ...............74....

4-18 Cumulative phosphorus load in the Summer3 pasture .............. ...............74....












4-19 A measured vs predicted scattergraph for the Summer3 pasture Ph loads ................... .....74


4-20 Monthly phosphorus load in the Winter4 pasture ................. ...............75..............


4-21 Cumulative phosphorus load in the Winter4 pasture ..........._...__......... ................. 75


4-22 A measured vs predicted scattergraph for the Winter4 pasture Ph loads ..........................75


4-23 Monthly phosphorus load in the Winter8 pasture ................. ...............76..............


4-24 Cumulative phosphorus load in the Winter8 pasture ................. ............................76


4-25 A measured vs predicted scattergraph for the Winter8 pasture Ph loads ..........................76


4-26 Monthly forage yield for the Summer 1 pasture ................. ...............77........... .


4-27 Monthly forage yield for the Summer 8 pasture ................. ...............77........... .


4-28 Monthly Totals of Enterprise Wide simulation results ................. .......... ...............79


A-1 Summer 1 Runoff ................. ...............83................


A-2 Summer 2 Runoff ................. ...............84................


A-3 Summer 3 Runoff ................. ...............85................


A-4 Summer 4 Runoff ................. ...............86................


A-5 Summer 5 Runoff ................. ...............87................


A-6 Summer 6 Runoff ................. ...............88................


A-7 Summer 7 Runoff ................. ...............89................


A-8 Summer 8 Runoff ................. ...............90................


A-9 Winter 1 Runoff ................. ...............91................


A-10 Winter 2 Runoff............... ...............92.


A-11 Winter 3 Runoff ................. ...............93.......... .....


A-12 Winter 4 Runoff ............... ...............94.


A-13 Winter 5 Runoff............... ...............95.


A-14 Winter 6 Runoff ............... ...............96.


A-15 Winter 7 Runoff ............... ...............97.











A-16 Winter 8 Runoff............... ...............98.


A-17 Summer 1 Phosphorus Loads................ ...............99.


A-18 Summer 2 Phosphorus Loads ................. ...............100........... ...

A-19 Summer 3 Phosphorus Loads ................. ...............101........... ...


A-20 Summer 4 Phosphorus Loads ................. ...............102........... ...

A-21 Summer 5 Phosphorus Loads ................. ...............103..............


A-22 Summer 6 Phosphorus Loads ................. ...............104........... ...

A-23 Summer 7 Phosphorus Loads ................. ...............105........... ...


A-24 Summer 8 Phosphorus Loads ................. ...............106........... ...

A-25 Winter 1 Phosphorus Loads ................. ...............107.............


A-26 Winter 2 Phosphorus Loads ................. ...............108.............

A-27 Winter 3 Phosphorus Loads ................. ...............109.............


A-28 Winter 4 Phosphorus Loads ................. ...............110.............

A-29 Winter 5 Phosphorus Loads ................. ...............111.............


A-30O Winter 6 Phosphorus Loads ................. ...............112....... ....

A-31 Winter 7 Phosphorus Loads ................. ...............113....... ....


A-32 Winter 8 Phosphorus Loads ................. ...............114.............









Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

DEVELOPMENT OF A MANAGEMENT FOCUSED DECISION SUPPORT TOOL FOR
OKEECHOBEE BASINT BEEF CATTLE AGROECOSYSTEMS

By

Sudarshan Jagannathan

December 2007

Chair: Gregory A. Kiker
Major: Agricultural and Biological Engineering

Agricultural enterprises require resource management that involve many trade-offs within

a complex ecological and financial environment. As an example enterprise within South Central

Florida, the MacArthur Agro-Ecology Research Center (MAERC) located on the Buck Island

Ranch (BIR) in Lake Placid Florida has a maj or obj ective to optimize its long term sustainability

in both ecological and economic facets. MAERC/BIR combines a research facility with a

commercial-scale, beef cattle enterprise (10,300 acres) to explore the role of long-term

ecological and social dynamics within sub-tropical grazing systems (www.maerc.org).

In order to maintain long term viability and sustainability, a balance between ranch

profitability and reduction of non point source pollution effects needs to be established and

studied. A possible solution to this challenge is to create a Decision Support System (DSS) for

beef cattle enterprises. Such a DSS could serve to communicate simulation results and metrics

effectively to the ranch operators, whose focus would be on profitability, as well as the

researchers and conservationists, whose focus would be on limiting the effects of non point

source pollution. Thus, the objective of this research project is to design and construct a decision

support model of a beef cattle ranch system to simulate selected beef cattle and ranch










management operations on a southern Florida beef cattle enterprise and to explore the

management decisions with respect to water resource factors such as runoff and nutrient loading.

The Questions andDecisions TM (QnDTM) model system was created to provide an

effective and efficient tool to integrate ecosystem, management, economic and socio-political

factors into a user-friendly model/game framework. This model is a unique and new

development since no other model before has modeled scenarios on a ranch-scale. The model is

also good in that it is more than just a hydrological model but also a decision support tool for

managers with a user interface that helps them in real-time decision making. The QnD model

links spatial components within geographic information system (GIS) files to the abiotic

(climatic) and biotic interactions that exist in an environmental system. QnD can be constructed

with any combination of detailed technical data or estimated interactions of the

ecological/management/social/economic forces influencing an ecosystem.

The specific QnD version has been developed for the BIR (QnD:BIR) using the conceptual

diagram which shows the integrated ecological and economic factors at the ranch-scale.

QnD:BIR uses elements of the Standardized Performance Analysis (SPA) method applied to BIR

to simulate elements of beef cattle production and economic dynamics. QnD:BIR uses

simplified water and phosphorous dynamics at a monthly time step generated from the long term

research from southern Florida beef and dairy cattle research. QnD:BIR utilizes existing

geographic information systems (GIS) coverages and monitoring data available from the

MAERC/BIR facility. QnD:BIR was tested on environmental data from BIR for the period of

2000 2003 for sixteen experimental pastures including both improved and native pastures.

Specifically, QnD:BIR simulation results of monthly runoff, phosphorus load and forage

production were compared with comparable field-scale data. Given the coarse monthly time










step, simulations of these factors were generally acceptable for use in the whole ranch

simulations. Given potential climate data for the area, specific scenarios were constructed to test

different management scenarios in terms of P loading and cattle production metrics. The

development of QnD: BIR provides a useful and modular system, capable of running various

scenarios depending on the setup for simulating both environmental and enterprise functions,

within an easy to use graphical interface with the ability to move cows and manage the enterprise

hands-on. Further model development and simulation could be expanded to allow more detail in

cattle response to temperature and surface water availability.

Qnd: BIR is a simple model that uses empirical relations with acceptable levels of accuracy

(and a Nash-Sutcliffe coefficient of at least 0.5 mostly). The model also takes into account

rainfall, water table depth, temperature, and soil characteristics for its hydrology and phosphorus

cycle.

However, since the model uses empirical relations, it cannot be applied in conditions that

differ vastly from the conditions present in BIR. Also, due to its simple nature it does not take

into account factors such as light, for ET, or drainage within and across pastures in the ranch and

into the canals.

Considering the significant positive qualities and certain limitations of the model, it can be

said that the model is to be used more as a guideline to point the manager in the right direction

for decision making than as a tool to provide exact values or measures for runoff or phosphorus

load in the long term.









CHAPTER 1
INTTRODUCTION



Management of agricultural enterprises often occurs within the context of complex

environmental and societal challenges including elements of economic, management and

political viewpoints as well as the often-explored technical perspectives. The quality of these

agro-ecological systems is significantly affected by the rapid growth in the state's population

over the last three and a half decades and concerns over non-point source pollution (2006

Integrated Water Quality Assessment Report-FDEP, 2006).

As an example, beef cattle operations in south central Florida are concerned with long-term

sustainability and viability under increasing regulatory pressures. Adding to this existing

challenge is the uncertainty of climate and environmental drivers to agroecosystems. Decision

Support Systems (DSS) linking water resources and agriculture should be cognizant of the

intersecting and sometimes conflicting goals of profitability, non-point source pollution effects

and adaptive management. Recent advances in the climate sciences, including improved

capabilities to forecast seasonal climate, have provided increased capabilities for developing

useful decision support tools in service of specific agriculture, forestry, and water resources

management (SECC, 2007). These decision tools should communicate simulation results to

decision-makers or stake-holders in their own language and metrics whenever possible. Three

fundamental questions arise when investigating water resource management within beef cattle

enterprises.

1. What decisions are currently made that effect runoff, water quality on beef cattle
ranches?

2. How accurate do forecasts have to be to be useful for beef cattle operations and
environmental regulators?









This research addresses some of these questions by exploring the role of management-

focused, agro-ecosystem models. The obj ectives of this research proj ect are the following:

1. Design and construct a decision support/scenario-based model of a beef cattle ranch using the
QnD model system to simulate selected beef cattle and range management operations on a
southern Florida beef cattle operation.

2. Test and calibrate the model using climate, hydrology, soil and forage monitoring data from
representative pastures.

3. Explore the impact of scenarios on management decisions with respect to water resource
factors such as runoff, nutrient loading, calf production and basic revenue/cost dynamics.

The MacArthur Agro-Ecology Research Center (MAERC) located on the Buck Island

Ranch (BIR), Lake Placid Florida provides a unique setting for production-related, agro-

ecological research. MAERC/BIR combines a research facility with a commercial-scale, beef

cattle enterprise (10,300 acres) to explore the role of long-term ecological and social dynamics

within sub-tropical grazing systems (www.maerc.org). Recent multi-disciplinary research efforts

at Buck Island Ranch (Swain et al., 2007; Arthington et al., 2007; Tanner and Mc Sorely, 2007;

Capece et al., 2007) have provided a useful dataset of climate, hydrological, nutrient, vegetation,

herbivore and production/economic data for further integration and model development.

Organization of this Thesis

In addition to the introduction, this thesis research is divided into four chapters and a

technical appendix. Chapter 2 provides a review of several items relevant to the research: an

overview of the Buck Island Ranch monitoring effort; a brief review of the existing hydrological

and nutrient modeling approaches that have been used in similar environmental systems; a

detailed look into some of the related obj ect oriented and decision support concepts and an

overview of the QnD modeling system. In Chapter 3, a detailed design of the QnD Modeling

effort within Buck Island Ranch has been described along with the methodology for model









calibration and testing. An analysis of the model's structure and the methodology used to model

the given system is presented, along with the results of model calibration and testing. The

following chapter (Chapter 4) presents the results of QnD model testing on specific pastures

(improved and native) within BIR. Chapter 5 provides an overall summary and lessons learnt

from the modeling effort as well as potential next steps. The technical appendix provides

additional results and background for greater clarification in terms of obj ect designs and model

performance on specific pastures.









CHAPTER 2
LITERATURE REVIEW

This chapter provides a review of the concepts and background used in developing a

decision support system for a beef cattle enterprise. An overview of Buck Island ranch, the

research site is given in order to help put things into perspective, along with a brief history of a

sample set models that have been developed to date. Following this review, a section is provided

that helps the reader to understand the need for a more obj ect oriented model and the obj ect

orientation concepts, which are discussed in detail. The obj ect oriented review section is

followed by a detailed analysis of the QnD model and its overall structure, hence providing for

an overall foundation of the research proj ect.

Study Site: Buck Island Ranch

Non-point source pollution is a maj or cause of concern in the south Florida water systems.

Lake Okeechobee is one of the largest and most important water bodies of this region and a very

popular site for research studies analyzing the effect of non-point source pollution on a typical

beef cattle agro-ecosystem (Martinez 2006, Yang 2006, Pandey 2007). According to Pandey

(2007), the MacArthur Agro-ecology Research Center, Buck Island Ranch (BIR, 4168, ha), Lake

Placid, Florida, USA (270 09'N, 810 12'W), shown in Figure 2-1, was chosen for a research site

since it is representative of the subtropical, wet-prairie agro-ecosystem that exists in the

Okeechobee watershed. For the proposed obj ective of this research too, this site would be well

suited. Buck Island ranch presents a diverse and rich ecological setting, due to its coupling of

sparse forests and wetlands in the midst of a commercial cattle ranch owned by the John D. and

Catherine T. MacArthur Foundation (Arthington et al., 2007; Swain et al., 2007). The ranch area

itself is at the center of a tributary basin of Lake Okeechobee, the Indian Prairie/Harney Pond

Basin, which has been drained for better pasture over several years (Arthington et al., 2007;









Swain et al., 2007, Pandey 2007). This ranch enterprise has been the center of research for

almost a decade due to the willingness and interest of the ranchers themselves to limit the effects

of non-point source pollution on the Lake Okeechobee watershed.

A maj ority of the current and past research proj ects at Buck Island ranch have focused on

the experimental pastures (Summer 1- 8, Winter 1-8) and to evaluate the effectiveness of the

Best Management Practices (BMP's) to control the non-point pollution effects(Pandey 2007,

Yang 2006). However, the BMP's mainly focus on reducing phosphorus loads in the region, and

since BIR is primarily a commercial cattle ranch, one of the main management goals is to

optimize the amount of beef production per unit area of land, or in other words improve the

productivity of the enterprise. This trade-off balance tends to be the responsibility of the ranch

managers to maintain. Cattle rotation is one of the practices that are commonly done at the ranch

in order to maintain optimal feed for the cattle. The rotations are designed such that a cattle herd

spends the maj ority of its time on the improved summer pastures (May October), where the

forage available is Bahiagrass (Paslpalum notatum), which has a higher forage quality. Pandey

(2007) explains the movement of cattle to be done due to two reasons : firstly, summer pastures

are fertilized (NH4NO3 56 Kg N/ha) (Arthington et al., 2007) in spring and, therefore, have

better forage quantity and quality compared to winter pastures which have never been fertilized

(Swain et al., 2007). Secondly, winter pastures are less intensively drained and as a result they

are regularly flooded during the rainy season in summer. For the stocking rate experiment of

1998 2003 (Swain et al, 2007), the summer and winter experimental pastures were chosen to

gauge the effect of different cattle stocking on the environment.

Experimental Pastures

The pastures in the ranch are divided into seasonal pastures depending upon where the

cows are placed seasonally. The seasonally demarcated pastures fall mainly into two categories:










summer pastures which tend to be improved, and winter pastures which tend to be native grass

pastures.

The summer pastures are eight approximately 20 ha (range = 19.0 to 22.1 ha) pastures

where in the cattle are stocked during the summer months. These pastures are well drained and

the maj or forage growth in this region is Bahiagrass (Paspalum notatum). Bahiagrass is generally

considered to be higher in nutritional value than the native species of grass that grows the winter

or unimproved pastures (Kunkle 2001).

The winter pastures similarly are eight experimental pastures of approximately 32.2 ha

(range = 30.3 to 34.1 ha). The forage on these pastures is not regulated and hence Bahiagrass is

not the maj or forage species in the region. Compared to the summer pastures, the winter pastures

are not as well drained and retain water during large rain events.

A network of shallow surface canals and drains carry all of the runoff water into the

Harney Pond Canal and then onward into Lake Okeechobee. As a result, monitoring these

pastures is essential to control the addition of phosphorus to the lake.

Overview of Past and Current Models

Buck Island ranch has been the obj ect of multiple research studies involved in hydrological

and nutrient model development. Many scientists and students from the University of Florida and

other institutions have conducted several research studies at the site. Stocking rate experiments

conducted during 1998 2003 have lead to several studies being conducted based of the data

collected from the experiment.

One such study involving the integration of Ranch Forage Production, Cattle Performance,

and Economics in Ranch Management Systems by Arthington et al, (2007) indicated that

stocking rates had a large effect on total production and profitability. However, stoclong rates

had minimal to no effect on forage utilization or cattle performance. With no offsets in improved









calving percentages, weaning weights or other measures of livestock performance, the inevitable

outcome of lower stocking rate is impaired profit potential (Arthington et al., 2007). Overall

changes in stocking rate has a direct, one-to-one relationship with ranch revenues. If stocking

rate effects surface water quality, there is a tradeoff between water quality improvement and

profits from breeding cows.

Moreover, an analysis of the soil phosphorus, cattle stocking rates, and water

quality for the region (Capece et.al., 2007) indicated better effectiveness of approaches focused

on decreasing phosphorus inputs and decreasing movement of accumulated soil phosphorus into

surface runoff would be more effective than approaches focused cattle management for reducing

P loads in surface runoff from cattle pastures. It also showed that the stocking rate had no

measurable effect on nutrients in surface runoff during 5 years of stocking treatments (Capece

et.al., 2007).

A Brief Summary of Hydrological and Nutrient Models

This section provides a history of agricultural modeling and presents an overview of a few

of the past and current models that have been developed for various landscapes in an attempt to

help familiarize readers with the history of modeling and put into better perspective this

modeling effort. Over the last few decades a variety of ecological and biological models have

been developed. The obj ective of this research entails the study of hydrology, nutrient movement

and forage growth, along with beef cattle enterprise management, and a few modeling efforts

focused on some of these fields are looked at in this section.

The Hydrological models can be classified on a scale ranging from distributed physical -

based to the lumped conceptual models. In the early 1970s the U.S. Environmental Protection

Agency (EPA) began sponsoring a series of water quality models in response to the Clean Water

Act, hence a maj ority of hydrological models used presently were developed during this time.









Early conceptual hydrological models used a representation of basic laws of hydrology

using differential equations and empirical algebraic equations for modeling different processes

(Yang, 2006). Some of the more popular models include Stanford Watershed Model (Crawford

and Linsley, 1966), the SSARR (Streamflow Synthesis and Reservoir Regulation) model

(Rockwood et al., 1972), Sacramento Soil Moisture Accounting Model (Burnash 1973), the HBV

model (Bergstrom 1976), the tank model (Sugawara et al., 1976), the Xinanjiang model (Zhao et

al. 1980), HEC-1 (Hydrologic Engineering Center, 1981) and the HYMO (Williams and Hann,

1983), CREAMS (Chemicals, Runoff, and Erosion from Agricultural Management Systems)

(Knisel, 1980) and the CREAMS derived GLEAMS (Groundwater Loading Effects of

Agricultural Management Systems) (Leonard et al., 1987). Recently, the dynamic changes in the

modeled areas are being explained by study in conceptual models of soil depletion, redistribution

and moisture replenishment. (Arnold and Fohrer, 2005, Yang 2006).

The next class of models is more physically-based than the lumped conceptual models,

which enable the more detailed representation of the physical watershed and hence require

simpler structure and fewer parameters. Some examples of semi-distributed physically based

models include SWAT (Soil and Water Assessment Tool) (Arnold et al., 1993) SWIM (Soil and

Water Integrated Model) (Krysanova et al., 1998, 2005), AGNPS (Agricultural Non- Point

Source pollution model) (Young et al., 1989), TOPMODEL (a TOPography based hydrological

MODEL) (Beven and Kirkby, 1979) and ANSWERS (Areal Nonpoint Source Watershed

Environment Response Simulation) (Beasley and Huggins, 1980), SDSM model (Singh 2001)

with their application varying from examining water quality to assessing the effectiveness of

BMP's on runoff and nutrient loads of the given watershed.









A third class of models is the completely physically-based distributed models. Some

examples of physically based models are MODFLOW (McDonald and Harbaugh, 1988), MIKE

SHE (Refsgaard and Storm, 1995) and GSSHA (Ogden 2001).

Overview of Forage Models

Forage models have been developed through time with varying complexities, ranging from

a simple empirical model like the Miami model (Leith 1975) to the more complex physiological

models like the DSSAT (Jones 2003). The model developed mainly depends on the location for

which it was developed and the types of forage it was dealing with. Yang (2006) has reviewed

the structure of additional forage models such as PAPRAN (Seligman 1981), CENTURY model

(Parton et al. 1987) ELM (Innis, 1978), ERHYM-II (White, 1987), GEM (Hunt et al., 1991),

CCGRASS (Verberne, 1992) and GEMT (Chen and Cougheour, 1994) and the Hurley Pasture

Model (Thornley and Cannell, 1997).

All the models discussed in the above section were used extensively in their time and are

applicable today. However, as the depth and computational requirements of modeling complex

ecosystems increases, the technical competence required by the model also compounds

exponentially. This raises the need for a complete rewrite of the program code for any and every

small change in the conditions or a change in the site on which it is being applied. A requirement

arises for a system that is capable of adapting to highly complex processes that might change

from one location to the other without having to rewrite the model, i.e. the issue of portability.

Object Oriented Systems Development

Obj ect oriented systems are beginning to develop as a practical solution to the issue of

model reusability. In recent times, the complex and highly computational model development is

turning to obj ect orientation concepts to better develop the model and at the same time keep the

design simplistic for example the ACRU 2000 (Kiker et al. 2006). The most recent version of the









ACRU (Agricultural Catchments Research Unit) (Schulze et. al, 1989) was reconstructed to be

an obj ect oriented model that described its landscape in terms of Components, Processes and

Data. The ACRU 2000 (Kiker et al. 2006) is developed in Java and has proven to be highly

extendable. Several modeling efforts have been conducted using this model in the south Florida

region and the research site of Buck Island Ranch. (Martinez 2006; Yang 2006; Pandey 2007).

The following discussion highlights the various facets of object orientation and its

application in the world of modeling. The use of obj ect orientation in modeling increased in the

early 1990's with several models being developed using the concept (Matsinos et al 1994, Mooij,

1996).

Obj ect Oriented Programming (OOP) originated as a development platform for physical

modeling in Simula-67 programming language. However, in the mid 1990's, it developed as the

dominant programming methodology, largely due to the influence of C++. In the past decade,

with the rise in popularity of Java programming language, the use of OOP concepts has become

more common, perhaps more importantly because of its implementation using a virtual machine

that is intended to run code unchanged on many different platforms. This feature of portability of

code is also being introduced now by Microsoft into the .Net framework.

The "magic" quarks (Armstrong, 2006) of obj ect orientation exists in the main components

of it namely, inheritance, encapsulation, polymorphism and abstraction along with obj ects,

instances and classes. The following sections provide some additional detail of these concepts in

an attempt to understand their application within environmental decision support systems.

A class is the basic unit in OOP. Classes are real world groups, which interact with each

other through relationships. A class can be a species, a group, any unit that has common









properties and implements common processes (Robson 1981, Rosson 1990). It can also be

defined as a set of obj ects that share a common structure and common behavior (Booch 1994).

An obj ect, on the other hand, is an instance of a class (Booch 1994) and can be anything

from fish to cows to grass to anything that is being modeled. It is important to understand that an

obj ect refers to the individual but not the whole group. It is further explained by Armstrong

(2006) as an individual, identifiable item, either real or abstract, which contains data about itself

and descriptions of its manipulations of the data. An obj ect of a class has all the properties of that

class and all of its parent classes. For example, an individual cow has all the properties and

processes of the cow species class as well as the mammal class which would be the parent class.

This concept of parent and child classes brings us to the first property of OOP, inheritance.

Inheritance was introduced as a part of the development of OOP in 1967 in the Simula

programming language. (Dershem, 1995). Some literature also inclines towards the idea that

inheritance is the only unique feature introduced by OOP (Henderson-Sellers 1992). Inheritance

has been defined as a mechanism by which obj ect implementations can be organized to share

descriptions (Wirfs-Brock, 1990) and also by Armstrong (2006) as a mechanism that allows the

data and behavior of one class to be included in or used as the basis for another class. Inheritance

signifies the property that any given class can be derived from another class, and it in turn

'inherits' all of the properties and the processes of that parent class. This property continues all

the way to the highest level of the hierarchy (Budd, 1991; Silvert, 1993). Such a hierarchical

structure ensures that only very specific properties need to be specified for each individual obj ect

and it inherits most of the higher level properties from its parent class. This reduces the

complexity of the code and increases the simplicity of the design itself (Moooij and Boersma,

1996).









Another important reason for the simplicity of design of obj ect oriented models is the

encapsulation property of OOP. It is described as a process used to package data with the

functions that act on the data or more commonly as a property that hides the details of the

obj ect' s implementation so that clients access the obj ect only via its defined external interface

(Wirfs-Brock 1990). Encapsulation is the containment of all of the processes and properties

required and performed by the obj ect of a class, within that class or its parents. This reduces the

coding overhead on the obj ect. For example, a cow or a fish having all of its properties and

processes within itself and allows another obj ect to make it perform a given process at a given

time. Encapsulation property makes such simplicity possible.

Polymorphism was used in software development and originates from it (Armstrong 2006).

She also goes on to indicate that the literature appears to inconsistently apply the concept of

polymorphism with some likening polymorphism to late binding or dynamic binding (Byard et

al, 1990). Bringing together these conceptualizations, Armstrong (2006) defines polymorphism

as the ability of different classes to respond to the same message and each implement the method

appropriately. In modeling terms, polymorphism as be explained by the use of an example of

feeding, which signifies grass to a cow but hunting game for a lion. The same feed process can

be used to initiate both processes with the respective classes executing the corresponding

processes of their species.

Data abstraction originated in the 1950's and is commonly defined as the property of OOP

to simplify complex real life situations by suppressing irrelevant details (Henderson-Sellers

1992, Ledgard 1996, Yourdon 1995).

With the knowledge of OOP also comes the need to know the advantages and

disadvantages of it. The obj ect orientation has several ups and downs discussed in detail in the









literature (Johnson 2000), but the most important fact about obj ect orientation is the ease of

model development. Moreover, the portability, reusability and the extensibility of the code are

the most appealing facet of obj ect orientation to the modeling world (Johnson 2000). The ability

to apply and use the model developed for one ecosystem and specific conditions and to easily

change the model components to adapt to a totally different set of conditions is the ideal scenario

for model design and code development.

However, intended users of this model are the ranchers who are responsible for making the

decisions and managing the operations of the beef cattle enterprise. This further enhances the

requirement for a link between the output of the complex hydrological and nutrient models and

the decision making capability of the ranchers. The need is for a decision support tool that

processes the data outputted by the models to a form which is easily interpreted by the decision

makers. Adding a spatial, geographical information systems module to it would further enhance

the authenticity and the confidence of the model, and also improve its ability to accurately model

the spatial variance of the landscape.

Decision Support Systems

Hydrologic models have served as a valuable tool for water resources management for

many years (Greene and Cruise, 1995). The pressure to develop better and more accurate models

requires the ability to better describe the landscape spatially. This can be achieved by the use of

geographical information system (GIS) within the model. A loose coupling of the simulation

engine and the GIS alone is not sufficient to assist the decision-makers/ stake holders to

efficiently make critical decisions. The necessary linkage is provided by the decision support

system, which processes the data outputted by the simulation engine and routes it to the GIS

module to represent it in a form that would be an effective assistant to a stake-holder. Reitsma

(1996) defines a decision support system (DSS) for water resources application as a computer-based










system, which integrates state information, dynamic or process information, and plan evaluation tools

into a single software implementation. In this definition, state information refers to data that

represent the system's state at any point of time, process information represents the first principles

governing resource behavior, and evaluation tools refers to software used to transform raw data into

information used for decision making (Satti, 2002)

A decision support system extends the scope of the simulation engine of any model to not only

include fixed scenarios that are pre-determined and preset, but allows a more complete view of the

various possible outcomes and options available to the decision-maker. In other words, it not only

looks into what would happen, but also what option could be available to change it. The wide range

of applications of DSS techniques for the study of water resources problems includes surface runoff,

river basin management, urban storm water management, groundwater contamination, have been

discussed in literature (Dunn et al., 1996; Jamieson and Fedra, 1996; Ito et al., 2001; Sample et al.,

2001).

The current modeling effort is undertaken using one such decision support system that supports

a GIS module and the above section is presented in order to help better understand the use and

advantages of a GIS integrated model.

Introduction to Questions and Decisions QnD

The Questions and Decisions (QnD) is a generic environmental modeling system that has

been developed using Java based object oriented programming according to Kiker et al. (2006).

The ideology behind this simple modeling structure and approach is to present the model as a

game (Figure 3-1) which will involve both managers and scientists. The game has appeal to both

the communities due to its ability to output data and interpret it according to the need of the user.

The managers or decision maker can use the simple user interface to assist them in making

management decisions without having to process numerical model output. The model uses









several useful and easy to understand methods to appeal to manager using warning light, tabbed

pane graphs indicating the trend of several important decision parameters and also a management

toolbar that contains management option that can be applied for the next time-step. The user

interface is developed using Java swing which provides a major advantage of platform

independence.

The model also integrates a geographical information system (GIS) module into the user

interface and has the ability to load several layers of shapefi1e to provide a better understanding

of the ecology and the landscape of the region. The GIS module is coupled to the Java model

using 'GeoTools-Lite', an open source GIS application programmable interface (API) for Java.

The GIS module allows the user of the model to do several operations on the spatial units such as

select, pan and zoom through a toolbar at the bottom of the screen. This allows the user to select

one or more of the pastures and apply some specific management action to those selected

pastures/spatial units, which accounts for an interactive and dynamic modeling experience. For

the scientists and the number crunchers, the model has also has a more conventional form of

output in the form of comma separates value Hiles (.csv) that can be loaded into Microsoft Excel.

An object oriented approach appeals to the world of modeling as the design of object,

classes and methods is easier to reflect from their real world equivalents (Kiker et al., 2006). The

benefits of an obj ect oriented approach have already been discussed in earlier sections. The

elemental obj ects in programming QnD are Components, Processes and Data (Kiker et al.,

2006). In QnD, all of the components are given a 'C' prefix, the processes, 'P' and the data a 'D'

prefix. So essentially the building blocks of QnD are the CComponents, PProcesses and the

DData. Components are obj ects of interest (Kiker et al., 2006); a component can be any of the

important physical players within the ecosystem, for example fish, grass, forest cover etc. The










Components describe the constituents and entities within a spatial area or spatial unit. Every

component has a set of data, sub-components and processes associated with it.

Processes are actions that involve Components, and Data are descriptive obj ects assigned

to components according to Kiker et al. (2006) and Kiker and Linkov (2006), in other words,

processes are the tasks that any component performs to interact in some way with both the

environment in general and other components. Processes use data to perform operations on the

components to show the interaction with other components. The data signifies the properties of

the components which are modified by the processes. Figure 2-2 describes a simplistic UML

look at the relationship between components, processes and data. The relationship that is shown

in the UML can be better described as, processes and data are elements of a component, and a

component can have one or more of each of these. A component can also contain sub-

components which can have their own processes and data associated with them. The processes

too can have sub-processes contained within it. When the QnD: BIR model is discussed in the

next chapter, a place of the sub-processes emerge as the maj or placeholder for all of the

calculations. In the UML use case diagrams, Figure 2-3 to Figure 2-6, the role of different

actors/players, the coders, model developers and the players, within the QnD system is made

clear.

Figure 2-3 describes the role of the coder or the code developer. Coders mainly interact

with the Java code in the model. The design, development and maintenance of the model code

itself, is the coder' s responsibility. The UML design and overview of the system is also designed

and maintained by the coder. A coder can interact with the players, the developers and the

outside world overall through the QnD website, with ideas and comments about the model









design. The model deployment is the Einal step of the development. QnD is deployed using Java

Network Language Protocol (JNLP) as described in the following sections.

The developer's role is depicted in Eigure 2-4. The model developer doesn't interact as

much with the code itself but primarily with the XML input files. The XML input file provide a

powerful and generic way to setup the QnD model, enabling the model developer to apply QnD

to different sites without having to change the java code at all. Through the XML the developer

can describe the modeled site, the components, processes and data discussed earlier. Moreover,

the user interface is also described in the XML by the model developer, all of the warning lights,

charts, graphs, GIS images and management options etc are designed and described in the XML

by the model developer. The developer interacts with both the players of the system, who are the

stakeholders, and the coder to best formulate the model development process. Once the model

has been developed, the calibration and validation of the model is the model developer' s

responsibility.

The players or the users of the model constitute the third class of people who interact

with the model. Figure 2-5 describes the interaction of the players to be minimal with the j ava

code or the XML Hiles. They mainly interact with the user interface of the model and the actual

output Hiles of the model. Their role as the user is to use the model to better understand the

implications of the decisions they are required to make. Players select the scenarios, the

management options and essentially run the model.

It is the interaction among these three actors that constitutes the maj or design and

development of the model, by generating a collaborative dialogue amongst the users and the

model developers, acquiring technical data and discussing both informal "rules of thumb" and

technical implications of management decisions (Kiker et al., 2006). QnD allows both hard data,









such as field-measured experiments, and soft data, such as experiential learning or general

impressions to be valid model inputs (Kiker et al., 2006). The result of the dialogue is conveyed

to the coder in cases where there is a requirement for code changes to the model.

Kiker et al., (2006) describes model development methodology as iterative and interactive,

involving alternative discussions with the stakeholders and model development, in order to best

understand the requirement of the managers and tailor the model to it. The object oriented nature

of the model coupled with the expert knowledge gained from the discussion with the

stakeholders provides the backbone of the QnD model.

Once an initial (prototype) version QnD has been developed, it can be used as a game to

stimulate further discussions between managers, scientists and stakeholders to try out different

management alternatives and investigate possible repercussions of those decisions (Kiker et al,

2006; Kiker and Linkov, 2006).

Once the development of the model is completed and the model has been put through

validation, it can be deployed online as a web-based model-game using Java Network Language

Protocol. Some of the earlier versions of QnD have been deployed as a game making it a good

resource for teaching and learning about the environment.

Overall the object oriented QnD model proves to be a powerful and easy to use decision

support tool, which couples an interactive design environment with a quick and efficient model

development and deployment cycle. It establishes an essential link between the research oriented,

complex hydrologic model and a simplistic user interface driver decision support tool used and

preferred by the managers, and hence is ideal for use in the current research study.











Wetla nds
Summer Experimecntal Pa stures
Winter Excperimental Past-uresr
Housing & Barns
Roads
Pastures


.. .L ~- .. IL ....... .

0.9 0 0.9 1.8 Miles



Figure 2-1.The figure shows the Buck Island ranch with its summer and winter experimental
pastures. (Kiker et al, 2006)


~2-~c~z~ 'I
i

~
%
TICil
uC~r





Figure 2-2. A simplistic UML look at the different components of QnD model (Kiker et al.
2006).


Interactions with OnD Web Site


Figure 2-3. Role of the coder or the code developer















Interactions with OnD Web 8ite


Figure 2-4. The developer's role


Interact ons w th QnD Web Site


Interactions with OnD GarneDriver Object


Figure 2-5. The interaction of the players is minimal with the j ava code or the XML files
























CSpatialUnit objects
have a specific reference
to a polygon in the base
layer shapefile.


SCHabitat objects are
associated with a
specific CSpatialUnit to
provide a non-spatially
specific container for
CLocal~ornponent
objects


+ homeHabitat


Wile CLocal~omponent
ejects are specifically tied to
CHabitat objects, they also
can access their location
satially.


Figure 2-6. The class diagram of a typical QnD system









CHAPTER 3
METHODOLOGY

The objective of this chapter is to provide a detailed account of the current version of the

QnD model applied at the Buck Island Ranch (QnD:BIR). The actual design and development of

QnD:BIR is discussed, including the major processes involved in the model. Some of the actual

objects (components, processes and data) included in the model development are further

explored in this chapter, followed by an account of some of the model calibration, validation,

error quantization and data representation techniques commonly used.

Design of an Enterprise-Level Model for Buck Island Ranch QnD:BIR

The modeling effort for the QnD: Buck Island Ranch was designed to be simplistic and based on

literature-derived concepts, empirical data and expert knowledge. Most of the relationships in

this version of QnD are either empirical, calculated from the data recorded at the research site or

is based on the basic laws of hydrology. Figure 3-1 shows a screenshot of the model with the

cow icons and the GIS coverage of BIR.

The basic spatial setup of the model divides the whole area of BIR into 68 spatial units

(CSpatialUnit), each representing the 68 pastures on the ranch. Each of these spatial units contain

a CHabitat, which the main holder of the other local components. In QnD:BIR, the habitat is

considered to be 'default' and is contained in all spatial units. Each of the CHabitats contains

several local components including CBahiagrass, CNativegrass, CUplandSoil, CWetlandSoil,

and a CHerd wherever a herd is present. This is further clarified with the help of the UML class

diagram in Figure 3-2. The grasses and soil have a percent area DData, which signifies whether

the selected spatial unit is an upland or a wetland, improved or native pasture. Moreover, another

DData (DImprovedPasture) property is defined to signify whether the current spatial unit is an

improved or a native pasture. There are several other DData objects and processes which are









global to the model which signify values used for calculation (DZero etc) or are input/output

variables of the model.

The main focus of this model is to simulate five major aspects of the Buck Island ranch

ecosystem and ranch management operations:

QnD:BIR Hydrology

The hydrology of this region has been previously modeled by in-depth models like

ACRU2000 (Campbell et al., 2001; Clark et al., 2001; Kiker and Clark, 2001; Martinez, 2006;

Yang, 2006; Pandey, 2007) and WAM (SWET, 2002), which have enabled highly detailed sub-

daily modeling. On the other hand, the QnD Buck Island Ranch model has followed a very

simplistic, deterministic method to model the hydrology of this region by using simple

relationships between the Ground Water level data and the amount of available storage in the

soil. This enables the development of an effective hydrological model in a short period of time

which models the actual data on a monthly scale within acceptable levels of accuracy.

The model structure is defined by a number of equations and relationships. One of the

major inputs for modeling the hydrology of any region is the rainfall data. For the QnD Buck

Island Ranch model, the Average monthly rainfall data is provided as input to the model. This

model uses Ground Water Level as an input factor for calculating the runoff. Ground water table

data is read in as input to the model. The height of the possible available column of storage is

calculated as the difference between the mean height of Buck Island Ranch from the sea level

and the height of the ground water table. i.e.

Possible Available Column of Storage (in meters) = Mean Height of Buck Island Ranch (in

meters above Sea Level) Mean Height of Ground Water Table (in meters above Sea

Level) (3-1)









The available water storage is dependent on the soil type. This value, deduced from the

Possible Available Column of Storage, is based on the porosity and the field capacity of the soil.

This factor accounts for the plant available storage and the gravitational storage i.e.

Available Water Storage (in mm) = Possible Available Column of Storage (in mm) *
(Porosity Field Capacity) (3 -2)

Moreover, an evapotranspiration (ET) factor is also added to this plant available storage

which is calculate as a function of average temperature of the area during the month. There are

other factors that affect ET, but since temperature is the major factor, it is taken into account in

the model.

The value of the possible runoff is calculated as the difference between the rainfall and the

actual available storage.

Runoff (mm) = Total monthly rainfall(in mm) (Available Water Storage (mm) ET

factor) (3-3)

Finally, the total runoff volume is calculated for each pasture, taking into effect its area and

the amount of runoff as a result of rainfall, i.e., the total runoff volume is calculated for every

pasture using the equation:

Runoff Volume (in L) = Area of the Pasture (square m) Runoff (mm) (3 -4)

To visualize the model using the obj ect oriented approach used by QnD, each of the values

being calculated represents a DData, and each relationship/equation is described by one or more

PProcesses. The hydrology component is governed by the soil component (CUplandSoil or

CWetlandSoil), which contains the PProcesses. The inputs are again DData values which are

either local to the spatial unit (DPercentAvailableStorage) or globally exist (DMonthlyRainfall).

In QnD terminology, the whole of hydrology is large a part of a single PProcess,

PCalculateRunoff, that governs it. This PProcess has includes several sub-processes which










perform the calculations based on the relationships described earlier. The detailed list of

processes supported by QnD and their explanations are part of appendix B. These sub-processes

systematically perform the calculations in the order they were setup, and the calculated results of

total runoff volume is stored in the DData, DRunoffVolume. It is important to note that QnD

only performs the processes at every time-step, if the component is present in specific spatial unit

that is being updated.

QnD:BIR Nutrients

The nutrient movement of this region has been previously modeled by in-depth models

like ACRU2000 (Campbell et al., 2001; Clark et al., 2001; Kiker and Clark, 2001) and WAM

(SWET, 2002) which have enabled highly-detailed modeling. On the other hand, the QnD Buck

Island Ranch model has followed a very simplistic, deterministic method to model the nutrients

of this region by using simple relationships. The simplistic approach enables the modeling of the

complete ranch, on a broad scale, which helps to better understand ranch dynamics and the effect

of the hydrology and nutrient cycle on ranch management and profitability. In this model,

nutrients in the region are divided into extractible phosphorus and stable phosphorus as

perceived from the standpoint of the model. The extractible phosphorus is responsible for all the

nutrient movement from the soil to the cows and in the hydrology. The stable phosphorus, as the

name suggests, is considered to be relatively stable and is always present in the soil. However,

every time step, there is a nutrient movement from the extractible phosphorus to the stable

phosphorus and vice versa at different rates. This movement is governed by transfer coefficients

that are used to transfer extractible to stable phosphorus and vice versa. This process of transfer

can be shown by a series of equations as following:

Extractible to Stable Phosphorus Transfer Amount (in kg) = Extractible to Stable
Transfer Coefficient Extractible Phosphorus (in kg). (3-5)









Stable to Extractible Phosphorus Transfer Amount (in kg) = Stable to Extractible
Transfer Coefficient Stable Phosphorus (in kg). (3 -6)

The runoff event triggers the movement of a fraction of the Extractible Phosphorus with the

runoff. This amount of phosphorus or in other words, the runoff phosphorus load (measured in

kilograms) is calculated in the model using an empirical linear relationship that is derived

directly from the measured BIR data.

This linear relationship is based upon the measured average runoff volume to the

measured average phosphorus load for the set of native and improved pastures. Figure 3-3 shows

the trend line characteristic of this linear relationship.

The obj ect oriented interpretation of these conditions and components, which is used by

QnD, describes the pools of phosphorus as a property of the soil component (CUplandSoil or

CWetlandSoil) of the spatial unit. Each of these components contains DData values which

signify the presence and amount of extractable (DExtractableP) and stable (DStableP).

PProcesses, PExtractableTo StableTransfer and P StableToExtractableTransfer interpret the

phosphorus transfer between the two pools, using the PTransfer process type, to the obj ect

oriented QnD model, incorporating the equations discussed earlier in the section.

The actual phosphorus load coming off of the spatial unit is governed by the

PExportInRunofflmproved and PExportInRunoff'Native processes which correspond to the

Runoff vs. PLoad relationships described earlier for the improved and native pastures

respectively. Both of these processes affect the total phosphorus load DData (DPLoad), updating

its value at every timestep.

QnD:BIR Forage Growth

The forage model component of QnD:BIR is again designed with the similar simplistic

approach as the other components of QnD. Forage is a very important component of the BIR










ecosystem and the sustenance of the ranch depends on the cows having enough forage to feed on.

Lack of forage growth also causes the cow/ calf condition to worsen and thereby result in

unhealthy cows and additional expenditure in buying supplemental feed for the cow. QnD:BIR

mainly considers there are two factors that affect the growth of forage, the relative rainfall and

the seasonal effect.

For the effect of relative rainfall, the average monthly rainfall for the whole period is

calculated for every month.

Average monthly rainfall (month = Jan) (mm) = All years) Monthly rainfall (mm) / N

(3-7)

The average monthly rainfall is then used by the model to calculate the relative rainfall

for the current month.

The relative monthly rainfall affects the forage growth as an empirical linear relationship.

The values of this relationship are calibrated to best suit the conditions present at BIR, with

within the confines of the acceptable results.

Forage growth rate = f (relative monthly rainfall) (using the linear relationship). (3-8)

Total grass biomass (Total Forage) = Forage growth rate Total grass biomass (3-9)

The seasonal effect on forage is more complex within the model than rainfall effect.

Moreover, it also accounts for the wilting of the grass during dry season. This is governed by an

empirical curve, which is calibrated to suit the site being modeled i.e. the Buck Island ranch.

The seasonal effect is governed by similar equations as mentioned above:

Forage growth rate = f (current month) (using the curve). (3-10)

Total grass biomass (Total Forage) (in 1000 kg) = Forage growth rate Total grass
biomass (in 1000 kg) (3-11)









Similar relations are used to model both native and improved pastures, where the herds

are present and the herds consume the grass at a constant rate per day per cow. The specific

value and its background are mentioned the next few subsections. With the coupling of the three

factors, QnD:BIR overall presents a good and simplistic model design based off of expert

opinion, literature and the ranchers view of their ecosystem.

Forage growth relations are incorporated into the model using two simple PRelationship

sub-processes within the PCalculateForageGrowth process. These two PRelationships signify the

rainfall effect and the seasonal effect on forage, defined earlier in this section. The forage

relationships are specified within the CBahiaGrass or the CNativeGrass components of the

model, which are to the most part similar.

QnD:BIR Beef Cattle Management

The cows are the most integral part of the Buck Island Ranch ecosystem. As mentioned

earlier, the ranch is primarily a beef cattle ranch and cows are a maj or asset and a very important

player in the ecological balance at the ranch. The model is primarily a management tool, with its

primary focus as the stakeholders and their interests, which, in this case, are the ranchers at the

Buck Island Ranch. Figure 3-4 shows a timeline for the ranch management operations, the model

is primarily based on these timelines to effectively simulate the beef cattle enterprise.

The modeling of the cows at the Buck Island Ranch has been developed based upon the

Eigure shown above. The model portrays three aspects of the life cycle of the calves, i) the

impregnation of the cows (breeding), ii) the birth of the calves, and iii) the time when the calves

are weaned and sold.

Breeding

The calves born on the ranch that can be further divided into three categories on the basis

of the part of the breeding season that the cows are impregnated in, namely: the calves born of









cows that are impregnated early in the season, which form the early cohort; the calves born of

cows that are impregnated in the middle of the season, which form the middle cohort; and the

calves born of cows that are impregnated late in the season, which form the late cohort. This

breeding season ranges from January to the end of April. The rate of impregnation during this

season depends upon the climatic conditions, i.e., temperature, rainfall, etc. and the condition of

the cow. From the analysis of the ranch Standardized Performance Analysis (SPA) data, the

average impregnation rate of the cows all through the breeding season in the ranch is between

75% and 80%. The model assumes an almost equal rate of impregnation throughout the breeding

season, only varying due to the climatic conditions. During the entirety of the breeding season,

all the herds of cows on the ranch are exposed to bulls for a time period ranging from 90-120

days, and the ratio of the number of bulls to cows is 1:25 (Source: Patrick Bolen).

Calving

The average gestation period of the cows is 9 months (Source: Patrick Bolen), hence the

first calves are born around November from the cows that were impregnated early in the

breeding season. The calving continues all the way through to the end of the following February.

The cows impregnated early in the breeding season give birth at the onset of the calving periods,

around the month of November, and the calves so born form the early cohort. Similarly, the

middle- and the Late Cohorts are born all the way through February. The population of the

cohorts can be calculated by the following equations:

Early Cohort Population (cow units) = Early Pregnancy Rate Cow Population (3-12)

Middle Cohort Population (cow units) = Middle Pregnancy Rate Cow Population (3-

13)

Late Cohort Population (cow units) = Late Pregnancy Rate Cow Population (3-14)

The pregnancy rates for early, middle and late cohorts are dependent on the condition of the cow.










Weaning and selling

From the time of birth, the calves gain 1.4 pounds a day, on an average (Kunkle et al.,

2001). The calves are weaned around the month of May. At this stage, it is ensured that the

calves reach their endured target-selling weight of about 492 pounds on an average.(SPA 2005)

The calves that fall short of the required mark could be fattened by using one or more techniques

such as supplemental feed or/ hormones. This is reflected in the model as a management decision

that the user can make while running the model. Once this is done, the calves are sold in cohorts.

The selling of the calves is also a management decision which is part of the user interface in the

model. This gives the user/stakeholder the option to sell any/all cohorts at the time he thinks is

right for the ranch.

Cow intake and waste

Int a ketttt~~~~~~~tttttt

The average intake per day per cow is about 25.9 pounds of dry matter (Kunkle, 2002).

Of all the food ingested by the cattle, the utilization rate is about 55%, i.e. the nutrition level of

the food ingested by the cattle (Kunkle 2002). This utilization rate is higher for Bahiagrass as

compared to the other varieties of forage (Kunkle 2002). This percentage of utilization of the

forage by the cows is calculated from the amount of grass/forage available on the pasture. After

the forage is ingested by the cattle, the total grass biomass is also calculated and updated.

Waste

The cattle waste on the ranch is modeled primarily for the phosphorus and nitrogen content

present in it. The amount of phosphorus present in the cattle waste is about 0.044 kg/cow and

about 0.04 kg/calf and the amount of nitrogen is about 0.019 kg/cow and 0.017 kg/ calf (ASABE

Standards 2007). Of the total amount of phosphorus and nitrogen present in the cattle waste, a









certain percentage is extractible. This value of extractible phosphorus and nitrogen is accordingly

updated from the phosphorus and nitrogen loads that are dropped on the soil.

The beef cattle management is more complex to visualize in obj ect oriented terms due to the

variety of factors, parameters and processes involved. To start with, the cattle herd can be looked

at as a component (obj ect) and all of the QnD:BIR beef cattle management is contained within

this local component. Several PProcesses are included within this component each signifying the

various maj or aspect of ranch management discussed earlier in this section (breeding, calving

and selling).

It is important to note that, each of these beef cattle management sections are implemented

for each of the three cohorts that the calves are divided into. This further complicated the design

of the QnD:BIR beef cattle management. An attempt has been made to retain the simplicity of

the design of the XMLs to assist any of the stakeholders, who might be interested in the design

of the model, to understand it easily.

QnD:BIR Ranch Incomes and Expenditures

Consideration of the financial details in the model starts with the previous year' s ending

balance as the current year' s beginning balance. This amount is considered to be a fixed value

for the purpose of this model. During the course of the management cycle, the ranch may incur a

variety of both revenue as well as expenses. The maj or source of revenue for the ranch is the sale

of the calves, which are sold at an average price of about $1.10/pound of calf (Source:

Standardized Performance Analysis, MAERC). Other sources of revenue may include SOD and

sale of pregnant cows or bulls. In our model, we look mainly at the sale of cows and the lifting of

SOD, both of which are management options available to the user. At the end of each monthly

time step, we calculate the total revenue gained by the ranch.









The ranch also incurs a number of management and maintenance expenses that include

feeding and grooming of the cattle. One of the maj or expenses of the ranch is providing

supplementary feed during the preconditioning period of the calves. To improve the total yearly

yield, the ranch also buys impregnated cows. Another source of expenditure at the ranch is the

hormones inj ected into the cows, which is an almost regular practise (Source: Patrick Bolen).

Apart from the abovementioned expenses, there are a number of other conditional expenses that

may be incurred by the ranch, one of which includes pumping water. During the years that the

ground water level is low, additional water is pumped into the canal and may cause electricity

overhead.

To analyze the various combinations of conditions and their corresponding financial

repercussions, the model is designed with the idea that the user of the tool, i.e., the rancher, gains

an overall management and financial perspective and can vary the conditions to study the effects.

This takes into consideration that within the model, all these options are mainly management

options which the user can set while he is running the model, the idea being that the financial

decisions should always remain in control of the rancher who is using the tool.

The obj ect oriented model interpretation of incomes and expenditures has far-reaching

applications, outside the field of hydrological and ranch operation modeling. QnD:BIR

establishes a link, on a simplistic level, between the world of economics and the object oriented

programming model. The model design handles the ranch's month to month economics by

tracking on a simplistic level, the incomes and expenditure of the ranch through the cow/calf

operation. The two DData values of DTotallncome and DTotalExpenses govern these two

values. And the different between them is considered to be the operating balance of the ranch

(DTotalOperatingAmount) The incomes and expenses are usually a result of the management










options which the PTotalExpenses and PTotallncome aggregate to update the values of

DTotallncome and DTotalExpenses. Incomes and expenditures can be a result of more than just

cow/calf operations, for example, pumping excess water etc. can also result in the ranch

incurring expenses. Hence this module of QnD:BIR is controlled at the global level, and does not

belong to any local component. However, as each of the local components contribute to the total

incomes and expenditures, the values are updated and the total operating amount is calculated at

the end of every timestep.

Overall, this module can be looked at as a set of processes or operations in the object

oriented sense, which is affect by individual obj ects or components through their respective local

processes. A similar simplistic design can be a starting point for the development of other

applications that need to translate financial or economic processes into object oriented

programming.

Addition of New Features into the QnD Model

Initial design and versions of QnD:BIR were reviewed by ranch management and

scientists who requested that a significant new feature would need to be added to the user

interface; moveable icons that represent cattle herds. The requested features were not present in

any of the previous models of QnD. Thus, an additional feature in QnD Buck Island Ranch is

the capability to add icons representing cows on the spatial units representing pastures. These

icons can be placed on the spatial units if the data suggests the presence of a cow herd on the

corresponding pasture. The cow icons can be moved as one moves the cows from one pasture to

another as part of a management decision. This capability to move cow herds is a very important

part of the management cycle of the operation on the ranch.

The technological challenge associated with the development of such a management option

and dynamic icon movement is what makes in a special feature of QnD:BIR. The dynamic










placement of the cow icons on the pastures to indicate the presence of a herd involves placing an

additional GIS layer, a marker point layer, on top of the existing GIS maps that are loaded at

startup. The GeoTools-Lite is generally a useful API for GIS, but the documentation of it is still

sparse which further complicates the task. This point layer is generates by checking each of the

spatial units that are being loaded from the XML files for the presence of a herd or in QnD terms

the presence of a CHerd component. If a CHerd is found in a spatial unit (CSpatialUnit), a new

point is added at the (x, y) location of the centroid on the new marker layer. The procedure is

replicated for all of the spatial units and the result is the placement of the cow icons on the user

interface GIS map of QnD:BIR.

Once the icons are placed onto the GIS map, moving the cows as a management option is

the next technical difficulty. QnD supports the movement of a component from one spatial unit

to another, wherein, all of the processes and data linkages are moved along with the component

to the destination spatial unit and the links are reestablished. In order to facilitate the

management option of moving cows, Java Swing objects were used to create the user interface

extensions and QnD component movement support was used to move the cows in the model

setup. When the cows are moved, instead of having to redraw the whole marker layer, QnD:BIR

simply loads the marker layer and deleted that one point corresponding to the CHerd being

moved and adds a new point at the destination spatial unit. The map is then refreshed in order to

reflect the changes made to the layers.

This management option provides future model developers, the option to move their

components during the course of a model run as a part of a management decision, which in turn

further expands the flexibility of the model itself.









The simplistic model design of QnD:BIR enables it to cover varied modules, on a ranch

scale. Moreover, the design also accommodates any future improvements to the model structure

and design relationships, hence making it highly extendable.

Model Calibration, Validation and Data Representation

The efficient development and working of a model requires checking the accuracy of the

results and increasing the robustness of the model. Model Testing is used to improve the

performance of the model by detecting the design shortcomings in the model algorithms and

using procedures like calibration and validation to improve the hardiness of the model as well as

its accuracy and performance. The following sections expand on the procedures used for model

calibration and validation:

Model Calibration

Model calibration is the methodology used to tune or update the model settings to suit the

site that it is being used for. Model calibration process involves identifying parameters that

within the model, that could allow the possibility of an error factor, namely parameters that have

been used from a general national average or derived from other studies at different sites where

the conditions might not be exactly the same as the chosen research site. Once these parameters

have been identified, their values are changed to suit the conditions being modeled. Moreover,

for the calibration period chosen, these parameters values are corrected within the allowable

limits to best capture/follow the observed data trends.

Model calibration is done in order to improve the model's accuracy, by adjusting the

parameters to best suit the historical data observed. However, the more complex the model, the

more parameters that can be changed, making the calibration process to be more and more

complex.









Model Validation

Model calibration prepares the model to best suit the conditions at the current research site.

A model is then validated by running it for the validation period without any further change in

parameters or setup after it has been calibrated. The results of these runs are then compared to

the observed/measures values of the parameters that are being modeled. This is done in order to

gauge the effectiveness of the model and its setup. During the validation period, several

statistical and other methods are used to calculate and quantify the error that is present in the

model. The following sections describe some of the commonly used statistical error

quantification methods.

Model Evaluation

Graphical representation of the model results does quantify the model and the general trend

between the observed and the model results. However, visual evaluation of the results alone is

not sufficient to gauge the effectiveness and accuracy of the model. Statistical analysis methods

are used to quantify the visual evaluation of the result by using several methodologies to quantify

the amount of error that is exhibited by the model results. The following section attempts to give

an overview of a few commonly used model evaluation techniques.

Statistical Representation

Quantitative methodologies of analysis are required to evaluate the results of a calibrated

model. This section looks at some of the commonly used methodologies that are used to validate

the results of this modeling effort, namely, Root Mean Square Error (RMSE), Pearson product-

moment correlation coefficient (R2), and Nash-Sutcliffe (NS) Coefficient (Nash and Sutcliffe,


1970). In all the following equations, P is the observed value, Oi is the model-simulated value,

and Nis the number of observations.









The Root Mean Square Error: RMSE is essentially the overall sum of squares errors

normalized to the number of observations (Hession et al., 1994)(Yang 2006). The RMSE is

calculated in the same units as the analyzing quantity. The following equation is used to calculate

the RMSE:


RM~SE = [4-O)
N 0 RMS < 00(3-15)

This value can be interpreted in term of the units of the modeled parameter. Due to the

presence of a quadratic term in the equation, a large error value has a greater effect and on the

other end, smaller values indicate better model performance (Evans et al., 2003)

Pearson product-moment correlation coefficient (R2), iS the measure of linearity between

two variables. R2 is probably the most popular measure of fit in statistical modeling. The values

of R2 can range between 0 and 1, with 1 being the 'perfect' match of measured and predicted.

The equation used to calculate the value of the coefficient (R2) iS given by:



R 2 I=
1~~~~~ (O O)2I(4-P
0 < R2(3-16)

The Nash-Sutcliffe (NS) coefficient (Nash and Sutcliffe, 1970) is one of the more

effective ways to indicate a goodness of fit. This method is also recommended by the American

Society of Civil Engineers (ASCE, 1993) as an effective instrument of model validation. An NS

value of 1 indicates a perfect fit and alternately, as the value approaches zero, the lesser the

accuracy of prediction of the model. The NS can be computed by using the following:











NS = 1 -=

-0
The NS is most effective when the coefficient of variation for the observed data set is large

(ASCE, 1993).

The modified form of Cgf was developed by Krause et al. (2005) to reduce the sensitivity

of Cafto large values:




Cs =1 1- with j=1 (3-18)




The overall model design is simplistic but the knowledge based iterative approach

strengthens the modeling effort. The strength of the model is measures through the process of

testing and validating the model, which is described in the following chapter.









CHAPTER 4
MODEL TESTING, RESULTS AND DISCUSSION

This chapter focuses on the results and the testing of the model. Model results from the

calibration to the validation stage are represented using a variety of graphical methods and an

analysis of the most and least favorable results is performed to identify the strengths and

weaknesses of the model.

The QnD:BIR decision support tool, as discussed in Chapter 3, is developed using Java

and obj ect orientation for the south Florida beef cattle agro-ecosystem. This model uses simple

relationships based on measured data, the laws of hydrology and results derived from

consultation with the ranch managers. The model was developed to be applied on a whole farm

for a variety of processes and events being modeled, ranging from hydrology and nutrient

movement, which is the maj or validation modules of the model, to cow/ranch management and

the income-expenditure cycle. The enterprise management and the income-expenditure cycle run

different scenarios to assist the decision makers to interpret the output of the hydrology and

nutrient engine and its effect on enterprise management and profitability. The time-step for the

version 1.0 of QnD Buck Island Ranch is set to be monthly to better suit it to the decision

timescale at which the ranch is being managed.

Model Inputs

QnD is capable of reading time series inputs from a file in comma-separated format. These

files have to be declared in the XML input files of the model which are read in and stored in hash

tables. For QnD: Buck Island Ranch, the inputs provided to the model include average monthly

rainfall and ground water table depth, which are being read from time series files. For the GIS

module, a GIS shapefile describing the modeled area is also a part of the input, which contains









the area and perimeter of each pasture, which is also read in as input. All of the input that is read

is stored into DDriverData obj ects.

As a part of the discussion about the inputs of the model, an overview of the input data

and its implications is warranted. Figure 4-1 is a graph of the total monthly rainfall and the

ground water levels of BIR. From the graph we notice that the time period that we are applying

the model is a combination of wet and dry periods. Early 2000 and 2001 were relatively dry

periods, with low ground water table. June September 2001 is a wet period with high rainfall

and as a result we notice a rise in the water table depth. This is again followed by a relatively dry

period between October 2001 and June 2002. Hence the water table too drops slowly. The point

to be noted here is that the water table is supported to some extent by the presence of the Harney-

Pond canal which is maintained at an almost constant height, and which contributes to the water

table. But this effect of the canal is already considered by the model by taking the water table

depths directly as input.

The following year, late 2002 2003 is a wet year with rainfall events sparred throughout

the year. This results in the water table staying relatively high throughout this period, which

presents two different scenarios. Firstly, late 2002 period, wherein, the rainfall is relatively low,

but the level of the water table does not drop, in fact there is a slight increase observed during

this period. Alternatively, the period of mid to late 2003 presents a different scenario of high

water table and large amount of rainfall. Each of these scenarios could affect the accuracy of the

model, as QnD:BIR directly relates rainfall and ground water depth with runoff. This implies that

a high water table would result in even a slight amount of rainfall causing heavy runoff, which

might lead the model to over-predict in this period.









This overview is designed to understand the input conditions that the model is being tested

upon and their implications and effects on the model.

Model Outputs

QnD outputs data in two modes, one of them is through the graphical user interface (GUI)

which would be graphs and indicator lights mainly intended for the decision makers to assist

them in managing the enterprise. These outputs can be customized to suit the requirement of the

enterprise being managed. For QnD: BIR, GUI outputs are the cow and calf population and the

operating-amount remaining. The other, more conventional, mode of output is the comma

separated file (.csv) with the numerical values of the output parameters. For QnD: BIR, runoff

volume, phosphorus load and grass biomass constitute the maj or outputs required for validation

of the model.

Model Calibration

The model was calibrated for the dry period of September 2000- January 2001 and a wet

period of July 2001 and August 2001, for the experimental pastures summer pastures 1-8 and

winter pastures 1-8. For the hydrological model, the parameter used for calibrating the model is

the percent available storage, which combines the effect of plant available storage and the

gravitational storage. Within the forage model, the rainfall effect and seasonal effect parameters

are the parameters that the model was calibration on.

Hydrology

During calibration, the hydrology of the model was calibrated based on the percent

available storage parameter. The calibrated value of this parameter was determined to be 0.23 8.

This number is large enough to account for the plant available storage and in also the effect of

evapotranspiration on the rainfall, which is not taken into consideration by the model separately.

The model overall was predicting the runoff amounts on the higher than the measured values.










Figure 4-2 shows a typical graph for one of the summer and one of the winter pastures during

calibration. This covers the trend shown in more detail in appendix A, which contains all the

results from all of the experimental pastures. Since part of the calibration period is one of the

driest times for the region, the model calibration should have helped it predict any future dry

patches with accuracy.

Nutrients

Calibration period model runs for nutrients followed a similar trend at the hydrology, with

the measured values being less than the model predicted. Figure 4-3 shows a typical graph for

one of the summer and one of the winter pastures during calibration. This covers the trend shown

in more detail in appendix A, which contains all the results from all of the experimental pastures.

Forage

The forage model calibration was done for the period of 2000-2001. The Buck Island ranch

forage data set does provide forage yield data for January, March, and December of 2000.

Moreover, the forage yield for the each pasture varies over a range for any given month.

QnD:BIR was calibrated within the range of values measured at BIR. Hence, the model

calibrated for the available ranges, in the absence of continuous data, for the calibration period.

Model Validation/Testing

Model testing for remaining time series data is described in this section. In order to give a

general idea of the trends of the model over the sixteen pastures, the highest and lowest

performance levels of the model for both the improved (summer) and native (winter) pastures.

Additional results are provided in appendix A.

Hydrology

Considering the overall predictions of the model, it can be seen that the overall model

tends to marginally overpredict the values of runoff, given the scale of the values, while missing









some of the smaller runoff events. Analysis of Summer 5 (Figure 4-2) Pasture indicated that the

model predicted the values quite closely with a Nash-Sutcliffe coefficient (NS-Ceff) of 0.6169, a

Root Mean Square Error (RMSE) of approximately 5.025 million liters, and a Normalized Mean

Square Error (nMSE) of 0.3 83 1. During the validation period between January 2001 and

December 2003 excluding the calibration months, the higher peak events occurred during the

months of May and September of 2001, 2002 and 2003, and January of 2003 of which the model

predicted most months with an acceptable degree of accuracy except those of September 2000,

October 2001, June 2002, and January 2003 where it missed some of the more significant events.

However, some of the smaller runoff events occurring during the early months of the year each

year were missed by the model.

The cumulative Runoffs, however, are generally close to the measured values, indicating

that the overall amount of runoff from the Summer 5 pasture (Figure 4-3) is being predicted with

better levels of accuracy. This indicates that though the model misses a few small events, it is

able to predict the total volume of runoff over a period of time. In the graph of the cumulative

runoffs, the curve of the measured values follows the curve of the predicted values very closely,

which, along with the high Nash-Sutcliffe coefficient value, makes Summer 5 the best

performance of the model for summer pastures.

On analyzing the measured vs predicted scatter graphs (Figure 4-4), it can be seen that

Summer 5 has one of the best prediction trend among all the summer pastures for runoffs. The

overall model does tend to marginally under-predict the values; however, the error value is not

very high. The measured values and the predicted values are also highly proportional.

Considering the least favorable of the model's performances, analysis of the Summer 4

pasture (Figure 4-5) indicated that the model did not perform as well. The graph indicates a










greater degree of over-prediction than other pastures with a NS-Ceff value of 0.5 100, an nMSE

of 0.4900 and a RMSE of approximately 4. 113 million liters. During the validation period

between January 2001 and December 2003, the model predicted the higher peaks that occurred

during the months of May and September of 2001, 2002 and 2003, and January of 2003, with a

lesser degree of accuracy than the best performance of the model. Moreover, the model missed

predicting a few of the events that occurred during the months of September 2000, October 2001,

December, January and February 2003.

The predicted cumulative runoffs for this pasture (Figure 4-6) are higher than the model

measured cumulative runoff values. The values begin together at the start of the validation

period, but move apart from September 2001, when the model missed a runoff event. Following

this period, the distance between the two curves increases since the model overpredicts the peaks

in 2002, except for certain points where they come marginally closer.

The measured vs predicted scatter graphs (Figure 4-7) for Summer 4 indicates that a

percentage of the values are over-predicted. The NS-Ceff value is also calculated to be the

minimum among the coefficient values of all the other summer pastures, which indicates that the

model has not proven as effective in modeling Summer 4.

Though the overall values were less accurate than the model's best performance, the

model still managed to capture the general trend of the peaks and the lows of the predicted

values quite accurately.

Moving to the native pastures, among all the model predictions for the native or winter

pastures, the best performance of the model is shown in Winter 4 (W4) (Figure 4-8), with a NS-

Ceff of 0.809, which is considered to be quite accurate, an nMSE of 0. 1908, and an RMSE of

approximately 5.608 million liters. The runoff graph for W4 shows a close match between the









measured values and the model predicted values, giving the model a highly acceptable degree of

accuracy. The graph shows accurately predicted high peaks for the months of of May and

September of 2001, 2002 and 2003, and January of 2003. However, the model still misses a few

of the smaller events, e.g., in the months of September 2000, May 2001, January, May and

August 2003.

The cumulative graph for the runoffs in W4 (Figure 4-9) gives a very clear indication of

the accuracy of the model in the abovementioned pasture. The curve for the measured values

follows the curve for the model predicted values very closely as seen in the graph. The values are

very close almost throughout the validation period, except for an instance in November 2002 and

one in September 2003, where the measured values differ from the predicted values with a

slightly greater margin.

The measured vs predicted scatter graph for W4 (Figure 4-10) indicates clearly the

accuracy of the model with respect to pasture W4. Very few values on the graph are under-

predicted, and the high proportionality also indicates the high accuracy of the values. The best fit

trend line also matches quite closely with the 1-1 trend line, overall implying a good model

performance for W4.

Among the native pastures, the model shows the least accuracy of performance on the

Winter 3 (W3) pasture (Figure 4-11). The result graphs indicate the presence of some

discrepancy between the measured and the predicted values for the month of July and October

2001, June, August, December 2002, and June, September and October 2003. For certain other

months like September 2000, November 2001, February, March, April and November 2003,

some less significant runoff events are missed by the model.









The cumulative runoff for W3 (Figure 4-12) is correspondingly reflective of the lower

accuracy of the model on this pasture. The curves follow closely until May 2001. However, after

this period the model over-predicts the runoff on three crucial occasions in September of 2001,

2002 and 2003 which causes the cumulative curves to sparse out rapidly.

The measured vs predicted scatter graph for the W3 (Figure 4-13) pasture indicates an

overprediction of some of the values. It can be seen from the graph that the best fit trend line

does not perfectly match the 1-1 trend line as it does in W4, which is the best performance of the

model .

Phosphorus Load

The analysis of the model performance in the various pastures shows that the model tends to

slightly overpredict the values of the peaks and misses a few smaller events, similar to the

performance of the model on different pastures for runoffs. However, over NS-Ceff values

overall are higher for the phosphorus load simulations, indicating the overall better performance

of the model in predicting phosphorus loads

Among the model performances in the various summer pastures, the pasture with the

most accurate model predictions is Summer 8 (S8) (Figure 4-14), with a NS-Ceff of 0.5960, an

nMSE of 0.4040 and an RMSE of 5.35 kg. The graph depicting the S8 nutrient load indicates the

close match between the measured and the predicted values. For this pasture, the most significant

events take place during the months of May and September 2001, 2002 and 2003, and December

2002. Though the model mirrors the measured values with an acceptable degree of accuracy,

there remain a few values that the model misses, as shown for the months of September 2000,

May 2001, May 2002, and September 2003.

The cumulative load graph (Figure 4-15) also indicates a good level of accuracy in model

performance. During the initial stages of the validation period, the predicted values closely









mirror the measured values. However, around the period of September 2001 and 2002, there is

an increase in the discrepancy between the values. This discrepancy varies until the end of the

validation period.

The measured vs predicted scatter graph (Figure 4-16) for S8 shows a good prediction

performance. Though a few of the values are under-predicted in the graph, the best fit trend line

matches the 1-1 trend line tolerably well and is also indicative of the accuracy of the model in

this regard.

Considering the pastures for which the performance of the model was less than

completely satisfactory, the pasture where there appears to be maximum discrepancy between

predicted and measured value is Summer 3 (S3) (Figure 4-17) with a low value of NS-Ceff

(0. 1484), with 0.85 16 and 5.205 Kgs respective values of the nMSE and RMSE calculated. In

this case, the model matched the high peak values for the months of May and September 2001,

2002 and 2003, and December 2002, though without the accuracy displayed by the model in

other pastures. The model also missed the values in may 2001, 2002 and September 2003. The

overall model however, catches the general trends of peaks and lows similar to the hydrology

module.

The graph for the cumulative nutrient load (Figure 4-18) indicates the lower level of

accuracy in the performance of the model for this particular pasture. The match or relationship

between the parameters remains close until May-September 2002, after which, the model over-

predicts a few values causing a gap between the measured and predicted curves.

The measured vs predicted scatter graph for S3 (Figure 4-19) indicates some under-

prediction. A number of the data values are overpredicted, but the best fit trend line is below the

1-1 trend line and is not very accurate.









Overall, for the summer pastures, the model captures most of the significant events and the

general trends of peaks and lows. This is noticed across all of the eight summer pastures. The

detailed graphs of these are included in appendix A.

On comparing the model performance for the various native pastures with regards to

nutrient loads, the pasture with the most accurate model performance is Winter 4 (W4) (Figure 4-

20), with a NS-Ceff value of 0.6946, an nMSE value of 0.3054 and an RMSE value of

approximately 0.8899 Kgs. In the graphs associated with this pasture, the values predicted values

closely match the data measured at BIR. The model predicts values with very marginal

discrepancy for the high load months of July and October of 2001 and 2002. In spite of missing a

few of the smaller events, the model still maintains a highly acceptable level of accuracy and W4

is therefore among the best performances of the model in native pastures for nutrients.

The cumulative load graph for W4 (Figure 4-21) shows a close relationship between the

predicted values and the measured values. In the middle of 2002, there is a slight discrepancy in

the values, which can also be seen in some of the other months, namely, the end of 2001, 2003

and the beginning of 2002. However, overall, the curve for the predicted values closely matches

the curve for measured values, thus indicating a good degree of accuracy in the model.

The measured vs predicted scatter graph for W4 (Figure 4-22) indicates a slight under

prediction with regards to the data points on the graph. However, the best fit trend line indicates

a slight under-fitting with respect to the 1-1 trend line.

Among the model performances for all of the native pastures, the pasture where the

model was the least effective is Winter 8 (Figure 4-23) with a NS-Ceff value of -0. 1755, an

nMSE value of 1.1755 and an RMSE of approximately 1.0439 Kgs. According to the data shown

in the graphs, the model closely predicts the values for the months of May 2001, May 2002, May










and January 2003. The model overpredicts the values for September 2001, 2002 and 2003 and

December 2002. However, it still manages to maintain an acceptable degree of accuracy as can

be seen from the graphs.

The cumulative graph of the nutrient loads for W8 (Figure 4-24) indicates a discrepancy

between the predicted and the measured values. The curve for the predicted values follows the

curve for the measured values for the initial few months of the validation period. From August

2001, the curves start drifting apart as the discrepancy in the values increases. The measured vs

predicted scatter graph for W8 (Figure 4-25) indicates a higher degree of overprediction than is

seen on the graphs for the other native pastures. The best-fit line seems to be above the 1-1 trend

line.

Forage Growth

Model testing was done for the period of 2001 2003. As mentioned earlier, the forage

data recorded at BIR is not continuous and each of the readings for every month has range of

values, and hence, the possibility of erroneous values. The model was tested only for the

improved controlled pastures (Summer 1 and Summer 8) (Swain et al, 2007), which were

maintained at a zero cattle stocking rate. Moreover, the cattle movement is a part of a

management decision in QnD:BIR and for the testing period of the model, no management

options were used during the simulation to efficiently gauge the accuracy of the model without

any external influences. The testing procedure however is not as comprehensive as it was for

hydrology and nutrients due to the discontinuous and varied nature of the measured data.

The overall performance of the model was gauged by comparing the simulated results with

the range of measured values for the chosen pastures. Analysis of Summer 1 pasture (Figure 4-

26) indicated that the model was acceptably accurate and within the measured data range for

most of the testing period. It can be noticed that the model over-predicted the forage yield for a










few months and missed one increase in yield growth over the testing period in September 2003.

The NS- Ceff was calculated to be 0.2440 which is acceptable but does not reflect the variable

nature of the measured data.

Analysis of Summer 8 pasture (Figure 4-27), revealed a similar trend with the model's

overall performance was within acceptable limits. The NS-Ceff value was calculated as 0.2930.

An important observation of the model's design indicated that the simplistic empirical approach

followed by the model, is unable to account for the different factors that influence the forage

growth. However, the model overall was effective in simulating the growth trends of Bahia

grass.

Summary of Model Testing

The model was tested against the BIR measured data for the period of 2001 2003. The

graphs obtained are included in the appendix A. Moreover, several error estimation methods

were used to gauge the performance of the model over the 16 experimental pastures which were

explained in detail in chapter 3. After analyzing the output of all of the methods and graphs,

though the model seems to perform good and bad over all of the pastures, largely, the model

captures the trends of the peaks and lows, occasionally missing/over-predicting a few

peaks/lows. The RMSE values do not vary largely over the pastures, and the cumulative and

monthly graphs indicate that the model performance is within the acceptable limits of accuracy.

Collectively, the model performed well, with error estimations within the acceptable limits of

modeling standards.

Enterprise Wide Simulations and Scenario Analysis

The model development and testing have been discussed in the previous sections, which is

a requirement to validate the authenticity of any model. This section is the actual application of

the model for the whole beef cattle enterprise and the 68 pastures in the farm. The model was run










for the period of 2000 2003 for the whole farm and the total runoff, phosphorus load, grass

biomass and the monthly operational expenses of the model were measured under different

simulated rainfall conditions apart from the measured rainfall which is provided as input.

The model was run for the whole farm simulation following the time schedule discussed

earlier in chapter 3 with the move cows options being enforced every summer and winter to

move the cows from summer to winter pastures and vice-versa. Moreover, apart from the actual

rainfall, the conditions of less than regular rainfall and more than regular rainfall were also run as

separate scenarios.

Scenario 1: Measured Rainfall

The first scenario is the regular measured conditions of rainfall and temperature. The cows

are alternated between the Summer 1-8 pastures during the summer and the Winter 1-8 pastures

during the winter for the period of the run. The output values that are taken up for this analysis

included total monthly runoff, total phosphorus load, total grass biomass and total operating

amount available with the ranchers per month, with the assumption that BIR started out with

$1000000 as capital when the simulation period started. The operating income is not only an

indicator of the ranch operating costs and incomes, but also serves as a pointer to the cow-calf

ranch management operations, as they are the maj or influence in deciding the profitability of the

ranch.

Scenario 2: Low Rainfall

The second simulation involved reducing the amount of rainfall over the period of the run

by calibrating the measured values of the rainfall input to 15 percent less than that of a regular

period, hence causing more severe drought conditions. All other input parameters were kept the

same as rainfall is the maj or driver of QnD:BIR, it was chosen as the parameter to run different










scenarios for. Also the cow movement was differed and other non-experimental pastures were

used to move the cows during the winter namely, South marsh west and South marsh center.

Scenario 3: High Rainfall

The other end of the spectrum to lower rainfall is the higher rainfall year. The idea is to see

if more rainfall necessarily has an effect on the profitability of the enterprise and on the general

hydrology and nutrient loads of the region. However the cow movement in this case too is

similar to the other scenarios, wherein, the cows are moved from summer or improved pastures

to the unimproved pastures in the winter.

All of these scenario runs were run to provide an understanding of how the model can be

useful both to researchers and the rancher to provide a starter for future predictions based on

climate prediction data which can be continued as a part of this research.

Results and Discussion

The results of the whole farm simulation runs provided results which follow similar trends

to those seen in the experimental pastures during the validation period, as can be seen from the

graphs in Figure 4-28. The operating amount values are interesting due to the sinusoidal nature,

which further implies that profitability over the longer term required more than just cow

movement and calf sales, without provided excess support for the calves/cows to grow and

flourish, Though some months make good profits, overall, in order to maintain sustainability, the

conditions do need to be favorable or the operating costs setup within this model, must be

reduced.











Table 4-1. List of values of Nash-Sutcliffe coefficient (Ceff), Normalized Mean Square Error
and Root Mean Squae Error (n million liters) for runoff in summer pastures.


Summer Summer2 Summer3 Summer4 Summer5 Summer6 Summer7 Summer8
NS-
0.583 0.587 0.613 0.510 0.617 0.561 0.536 0.572
Ceff


nMSE 0.417 0.413 0.387 0.490 0.383 0.439 0.464 0.428


RMSE 4.179 4.368 4.346 4.113 5.025 5.417 5.164 4.809



Ceff m 0.57 0.5940 0.6210 0.5910 0.6260 0.5820 0.5380 0.5780




Table 4-2. List of values of Nash-Sutcliffe coefficient (Ceff), Normalized Mean Square Error
and Root Mean Squae Error (n million liters) for runoff in winter pastures.


Winter Winter2 Winter3 Winter4 Winter5 Winter6 Winter7 Winter8

NS- Ceff 0.729 0.674 0.382 0.809 0.695 0.738 0.715 0.705


nMSE 0.271 0.326 0.618 0.191 0.305 0.262 0.285 0.295


RMSE 5.902 7.231 6.977 5.608 8.806 6.687 8.215 7.151



Ceff m 0.656 0.604 0.491 0.663 0.616 0.633 0.649 0.618











Table 4-3. List of values of Nash-Sutcliffe coefficient (Ceff), Normalized Mean Square Error
and Root Mean Suae Error (nKgs for load in summer pastures.


Summer Summer2 Summer3 Summer4 Summer5 Summer6 Summer7 Summer8

Ceff 0.539 0.436 0.148 0.464 0.368 0.337 0.444 0.596


nMSE 0.460 0.564 0.851 0.535 0.631 0.662 0.555 0.404


RMSE 5.680 4.623 5.205 5.233 6.479 5.537 6.940 5.359


Ceff m 0.563 0.518 0.403 0.524 0.513 0.471 0.567 0.587




Table 4-4. List of values of Nash-Sutcliffe coefficient (Ceff), Normalized Mean Square Error
and Root VMean Su e Error (n Kgs)for load inwinter pastures.



Winter1 Winter2 Winter3 Wi nte r4 WinterS Winter6 Winter7 Wi nte r8

Ceff 0.411 0.534 0.504 0.694 0.287 0.586 0.461 -0.175



Nmse 0.588 0.465 0.495 0.305 0.712 0.413 0.538 1.175



RMSE 1.321 1.436 1.511 0.889 3.106 1.552 2.239 1.0439



Ceff m 0.383 0.511 0.436 0.567 0.476 0.572 0.529 0.303















S Runoff

45
g 40

& 35



10 ------- Measured

O .* *, 1 Model Predicted


Moth


Fiur 42Mothy uoff inthe Smmrpate


SS CuulaiveRuoff
250o oo oo







0 Moel Prdicte








c 0 00 0 0 0 0



Moth


Figure 4-3.Cumulative runoff in the SummerS pasture


40 00

35 00

30 00

25 00-

2000

15 00

10 00

5 00

0 00
0 00[


o Pred cted vs Measured






y=0799x
R2 eggs

Ceff= 0617


10 000 20 000

IMiasured


?0 000 40~ 000

Million;


Figure 4-4. A measured vs predicted scattergraph for the SummerS pasture runoffs



























..


S4 Cumulative Runoff

250



,150 '
10 ."

50 .-***** ------- Measured

O Model Predicted



Mot


S4 Runoff


Measured

-Mo~del Predicted


Orl~~PI~PI~N
dOOdOOd
0000000
rrlNNCIINNN
~CZ-~C-I-I1
armmarmic~m


Month


P)yr
OC
Nr*


Figure 4-5.Monthly


runoff in the Summer4 pasture


Figure 4-6.


Cumulative runoff in the Summer4 pasture


S4 Measured vs Predieted Scatte~graph


O Predc~tedys Measured






Ceff = 0510


3E 00

3C 00

26 00

2C 00

li 00

1C 00

5 CO

0 CO
C 000


10 000 20 000 3C 000 40 003

Measu red vlon


Figure 4-7. A measured vs predicted scattergraph for the Summer4 pasture runoffs













WV4- Runoff


70
60
5 0


5 20-
---- Measured
10 I ,,~
." I 1 Ve; t Model predicted


IVonh


WV4 Cumulative Runoff
400
'350 *
sr
2 00 *
r: r 150

100, I CrrrE.. Measured
50J
0 .f -Model Predicted


ooo~~Months

Fiur 49.uultie uoff n heWite4 asur
W4Mesue v PedcedError e -


Figure 4-8.Monthly


runoff in the Winter4 pasture


^PFreuteJ i Maf.1.sur~e




Ceff= 0800


1 1 1 |I I_|

5III

11111
I I [ ll|


II30 I ?l_ 2 ll||

Meas~rec


( 1_ |_1_11 4 1 _1|l
1\llI n;n


Figure 4-10. A measured vs predicted scattergraph for the Winter4 pasture runoffs















WN3- Runoff

70
P,60
S50



5 20 -II 's ------- Measured
10-
:..~~ Mod Moel Prededice
000~~N~m m
0000000000000000

0000000Months0


W3 Cumulative Runoff

350
S300
250 ..
200
-1 1 50,
"E 100
.***** ------- Measured
50
0 Model Predicted



ooo~~Months


Fiur 412Cuultie uof i te iner pstr


Figure 4-11.Monthly


runoff in the Winter3 pasture


W3MPueasured vs F 0.1 -.1-..-.irl--; .-[


40,00


j5000


25 00

20 00

15 00

10 00

5 CO

0 CO
0 00E


1 Pre leJe~ vs Measurac


v= 1256x
R 0 7:5

Ceff = 0382


10300 2C 000 30000 403000


Measured


w1II am


Figure 4-13. A measured vs predicted scattergraph for the Winter3 pasture runoffs

































































58 Measured vs Predicted Scattergraph


58 Phosphorus Load
40
S35
3 0 .1
,25
vl20
~15
S10 *******r I "" Measured
a 5-
O :1 *, -~11, I~d~l Model Predicted


ooo3~Months

Figre4-4.onhl poshouslod i h SmeSatr

SS -Cumlatie Pospousla






0 odel Pedicte







Figure 4-15.Cmuntlatv phosphorus load in the Summer8 pasture


o Pred cted vs Measure

y=0679x
R= 0 604

Ceff = 0596


100 10 000 20 000 30 000
Measured in Kgs


40 000 50 000


Figure 4-16. A measured vs predicted scattergraph for the Summer8 pasture Ph loads















SS Phosphorus Load
35
'30
S25
.9 20
'15-
10 *** Mea~sured

0 Model Predicted


oaoIV~onthsm












0 Moel Prdicte







Figure 4-18.Cmuntlatv phosphorus load in the Summer? pasture


S3 Measured vs Predicted Scattergraph


oPredictecivsMeasuni

y-0 913x
R2= 0467

Ceff = 0148


500o


)000 20000 30000 40000 5000C
Measured in Kgs


Figure 4-19. A measured vs predicted scattergraph for the Summer3 pasture Ph loads
















W4 Phosphorus Load

9D




2 ****M aue
1 .. oe rdce








Fiue4-20.otl phosphoru loditeWntr astured





..**
20

10 .* **** Mesure
00 0053 0





IVonths




Figure 4-21.Cmuntlatv phosphorus load in the Winter4 pasture


WV4 Measured vs Pr
1600

14 00

1200o

10 00

8 00

6 00

4 010 g ,'~ o

200 "

000


edicted Scattergraph


OPredlCtedv5 Measuni



y=0917x
R'- O 724

Ceff = 0694


0 00 0 000 10 000 10 000 20 000

Measured in Kgs



Figure 4-22. A measured vs predicted scattergraph for the Winter4 pasture Ph loads





























** ..


Figure 4-23.Monthly phosphorus load in the Winter8 pasture




W8 Cumulative Phosphorus Load

35





e 5 1
O 0 ....f- M de rdce


iMonths



Fiue42.uuaiepopou odi the Winterpasture


******* Measured

-Modei Predicted


000i
000


00000rr
00000


Months


W

14 00

12 00

10 00

800



200 O
DO
4 00


8Measured vs Predicted ~cattergraph


Y= 1.182%


Ceff= -0 175


O

5oo 000 100 15 000

Measured laKgs


Figure 4-25. A measured vs predicted scattergraph for the Winter8 pasture Ph loads


WS Phosphorus Load
















Summer 1- Monthly Forage Yield

10 10



4 149



m 3 --cft TI 1n.t 3 A Mean measured

e~ Model, Predchted



Time peio inmnt


Figur 4-6Motl foag yil fo h ume atr


000000~~~~33~~NNNNmmmmmm
00000~00005000000000000~
00000 00005000000000000
NNN~NNNN~N~N~~NNNNNNN~~N
,,
i m u7 ~ C" i 3 P7 LO ~ n rl im UI r- Ln ri i m LR ~~ r(
i

Time period(in months)


Figure 4-27.Monthly forage yield for the Summer 8 pasture


Summer 8 Monthly Forage Yield


-10









A Mean measure

-Model P-edicted













Total Forage Yield

4400
4300
ar 4200
fi 4100



E 3700
S3600 -Total Forage Yie d
3500
3400




;Months


Total Monthly Runoff





i 4

S2-
1 -Totai Monthly Runoff
0c





Months


Total Phosphorus Load
6000


S4000
~L3000


100 -o,11 1111 -Total Phosphorus Load
0
0 0 0 ~rrv r r



Months














Total Operating Amount


1200000
1000000
S800000
S600000
E 400000
200000


-Total Operati ng
Amount


000~~NNNmm
aaaaaaaaga
b 1; U r t;
m=Joa-ovov'a~Pc~
~4vlu_ ~~O

Months


Figure 4-28. A) Total Forage Yield in the ranch B) Total monthly runoff in the ranch. C) Total
phosphorus load in the ranch. D) Total operating amount at the ranch.









CHAPTER 5
CONCLUSION AND FUTURE WORK

Conclusions

Beef cattle enterprises and their management face several complex management and

political challenges in an already fragile ecosystem of south Florida. On one end, there exists a

struggle to maintain profitability and sustainability, and on the other, the effort to conserve the

fragile ecosystem of the region and hence the political pressure from the environmental

protection agencies. The current need is for a system that models the ecological issues of non

point source pollution, and interprets the results in a user friendly decision oriented format.

The obj ective of this research was to design and develop one such decision support tool

with the capability to model the ecological processes and interpret them as well. QnD: BIR was

intended as a model that can be used by both the research community and the ranchers

themselves with the intention of assisting the decision making process for managing the ranch.

Moreover, the research site chosen, Buck Island Ranch (BIR), provides a unique setting for

production-related, agro-ecological research. MAERC/BIR combines a research facility with a

commercial-scale, beef cattle enterprise (10,300 acres) to explore the role of long-term

ecological and social dynamics within sub-tropical grazing systems.

QnD: BIR is based on literature knowledge, actual laws of ecology, previous modeling

efforts and expert wisdom from researchers and ranchers, the intended users of the model. The

unique iterative model development methodology of QnD allows very close participation with

the researchers and ranchers during the whole process of model development to cater it to their

requirement. Moreover, the model is developed as an enterprise wide model, to be used on all of

the pastures within MAERC/BIR. The scale is yet another aspect that makes QnD:BIR unique.










Once the development stage was completed, QnD:BIR was tested on environmental data

from BIR for the period of 2000 2003 for sixteen experimental pastures including both

improved and native pastures. Specifically, QnD:BIR simulation results of monthly runoff,

phosphorus load and forage yield were compared with comparable field-scale data. After

analyzing the output of all of the statistical error estimation methods and graphs, largely, the

model captures the trends of the peaks and lows, occasionally missing/over-predicting a few

peaks/lows. The hydrology and nutrients simulations generally follow the trend mentioned

above. The forage yield simulations are also to the most part accurate and within the range of the

measured values. Moreover, the statistics of error estimation also indicate that the model's

performance is within the acceptable limits, given the coarse monthly time step. The model also

has added modules to simulate ranch cow/calf production and the incomes and expenditures

management which result in making it a complete enterprise wide decision support tool.

Future Research Recommendations

This modeling effort provided generally acceptable results and a helpful decision support

tool for researchers studying sustainable ranching and ranchers to assist them in managing the

ranch operations. However, as in any modeling effort, there is scope for future advancements

within the model. QnD:BIR design methodology allows such advancements to be integrated

easily into the model. The following are some recommendations for future work in this area.

Integrate Future Climate Predictions and Analyzing Different Scenarios

Climate prediction data from the Southeast Climate Consortium (SECC) can be integrated

into the model to simulate future ranch operation and ecological trends. The climate predictions

of future rainfall and temperature can be used as external input to the model. Various

management scenarios can be run on the model. The analysis of the output would assist the ranch

managers to assess the future of the ranch operation and help them prepare for it.









Improvement of the Cattle Production Module

The model currently has basic relationships governing cow production and movement.

More research could be done into various factors influencing the cow/calf production operation

ranging from buying of pregnant Heifers to culling of older cows. Moreover, detailed cow

movement records are being collected and analyzed recently at MAERC/BIR. This can be used

to further improve the cow herd movement simulations under various scenarios.

Integration of a More Complex Model into QnD

The hydrology and nutrient systems of QnD:BIR are developed using a simplistic

approach. In order to more accurately model the hydrology of this region, more complex models

like ACRU 2000 or the Century model can be integrated into QnD such that QnD:BIR uses the

complex structure of these models and their outputs to better analyze their effect on the ranch

operations of MAERC/BIR.

Integration of a More Advanced GIS Application Programmable Interface for Java

During the course of development and testing of this model, it can be noted that the GIS

interface that is used in QnD:BIR is an older version and does not allow the developer to use the

complete range of GIS capabilities. GIS is a powerful tool in spatial analysis, and further

exploration of some of the advanced features of GIS can empower the model to be more spatially

aware. This feature would help the model better gauge the topology of the region and hence

better predict the hydrology of the area. Moreover, a more advanced GIS link would also enable

the use of more than one principal layer to select and run the model simulations and hence

expand the modeling capacity of the model.

Overall, QnD:BIR modeling effort is a synthesis of both scientific data and expert

knowledge to create an easy to use decision support system with applications in education and

future research in modeling.



















S1- Runoff

40
S35
30
S 25
S20
15 1





Mot


Sl Cumulative Runoff

200
S180
.r 160
140 '"
,120

100

40 Measured
20
0 Model predicted

000000000000



Month


APPENDIX A
MODEL RESULTS AND GRAPHS


S1 Measured vs Predictedi Scattergraph


40 00

35 00

30 00


O Pred1te vs Measured




y= 1 077x
R2= 0 743


n0 non in0 ann 0n non 0o non an n


Millions


Measured


Figure A-1. A) Monthly runoff in the Summerl pasture. B) Cumulative runoff in the Summerl
pasture. C) A measured vs predicted scattergraph for the Summerl pasture runoffs














52- Runoff
40
S35
30
E 25
S20
15 1
10 ** Actual








20 00 Q 0 00 QO








Month


S2 Measured vs Predicted Scattergraph


oPredicted vs Measured






y=0910~x
R2= 0 67~6


1 000 10 000 20 000 30 000 40 000


Measured


Millions


Figure A-2. A) Monthly runoff in the Summer2 pasture. B) Cumulative runoff in the Summer2
pasture. C) A measured vs predicted scattergraph for the Summer2 pasture runoffs














53- Runoff

40
0 35
30




S10 ~ ~M .......--Mea sured

0 11 *. C1 Model Predicted

00C000000000



Month


oPredleted vs Measured




y= 0944x
R2= 0695


000 20 000

Measured


30 000 40 000
Millons


Figure A-3. A) Monthly runoff in the Summer3 pasture. B) Cumulative runoff in the Summer3
pasture. C) A measured vs predicted scattergraph for the Summer3 pasture runoffs


S3 Cumulative Runoff

250



S15 0
10

50 r-- ****** Measured
O Model Predicted



ooo~~~NMont

























.. *


S4 Cumulative Runoff

250






O odl reice




Monthrdite


Measured

SModel Predicted


0000000C~C
000000C




Month


0000
0000


S4 Measured vs Predicted Scatte graph


o Predr~ted.5).1PTIured


5c -Iil
I Il


II J Ill)I I o lln l
Measu red fllln


Figure A-4. A) Monthly runoff in the Summer4 pasture. B) Cumulative runoff in the Summer4

pasture. C) A measured vs predicted scattergraph for the Summer4 pasture runoffs


S4- Runoff





























i:


S5 Cumulative Runoff

250




~n150




50 I ------- Measured

O Model Predicted
00 0 00 0 0
000000000000


S5 Runoff


----- Measured

SModel Predicted


900~~
0000
tI000
NCVNCU
~$n_~
(OHL~:(C~


31~N~INmmm
00000000
00000000
CU N CV CV N N CV N
Hc~m L~" ~
r r

Months


Months


L


S5Measured vs Predicted Scattergraph


40 [IC
5
_35[IC
L

30 OC

25 OC


O Pred cted ..1 eauted






. = llFlig
5= = 1 6?


15 UI.




I1111


Al[III 000 l ll ? ll 4111 Ill

Measured M.illicon;


Figure A-5. A) Monthly runoff in the SummerS pasture. B) Cumulative runoff in the SummerS

pasture. C) A measured vs predicted scattergraph for the SummerS pasture runoffs

















SG Runoff

8 40
S35
30

E 25
S20-
cc* 15-
10 .- -I Ir Measured

0 Model Predicted



bbo~~Months


*

..**1


,,


**** *** Measured

-Model Predicted


OO~~~N
005000
003000
(UNNNNN
3r CZ r ~ b.111
m~njm~m


NNmmm
000~0
0000tr
NNNNN
~-11T~b.
mgmmw


Months


56Measured vs Predicted Scattergraph


4 Predictedys Measuree







y=0711x
FF=0609


I non in nn 70 ann 3n Inn 4nn on


Measured


Figure A-6. A) Monthly runoff in the Summer6 pasture. B) Cumulative runoff in the Summer6

pasture. C) A measured vs predicted scattergraph for the Summer6 pasture runoffs


56 Cumulative Runoff
















57 -runoff

40




~20
15
10 ....---- Measured

0 ** Model Predicted




OOC3~Months


57 Cumulative Runoff


Measured

-Model Predicted


ooooo
~Ur;r~N~~
r P.
mm(Umr~
~CO"~


,c~~VCi""~
~nn nn
ooooooo
rucrlrururururu
nrl~rzn
(Umm(Urclmar
vr~~vr~yvr


Months


40 00

j5 00

30 00

25 00

20 00

15 00

10 00

5 CO

0 CO
0 00E


+PFre leJe vs Measurac







y= 0742x
RE= 3607


10300 2C 1000

Measured


30000 400000

1\1II cn;


Figure A-7. A) Monthly runoff in the Summer7 pasture. B) Cumulative runoff in the Summer7

pasture. C) A measured vs predicted scattergraph for the Summer7 pasture runoffs














SB Runoff
vl40
35
30
L"25
20 2
-- 15
10 II U .L ......... Measured

0 "" ". Model predicted


oooIV~onthsn


SS Cumuatie unf







C Moel Prdicte



SS0 0000000000nof
00 0 00 0 0

Month


o Predicted vs Measured





y=0841x
R = 0649


00 10 000 20 000 SO 000 40 OC


MIlI ons


Measured


Figure A-8. A) Monthly runoff in the Summer8 pasture. B) Cumulative runoff in the Summer8

pasture. C) A measured vs predicted scattergraph for the Summer8 pasture runoffs














W1- Runoff

7D



~n4D

E 2D
Measured

,,. -l Model predicted

000000000000



Months


W1- Cumulative Runoff

350
30 .**
250
1t5
1 00
J 50

C Model Predicted
c00000000000
000000000000



Months


W1Measured vs Predicted Scattergraph7


e Pred cted vs Measured



y=0988ix
R2=079


0000 10 000 20 000 30 000 40 000

Measured Mlin



Figure A-9. A) Monthly runoff in the Winterl pasture. B) Cumulative runoff in the Winterl
pasture. C) A measured vs predicted scattergraph for the Winterl pasture runoffs















W2 Runoff

60





20 -3
Measured

Model Predicted








400 00000
300 00000






200
3 50
100 ...--M aue


50
0 Model Predicted



000~IVNonths


40 00

-35 00

30 00

25 00

20 00

15 00


OPrPed cted vs Measurt


y= 0796x
R2=0706


0000 10 000 20 000 30 000 40 000

Measured Millons



Figure A-10. A) Monthly runoff in the Winter2 pasture. B) Cumulative runoff in the Winter2

pasture. C) A measured vs predicted scattergraph for the Winter2 pasture runoffs





W3 Cumulative Runoff
350
S300
250 ..
200
S150
** "" Measured
50
SModel Predicted


lvonh


*** Measured

-Model Predicted


0000000lm
0000000


0000000lP
0000000


Months


W3 Measured vs P~edicted Scattereraph


)PrcICC ICJ vSMCDSur:C


= 1 2563
R= 0 755


10 00

5 CO

0 CO1
0 00[


10300 2C 000 30000 403000

Measured W1IlI en;


Figure A-11i. A) Monthly runoff in the Winter3 pasture. B) Cumulative runoff in the Winter3
pasture. C) A measured vs predicted scattergraph for the Winter3 pasture runoffs


W3- Runoff













WV4- Runoff


70
60
5 0


2 20-
---- Measured
10 I ,,~
." I 1 Vle t Model predicted


IVonh


W4 Cumulative Runoff
400
350
c 300
250
S 200 -
S150
1UU I I Measured
50
0 Model Predicted





400 00 0 0 0 0
U PeclUeO vsMeaura
35 00 ~N~~
T0 0 o 4 /=0 09
~e' P= 0 82


:0 00

15 00

10 00

5 CO

0 COI
0 01(


10300 2C 000 :0000 43000
a....\II ...Mcln;


Figure A-12. A) Monthly runoff in the Winter4 pasture. B) Cumulative runoff in the Winter4
pasture. C) A measured vs predicted scattergraph for the Winter4 pasture runoffs













W5 Runoff

80
S70


c 8 40
cr 30
20 rr rrn Measured
10 r srr
.* Model Predicted







400 ;d000

200 00000





Months


Figure A-13. A) Monthly runoff in the WinterS pasture. B) Cumulative runoff in the WinterS
pasture. C) A measured vs predicted scattergraph for the WinterS pasture runoffs


















































































Figure A-14. A) Monthly runoff in the Winter6 pasture. B) Cumulative runoff in the Winter6
pasture. C) A measured vs predicted scattergraph for the Winter6 pasture runoffs


W6 Runoff

70
S60
50 "
6 40
2n 0 ~L

,* ,q ------- Measured
10 s
d' -~ Model predicted





Months


W6 Cumulative Runoff
400
S350 ..
300
250


'100 .***** .....- Measured
50
0 p -Model Predicted

000000000000



Months














WN7- Runoff

70
~60
50
a 40 *' ,-
30-
OC
S20-
,*, ------ Measured
10-
0 ** Model Predicted

000000000000




Months




W7 Cumulative Runoff
450
400
S350
'ts 300



100 --------- Measured
0 Model Predicted


000~~Months


W7 Measured vs Predicted Scatter graphic


4000

3500

30 00

25 00

20 00

1500


O Pred cted vs Meas



y=0 704x1
R"= 0 45


10 00

500

0 001
n1 ann 1


in ann 7n ann si onn an an


Measured


Figure A-15. A) Monthly runoff in the Winter7 pasture. B) Cumulative runoff in the Winter7
pasture. C) A measured vs predicted scattergraph for the Winter7 pasture runoffs














W8 -Runoff
7D
S6D

4D

5 2D
-. ** ---- --- Measured
1D
0 1\~ 1 Model Predicted







400C00 dC


r:IF






Months


Wu28Measured vs Predicted Scattergraph


O Pree ~eJi vs Measurne


y=0751r
R2=073?


5 CO -

0 CO *4
0 00E


10300 2C 000 20000 403000

Measured wlin


Figure A-16. A) Monthly runoff in the Winter8 pasture. B) Cumulative runoff in the Winter8
pasture. C) A measured vs predicted scattergraph for the Winter8 pasture runoffs















S1- Phosphorus Load
40
u 35
30
S25
20
0 15
10 -o Measured
5 Model Predicted







20000 C



so :
40/ **** M asre




Months


51Measured vs Predieted Scattergraph


45 00 -

40 00

35 00

30 00

25 00

20 00

15 00

10.00

5.00

0 00
0000


+ Predictedys Measured


y=0696x
R2 0 556


70000 30000 40000 50000


Measured in Kgs



Figure A-17. A) Monthly phosphorus load in the Summerl pasture. B) Cumulative phosphorus
load in the Summerl pasture. C) A measured vs predicted scattergraph for the
Summerl pasture Ph loads














S2- Phosphorus load

35
S30
-o25



2 10
rat. ******* M/easurec
O 5-
I II. I F- Model Predicted
0 *~NNN
000000000000



Months


S2 Cumulative Phosphorus Load
160



240


Moth


S2 Measured vs Pirodicted Scattorgrarph


oPred cted vs Measured

y = 887x
R2= 0562


00 10 000 20 000 30 000
Measured in Kgs


40000 50000


Figure A-18. A) Monthly phosphorus load in the Summer2 pasture. B) Cumulative phosphorus
load in the Summer2 pasture. C) A measured vs predicted scattergraph for the
Summer2 pasture Ph loads














S3 Phosphorus Load
35
'30
S25
.9 20
L15-
10 *** Mea~sured

0 Model Predicted


oao~~Months


S3- Cmulatve PospousLa
160o oo oo






100
8 10

a 20 .



Moth


S1 Mpaclured vz PrejiCted SCaitterg~raph


= PIIfIIV MPR .:
F = .4tl7


IIII


Measured in Kgs


00IIIIII


Figure A-19. A) Monthly phosphorus load in the Summer3 pasture. B) Cumulative phosphorus
load in the Summer3 pasture. C) A measured vs predicted scattergraph for the
Summer3 pasture Ph loads
















S4 Phosphorus Load

no
sc 35
S30
m 25



"0



< 0 -
P. ~ModelPMonthse


-- Measured

-Model Predicted


rlrl~
88ooo
onoon
znczn
mplmm~


~NmmPr
00~0C
Onc)i3c~
NNCI~~~
Z d C Z C-
rrr~rsme
~CI'~L"


Months


54 Measured vs Predicted Scattergraph
45 00

40 00

35 00

30 00 -1 O '

25 00

20 00-





5.00

000


oPredictedvs Measured



y=0.77x
R 0 527


0 000 10 000 20 000 30 000 40 000 50 000

Measured in Kgs



Figure A-20. A) Monthly phosphorus load in the Summer4 pasture. B) Cumulative phosphorus

load in the Summer4 pasture. C) A measured vs predicted scattergraph for the

Summer4 pasture Ph loads






102


S4 Cumulative Phosphorus Loa d

































































SS Measured vs Predicted Scattergraph



.,'



,


o / co

0


00


45 00

40 00

3500

30 00

25 00

20 00

15 00

10 00

500

000
0 0


OPredlCtedv5 Measuni

y=0631x
R= 0 423


20 000 30 000

Measured in Kgs


40 000 50 00C


Figure A-21. A) Monthly phosphorus load in the SummerS pasture. B) Cumulative phosphorus
load in the SummerS pasture. C) A measured vs predicted scattergraph for the

SummerS pasture Ph loads


SB Phosphorus Load







" 0









2 50




50 to Measured

0 Modlel Predicted



oooo~Months















SS Phosphorus Load
35


a 25
S20


10 ****** Measured

.. -I P Model Predicted
0 **~~N ~
000000000000

0000MonthsO


S6 Cumulative Phosphorus Load
160
?s 140 1 ,

"100
~n80 1'

n 4U I ******* Measured
a20
c rees Model Predicted


Month


S6 Measured vs Predicted Scatter graph






+6






O ~


O Predleted vsMeasured


y= 0 717x
R2=0432


00 10 000 20 000 30 000
Measured inKgs


40 000 50 OC


Figure A-22. A) Monthly phosphorus load in the Summer6 pasture. B) Cumulative phosphorus
load in the Summer6 pasture. C) A measured vs predicted scattergraph for the
Summer pasture Ph loads














S7 Phosphorus Load
45
2Z 40
.E 3 5
30;
25 ..
S20 .

CP 10 -' ******* Measured
S5
0 : ** Model Pred cted


Mont s


S7 Cumulative Phosphorus Loa d
300
.250 ..


in150


5 10 ......, ******* Measured
0 -Model Predicted


Month


S7 Measured vs Predicted Scatter graph


a Predicted vs Measured

y= 0524x
RZ=0495


O




10
10
000 100 200 000 40
Mesre n g


50000


Figure A-23. A) Monthly phosphorus load in the Summer7 pasture. B) Cumulative phosphorus
load in the Summer7 pasture. C) A measured vs predicted scattergraph for the
Summer pasture Ph loads














58 Phosphorus Load
40
S35

,25 '
vi20 .
~15
S10 *******r I "" Measured
S5-
0 N Model Predicted


ooo3~Months


S8 Measured vs i-----: n.. 1 .-:I .ir. .2 ,.:


o Pred cted vs er:.urel

y= Il)791
P-=lmil4


Mleasuredin Kgs


40000 50000


Figure A-24. A) Monthly phosphorus load in the Summer8 pasture. B) Cumulative phosphorus
load in the Summer8 pasture. C) A measured vs predicted scattergraph for the
Summer pasture Ph loads


S8 Cumulative Phosphorus load
250

.E 200 ,

D 150 .

S100

50 ** ******* Measured
o
0 -Model Predicted
00 00 00 0 0
000000000000

ooaooMonths














W1- Phosphorus Load
12




c 2
0 P- M d l..., ..
0 0 0 0 0 0 0 0
oc co oo oc
40404
Motl


W1- Cumulative Phosphorus Load
40
S35
30
S25 -
in20
$15 ...
~n10 ******* Measured

0 Model Predicted


Month


W1Measured vs Predicted Scattergraph


oPredlCtedv5 Measuni


y=0719x
R2= 0455


14 nn


000 wet
0 000


S10 000
Measured inKgs


15 000 20 00C


Figure A-25. A) Monthly phosphorus load in the Winterl pasture. B) Cumulative phosphorus
load in the Winterl pasture. C) A measured vs predicted scattergraph for the Winterl

pasture Ph loads







107
















W2- Phosphorus Load

10


5y
4
c*** M aue
1 C- M de rdce
0

Month


W2- Cumulative Phosphorus Load

50

S40 ...*
~r35 .. *

30 .


P 10 *** Measured

0 Model Predicted
000000000000
000000000000




Montil5


W2 Measured vs Predicted Scattergraph


1 B.00

14 00

12.00

1000

8 00

6 DD





0 [00


o Pre led a Muerrl

y= 0.579x
R' = 0581


Measured In Kgs


15 000 20 000


Figure A-26. A) Monthly phosphorus load in the Winter2 pasture. B) Cumulative phosphorus
load in the Winter2 pasture. C) A measured vs predicted scattergraph for the Winter2

pasture Ph loads














W3- Phosphorus load
14
00




n 0
00 0 00 0 0
00 00 00 00
404040






L 25

10 ******* Measured
r5
0 -l Model Predicted



0000~~Month


W3 Measured vs Predicted Scattergraph


o Predleted vs MeasurE


R2=0492


Measured in Kgs


15 000 20 000


Figure A-27. A) Monthly phosphorus load in the Winter3 pasture. B) Cumulative phosphorus
load in the Winter3 pasture. C) A measured vs predicted scattergraph for the Winter3

pasture Ph loads














W4 Phosphorus Load



.57






Moth


WN4 Cumulative Phosphorus Load
40
? 35
30
25 **

S15 ******
~o10 ,------ Mecasured
F05
0 O .... -Model Predicted


ooo~IV~onths


WN4 Measured vs Prodicted Scattergraph


OPredletedvs Measuni


y=0917x
R2= 0 724


0 O


0 000 5 000 10 C00
Measured in Kgs


15 000 20 00C


Figure A-28. A) Monthly phosphorus load in the Winter4 pasture. B) Cumulative phosphorus
load in the Winter4 pasture. C) A measured vs predicted scattergraph for the Winter4

pasture Ph loads















W5 Phosphorus Load
18
S16


1 l


r. ******* Measured

g 7.. -IModel ar-edicted



Moth


WN5 Cumulative Phosphorus Load

90
v 80 .
.5 70 *
S60
.9 50 .**
~40
so ....
S20 ." ****** Measured
a 10
0 ,,, Modei Predicted

000003000000C
000003000000C



Months


oPredletedvs Measured


y=0298x
R2=0369


W5 Measured vs Predicted Scattergraph


I
O

O
o .~ o

I


00


0 00


5 000 10 000

Measured in Kgs


15 000 20 00C


Figure A-29. A) Monthly phosphorus load in the WinterS pasture. B) Cumulative phosphorus
load in the WinterS pasture. C) A measured vs predicted scattergraph for the WinterS

pasture Ph loads






111














W6 Phosphorus Load
16
~c14
12 *2



2-

0 1. odel Predicted


Moth



6005006
50O








Months


W6 Measured vs Predicted Scattergraph


O Predleted vs Measured


y= 0547x
R2=0612


00 5000 10 000
Measured in Kgs


15 000 20 OC


Figure A-30. A) Monthly phosphorus load in the Winter6 pasture. B) Cumulative phosphorus
load in the Winter6 pasture. C)A measured vs predicted scattergraph for the Winter6

pasture Ph loads














W7 Phosphorus Load
18
P 16
.5 14
.2 10 : e










S60



O 20
0 Model Predicted


"Months


W7 Measured vs Predicted Scattergraph


OPredlctedvs Mve~asured


y=0411x
R2= 0495


0 /


00


600

4 00

200


0 0


000 10 000
Measured in Kgs


15 000 20 00C


Figure A-31. A) Monthly phosphorus load in the Winter7 pasture. B) Cumulative phosphorus
load in the Winter7 pasture. C)A measured vs predicted scattergraph for the Winter7

pasture Ph loads





113



























:: .* .


W8 Cumulative Phosphorus Load
35
be
30
o25 ...
a 20 .=*
S15

n ******* Measured
0 5
0 ......f Model predicted



oo3~~Months


W8Measured vs Predicted Scattergraph

















000 5000 10 000 15 000 20 OC

Measured in Kgs


******* Measured

-Modei Predicted


00000C\
00000


000


Months


o Predleted vs Measured


y=1 182x
R2 = 0534


Figure A-32. A) Monthly phosphorus load in the Winter8 pasture. B) Cumulative phosphorus
load in the Winter8 pasture. C)A measured vs predicted scattergraph for the Winter8

pasture Ph loads


WS Phosphorus Load









LIST OF REFERENCES


American Society of Civil Engineers (ASCE) Task Committee on Definition of Criteria for
Evaluation of Watershed Models of the Watershed Management Committee. 1993. Criteria
for evaluation of watershed models. Journal of Irrigation and Drainage Engineering 119(3):
429-442.

Arnold JG, Allen PM, and Bernhardt G. 1993. A comprehensive surface-groundwater flow
model. Journal of Hydrology 142:47-69.

Arnold JG and Fohrer N. 2005. SWAT2000: current capabilities and research opportunities in
applied watershed modeling. Hydrological Processes 19:564-572.

Arthington, JD, Roka, FM, Mullahey, JJ, Coleman, SW, Lollis, LO and Muchovej, RM (2005 in
review). Integrating Ranch Forage Production, Cattle Performance and Economics in
Ranch Management Systems. J. Range Manage.

Arthington, JD, FM Roka, JJ Mullahey, SW Coleman, LO Lollis, and RM Muchovej. 2006.
Integrating ranch forage production, cattle performance and economics in ranch
management systems. Rangeland Ecology and Mgmt. 60(1).

Beasley DB and Huggins LF. 1980. ANSWERS (Area Non-point Source Watershed
Environment Response Simulation), User' s manual. Purdue University, West Lafayette.

Beven KJ and Kirkby MJ. 1979. A physically-based variable contributing area model of basin
hydrology. Hydrology Science Bulletin 24(1):43-69.

Booch, G. 1994. Obj ect Oriented Analysis and Design with Applications. Benj amin/ Cummings,
Redwood City, CA, .

Byard, C. Object-oriented technology a must for complex systems 1990. Computer Technology
Review 10, 14, 15-20.

Campbell KL, Kiker GA, and Clark DJ. 2001. Development and testing of a nitrogen and
phosphorus process model for Southern African water quality issues. 2001 ASAE Annual
International meeting. Paper No. 012085, St. Joseph, MI.: ASAE.

Campbell, KL, Capece, JC and Tremwel, TK 1995. Surface/Subsurface Hydrology and
Phosphorus Transport in the Kissimmee River Basin, Florida. Ecological Engineering Vol
5(2): 301-330

Capece, JC, Campbell, KL, Bohlen, PJ, Graetz, DA and Portier, K. (2007). Water Quality
Impacts of Beef Cattle Ranches in the Lake Okeechobee Basin. Rangeland Ecology and
Management.

Capece, CJ, KL Campbell, PJ Bohlen, DA Graetz, and KM Portier. 2006. Soil Phosphorus,
Cattle Stocking Rates and Water Quality in Subtropical Pastures in Florida. Rangeland
Ecol. and Mgmt. 60(1).










Chen DX, and Coughenour MB. 1994. GEMT & a general model for energy and mass transfer of
land surfaces and its application at the Fife sites. Agricultural and Forest Meteorology 68:
1451171.

Clark DJ, Kiker GA, and Schulze RE. September 2001. Obj ect-oriented restructuring of the
ACRU agrohydrological modeling system, Tenth South African National Hydrology
Symposium 26-28.

Crawford NH and Linsley RS. 1966. Digital Simulation in Hydrology: The Stanford Watershed
Model IV. Technical Report no. 39, Department of Civil Engineering, Stanford University,
Palo Alto, CA.

Dershem, HL and Jipping, MJ 1995. Programming Languages: Structures and Models. PWS
Publishing Company, Boston, MA.

Dunn, SM, Mackay, R., Adams, R., Oglethorpe, DR, 1996. The hydrological component of the
NELUP decision-support system: An appraisal. Journal of Hydrology 177, 213-235.

Evans BM, Sheeder SA, and Lehning DW. 2003. A spatial technique for estimating streambank
erosion based on watershed characteristics. Journal of Spatial Hydrology, Vol.3, No. 1.

FDEP (Florida Department of Environmental Protection). 2006. Integrated Water Quality
Assessment for Florida: 2006 305(b) Report and 303(d) List Update. Florida Department
of Environmental Protection. Division of Water Resource Management, Bureau of
Watershed Management, Tallahassee, Florida.

Greene, RG, Cruise, JF, 1995. Urban watershed modeling using geographic information system.
Journal of Water Resources Planning and Management 121(4), 3 18-325.

Henderson-Sellers, B. 1992. A Book of Object-Oriented Knowledge. Prentice- Hall, Englewood
Cliffs, NJ.

Hydrologic Engineering Center. 1981. HEC-1, Flood Hydrograph Package--User's Manual. US
Army Corps of Engineers: Davis, CA.

Innis GS. 1978. Grassland simulation model. Ecological Studies 26. New York: Springer-
Verlag.

Ito, K., Xu, X., Jinno, K., Kojiri, T., Kawamura, A., 2001. Decision support system for surface
water planning in river basins. Journal of Water Resources Planning and Management
127(4), 272-277.

Jamieson, DG, Fedra, K., 1996. The 'WaterWare' decision-support system for river-basin
planning. 1 Conceptual design. Journal of Hydrology 177, 163-175.

Jones JW, Hoogenboom G, Porter CH, Boote KJ, Batchelor WD, Hunt LA, Wilkens PW, Singh
U, Gij sman AJ, and Ritchie JT. 2003. The DSSAT cropping system model. Europ. J.
Agronomy 18: 235-265.










Kiker GA and Clark DJ. 2001. The development of a Java-based, object-oriented modeling
system for simulation of Southern African hydrology. 2001 ASAE Annual International
Meeting. ASAE Paper no. 012030. St. Joseph, MI.:ASAE.

Kiker, GA, Rivers-Moore, NA, Kiker, MK and Linkov, I. 2006. QnD: A modeling game system
for integrating environmental processes and practical management decisions. (Chapter in
Morel, B. Linkov, I., (Eds) "Environmental Security and Environmental Management: The
Role of Risk Assessment." Springer, Netherlands. Pp:151-185.

Kiker, GA and Linkov, I. 2006. The QnD Model/Game System: Integrating Questions and
Decisions for Multiple Stressors. (Chapter in Arapis, G., Goncharova, N. and Baveye, P.
(Eds) "Ecotoxicology, Ecological Risk Assessment and Multiple Stressors Springer,
Netherlands. Pp:203-225.

Knisel WG (ed.). 1980. CREAMS: A field scale model for chemicals, runoff, and erosion from
agricultural management systems. USDA Conservation Research Report 26: 643.

Krysanova V, Miiller-Wohlfeil DI, Becker A. 1998. Development and test of a spatially
distributed hydrological/water quality model for mesoscale watersheds. Ecological
Modelling 106: 261-289.

Krysanova V, Hattermann F, and Wechsung F. 2005. Development of the ecohydrological model
SWIM for regional impact studies and vulnerability assessment. Hydrologic Processes 19:
763-783.

Kunkle, WE, J. Fletcher, and D. Mayo. 2002. Florida cow-calf management, 2nd edition -
Feeding the cow herd. University of Florida/IFAS Extension Electronic Data Information
Service.

Ledgard, H. 1996. The Little Book of Obj ect-Oriented Programming. Prentice- Hall, Upper
Saddle River, NJ.

Leith H. 1975a. Modeling the primary productivity of the world. In: Leith H, Whittalcher, RH.
(Eds); Primary Productivity of the Biosphere. Springer-Verlag, New York.

Leonard RA, Knisel WG, and Still DA. 1987. GLEAMS: Groundwater loading effects of
agricultural management systems. Transactions of the ASAE 30 (5): 1404-1418.

Martinez, CJ 2006. Obj ect oriented hydrologic and water quality model for high-water-table
environments. PhD diss. Gainesville, F.L..: University of Florida, Department of
Agricultural and Biological Engineering.

McDonald MG and Harbaugh AW. 1988. A Modular Three-dimensional Finitedifference
Ground-water Flow Model. US Geological Survey, Techniques of Water Resources
Investigation Book 6, Chapter Al; 586 pp.










Mufioz-Carpena, R. A. Ritter and Y.C. Li. 2005. Dynamic factor analysis of groundwater
quality trends in an agricultural area adj acent to Everglades National Park. Journal of
Contaminant Hydrology

Ogden, FL, and A. Heilig, 2001, Two-dimensional watershed scale erosion modeling with
CASC2D, in Landscape Erosion and Evolution Modeling, R. Harmon, and W.W. Doe III,
eds., Kluwer Academic Press, New York,

Pandey, V. 2007 Analysis And Modeling Of Cattle Distribution In Complex Agro-Ecosystems
Of South Florida. PhD diss. Gainesville, F.L.: University of Florida, Department of
Agricultural and Biological Engineering.

Parton WJ, Schimel DS, Cole CV, and Ojima DS. 1987. Analysis of factors controlling soil
organic matter levels in Great Plains Grasslands. Soil Society of America Journal 51:
1173+1179.

Refsgaard JC and Storm B. 1995. MIKE SHE. In Computer Models of Watersheds Hydrology,
Singh V (ed.). Water Resources Publication: Highlands Ranch. CO; 809-846.

Reitsma, RF, 1996. Structure and support of water-resource management and decision-making.
Journal of Hydrology 177(3-4), 253-268.

Robson, D. Object-oriented software systems 1981. Byte 6, 8, 74-86.

Rosson, M. and Alpert, SR The cognitive consequences of obj ect-oriented design 1990. Human
Computer Interaction 5, 4, 345-379.

Rockwood DM, Davis ED, and Anderson JA. 1972. User Manual for COSSARR Model. US
Army Engineering Division, North Pacific: Portland, OR.

Sample, DJ, Heaney, JP, Wright, LT, Koustas, R., 2001. Geographic information systems,
decision support systems, and urban strom-water management. Journal of Water Resources
Planning and Management 127(3), 155-161.

Satti S 2002, Gwrapps: A Gis-Based Decision Support System For Agricultural Water Resources
Management, MS thesis Gainesville, F.L.: University of Florida, Department of
Agricultural and Biological Engineering

Seligman NG and Van Keulen H. 1981. PAPRAN: A simulation model of annual pasture
production limited by rainfall and nitrogen. p. 192-220. In M.J. Frissel and J.A. van Veen
(ed.) Simulation of nitrogen behaviour of soil-plant systems. Pudoc. Wageningen, the
Netherlands.

Sugawara M, Ozaki E, Wantanabe I, and Katsuyama Y. 1976. Tank model and its application to
Bird Creek, Wollombi Brook, Bihin River, Sanaga River, and Nam Mune. Research Note
11, National Center for Disaster Prevention, Tokyo, Japan.










Teague, W.R., Ansley, RJ, Pinchak, W.E. AND McGrann, J. 1995. A research-rancher
partnership to achieve sustainable use of rangeland. Texas TechResearch Highlights

Swain, H., PJ Bohlen, KL Campbell, LO Lollis, and AD Steinman. 2006. Integrated Ecological
and Economic Analysis of Ranch Management Systems. Rangeland Ecology Mgmt. 60(1).

Tanner, GW and McSorley, R. (2007). Bioindicators in ranch management systems. Rangeland
Ecology and Management.

Thornley JHM and Cannell MGR. 1997. Temperate grassland response to climate change: an
analysis using the Hurley Pasture Model. Annals of Botany 80: 205- 221.

Verberne ELJ. 1992. Simulation of the nitrogen and water balance in a system of grassland and
soil. Nota 258. DLO-Institut voor Bodemvruchtbaarheid, Oosterweg 92, 9750 RA, Haren,
Netherlands.

White JR. 1987. ERHYM-II: model description and user guide for the BASIC version. US
Department of Agriculture, Agricultural Research Service, ARS-59, Washington, D. C.

Williams JR and Hann RW. 1983. HYMO: Problem-oriented language for hydrologic
modeling--User' sManual. USDA: ARS-S-9.

Wirfs-Brock, RJ and Johnson, R.E. 1990. Surveying current research in object-oriented design.
Commun. ACM 33, 9, 104-124.

Yang, L. 2006. Coupled simulation modeling of flatwoods hydrology, nutrients, and vegetation
dynamics. PhD diss. Gainesville, F.L.: University of Florida, Department of Agricultural
and Biological Engineering.

Young RA, Onstad CA, Bosch DD, and Anderson WP. 1989. AGNPS: A nonpoint source
pollution model for evaluating agricultural watersheds. Journal of Soil and Water
Conservation 44(2): 168-173.

Yourdon, E., Whitehead, K., Thomman, J., Oppel, K. and Nevermann, P. 1995. Mainstream
Obj ects: An Analysis and Design Approach for Business. Yourdon Press, Upper Saddle
River, NJ.

Zhang, J., CT Haan, TK Tremwel, and GA Kiker. 1995. Evaluation of phosphorus loading
models for south Florida. Transactio ns of the ASAE 3 8(3): 767-773.









BIOGRAPHICAL SKETCH

Sudarshan Jagannathan was born in Mumbai, India, in 1984. He graduated with a Bachelor

of Engineering in Computer Science and Engineering from Osmania University of Hyderabad.

Shortly thereafter, he moved to Florida to pursue his graduate studies at the University of

Florida, in 2005. He pursued a concurrent Master of Science degree specializing in Agriculture

and Biological Engineering and in Computer Engineering.





PAGE 1

1 DEVELOPMENT OF A MANAGEMENT FO CUSED DECISION SU PPORT TOOL FOR OKEECHOBEE BASIN BEEF CATTLE AGROECOSYSTEMS By SUDARSHAN JAGANNATHAN A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Sudarshan Jagannathan

PAGE 3

3 To my father for all the support he has always shown me; to my mother for being a tremendous influence in my life, I would like to dedicate my thesis and all my achievements at the University of Florida. Her memory and blessi ngs have brought me this far.

PAGE 4

4 ACKNOWLEDGMENTS I am greatly indebted to Dr. Gregory A. Kiker for his complete support of my research interest in the field of Agricultural and Biologica l Engineering. I am sincerely thankful to him for his enthusiastic and generous support which he lped me learn the subject and his continued encouragement of my model deve lopment. His genuine interest in his work and the people he works, with have earned my highe st regard. Without his support and tireless help, it would not have been possible for me to have successfully completed my research. I would like to thank Dr. Rafael Munoz Carpena and Dr. Clyde Kiker for se rving on my committee and taking the time to meet with me and share their valuable insights about my research. I would like to acknowledge the immense value of the support and help of Dr. Chris J. Martinez, for the Buck Island Ranch data sets he provided and for his valuable inputs in the model development process. I would like to thank Mr. Gregory Hendricks for his time and for sharing his knowledge of hydrology and ecology with me, and Ms. Thej aswini Somasundaram for her support and encouragement throughout the course of this prog ram. I would like to also sincerely thank the faculty and staff of the department of Agricult ure and Biological Engineer ing for their technical and moral support. I would also like to express my thanks to all my friends and colleagues for their constant encouragemen t of all my endeavors. Most importantly, I would like to thank my fa ther for his support and encouragement, and the belief he has always had in me.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES................................................................................................................ .........8 ABSTRACT....................................................................................................................... ............11 CHAPTER 1 INTRODUCTION..................................................................................................................14 2 LITERATURE REVIEW.......................................................................................................17 Study Site: Buck Island Ranch...............................................................................................17 Experimental Pastures.....................................................................................................18 Overview of Past and Current Models....................................................................................19 A Brief Summary of Hydrologi cal and Nutrient Models................................................20 Overview of Forage Models............................................................................................22 Object Oriented Systems Development..................................................................................22 Decision Support Systems......................................................................................................26 Introduction to Questions and Decisions QnD.....................................................................27 3 METHODOLOGY.................................................................................................................36 Design of an Enterprise-Level Model for Buck Island Ranch QnD:BIR.............................36 QnD:BIR Hydrology.....................................................................................................37 QnD:BIR Nutrients.......................................................................................................39 QnD:BIR Forage Growth.............................................................................................40 QnD:BIR Beef Cattle Management..............................................................................42 Breeding...................................................................................................................42 Calving.....................................................................................................................43 Weaning and selling.................................................................................................44 Cow intake and waste...............................................................................................44 QnD:BIR Ranch Incomes and Expenditures.................................................................45 Addition of New Features into the QnD Model..............................................................47 Model Calibration, Validation and Data Representation........................................................49 Model Calibration............................................................................................................49 Model Validation.............................................................................................................50 Model Evaluation............................................................................................................50 Statistical Representation................................................................................................50 4 MODEL TESTING, RESU LTS AND DISCUSSION...........................................................53

PAGE 6

6 Model Inputs................................................................................................................... ........53 Model Outputs.................................................................................................................. ......55 Model Calibration.............................................................................................................. .....55 Hydrology...................................................................................................................... ..55 Nutrients...................................................................................................................... ....56 Forage......................................................................................................................... .....56 Model Validation/Testing.......................................................................................................56 Hydrology...................................................................................................................... ..56 Phosphorus Load.............................................................................................................60 Forage Growth.................................................................................................................63 Summary of Model Testing....................................................................................................64 Enterprise Wide Simulations and Scenario Analysis.............................................................64 Scenario 1: Measured Rainfall.....................................................................................65 Scenario 2: Low Rainfall..............................................................................................65 Scenario 3: High Rainfall.............................................................................................66 Results and Discussion....................................................................................................66 5 CONCLUSION AND FUTURE WORK...............................................................................80 Conclusions.................................................................................................................... .........80 Future Research Recommendations.......................................................................................81 Integrate Future Climate Predictions and Analyzing Different Scenarios......................81 Improvement of the Cattle Production Module...............................................................82 Integration of a More Complex Model into QnD............................................................82 Integration of a More Advanced GIS App lication Programmable Interface for Java.....82 APPENDIX A MODEL RESULTS AND GRAPHS.....................................................................................83 LIST OF REFERENCES.............................................................................................................115 BIOGRAPHICAL SKETCH.......................................................................................................120

PAGE 7

7 LIST OF TABLES Table page A-1 List of values of Nash-Sutcliffe coeffi cient (Ceff), Normalized Mean Square Error and Root Mean Square Error (in million liters) for runoff in summer pastures................67 A-2 List of values of Nash-Sutcliffe coeffi cient (Ceff), Normalized Mean Square Error and Root Mean Square Error (in million l iters) for runoff in winter pastures...................67 A-3 List of values of Nash-Sutcliffe coeffi cient (Ceff), Normalized Mean Square Error and Root Mean Square Error (in tons Kgs) for load in summer pastures..........................68 A-4 List of values of Nash-Sutcliffe coeffi cient (Ceff), Normalized Mean Square Error and Root Mean Square Error (in million l iters) for runoff in winter pastures...................68

PAGE 8

8 LIST OF FIGURES Figure page 2-1 Buck Island ranch with its summe r and winter experimental pastures..............................32 2-2 A simplistic UML look at the di fferent components of QnD model ................................33 2-3 Role of the coder or the code developer............................................................................33 2-4 The developers role...................................................................................................... ....34 2-5 The interaction of the players is mini mal with the java code or the XML files................34 2-6 The class diagram of a typical QnD system.......................................................................35 4-2 Monthly runoff in the Summer5 pasture............................................................................69 4-3 Cumulative runoff in the Summer5 pasture.......................................................................69 4-4 A measured vs predicted scatterg raph for the Summer5 pasture runoffs..........................69 4-5 Monthly runoff in the Summer4 pasture............................................................................70 4-6 Cumulative runoff in the Summer4 pasture.......................................................................70 4-7 A measured vs predicted scatterg raph for the Summer4 pasture runoffs..........................70 4-8 Monthly runoff in the Winter4 pasture..............................................................................71 4-9 Cumulative runoff in the Winter4 pasture.........................................................................71 4-10 A measured vs predicted scatterg raph for the Winter4 pasture runoffs.............................71 4-11 Monthly runoff in the Winter3 pasture..............................................................................72 4-12 Cumulative runoff in the Winter3 pasture.........................................................................72 4-13 A measured vs predicted scatterg raph for the Winter3 pasture runoffs.............................72 4-14 Monthly phosphorus load in the Summer8 pasture...........................................................73 4-15 Cumulative phosphorus load in the Summer8 pasture......................................................73 4-16 A measured vs predicted scattergra ph for the Summer8 pasture Ph loads........................73 4-17 Monthly phosphorus load in the Summer3 pasture...........................................................74 4-18 Cumulative phosphorus load in the Summer3 pasture......................................................74

PAGE 9

9 4-19 A measured vs predicted scattergra ph for the Summer3 pasture Ph loads........................74 4-20 Monthly phosphorus load in the Winter4 pasture..............................................................75 4-21 Cumulative phosphorus load in the Winter4 pasture.........................................................75 4-22 A measured vs predicted scattergra ph for the Winter4 pasture Ph loads..........................75 4-23 Monthly phosphorus load in the Winter8 pasture..............................................................76 4-24 Cumulative phosphorus load in the Winter8 pasture.........................................................76 4-25 A measured vs predicted scattergra ph for the Winter8 pasture Ph loads..........................76 4-26 Monthly forage yield for the Summer 1 pasture................................................................77 4-27 Monthly forage yield for the Summer 8 pasture................................................................77 4-28 Monthly Totals of Enterprise Wide simulation results......................................................79 A-1 Summer 1 Runoff........................................................................................................... ....83 A-2 Summer 2 Runoff........................................................................................................... ....84 A-3 Summer 3 Runoff........................................................................................................... ....85 A-4 Summer 4 Runoff........................................................................................................... ....86 A-5 Summer 5 Runoff........................................................................................................... ....87 A-6 Summer 6 Runoff........................................................................................................... ....88 A-7 Summer 7 Runoff........................................................................................................... ....89 A-8 Summer 8 Runoff........................................................................................................... ....90 A-9 Winter 1 Runoff........................................................................................................... ......91 A-10 Winter 2 Runoff.......................................................................................................... .......92 A-11 Winter 3 Runoff.......................................................................................................... .......93 A-12 Winter 4 Runoff.......................................................................................................... .......94 A-13 Winter 5 Runoff.......................................................................................................... .......95 A-14 Winter 6 Runoff.......................................................................................................... .......96 A-15 Winter 7 Runoff.......................................................................................................... .......97

PAGE 10

10 A-16 Winter 8 Runoff.......................................................................................................... .......98 A-17 Summer 1 Phosphorus Loads.............................................................................................99 A-18 Summer 2 Phosphorus Loads...........................................................................................100 A-19 Summer 3 Phosphorus Loads...........................................................................................101 A-20 Summer 4 Phosphorus Loads...........................................................................................102 A-21 Summer 5 Phosphorus Loads...........................................................................................103 A-22 Summer 6 Phosphorus Loads...........................................................................................104 A-23 Summer 7 Phosphorus Loads...........................................................................................105 A-24 Summer 8 Phosphorus Loads...........................................................................................106 A-25 Winter 1 Phosphorus Loads.............................................................................................107 A-26 Winter 2 Phosphorus Loads.............................................................................................108 A-27 Winter 3 Phosphorus Loads.............................................................................................109 A-28 Winter 4 Phosphorus Loads.............................................................................................110 A-29 Winter 5 Phosphorus Loads.............................................................................................111 A-30 Winter 6 Phosphorus Loads.............................................................................................112 A-31 Winter 7 Phosphorus Loads.............................................................................................113 A-32 Winter 8 Phosphorus Loads.............................................................................................114

PAGE 11

11 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science DEVELOPMENT OF A MANAGEMENT FO CUSED DECISION SU PPORT TOOL FOR OKEECHOBEE BASIN BEEF CATTLE AGROECOSYSTEMS By Sudarshan Jagannathan December 2007 Chair: Gregory A. Kiker Major: Agricultural and Biological Engineering Agricultural enterprises require resource management that in volve many trade-offs within a complex ecological and financial environment. As an example enterprise within South Central Florida, the MacArthur Agro-Ecology Research Center (MAERC) located on the Buck Island Ranch (BIR) in Lake Placid Florida has a major obj ective to optimize its l ong term sustainability in both ecological and economic facets. MAERC/ BIR combines a research facility with a commercial-scale, beef cattle enterprise (10,300 acres) to explore the role of long-term ecological and social dynamics within s ub-tropical grazing syst ems (www.maerc.org). In order to maintain long term viability and sustainability, a balance between ranch profitability and reductio n of non point source pollution effect s needs to be established and studied. A possible solution to th is challenge is to create a Decision Support System (DSS) for beef cattle enterprises. Such a DSS could se rve to communicate simulation results and metrics effectively to the ranch operators, whose focu s would be on profitabi lity, as well as the researchers and conservationists, whose focu s would be on limiting the effects of non point source pollution. Thus, the objective of this research project is to design and construct a decision support model of a beef cattle ranch system to simulate selected beef cattle and ranch

PAGE 12

12 management operations on a southern Florida be ef cattle enterprise and to explore the management decisions with respect to water resour ce factors such as runoff and nutrient loading. The Questions and Decisions (QnD) model system wa s created to provide an effective and efficient tool to integrate ecosystem, management, economic and socio-political factors into a user-friendly model/game fr amework. This model is a unique and new development since no other model before has mode led scenarios on a ranch-scale. The model is also good in that it is more than just a hydrological model but also a decision support tool for managers with a user interface that helps th em in real-time decision making. The QnD model links spatial components within geographic info rmation system (GIS) files to the abiotic (climatic) and biotic interactions that exist in an environmental system. QnD can be constructed with any combination of detailed technical data or estimated interactions of the ecological/management/social/economic forces influencing an ecosystem. The specific QnD version has been developed for the BIR (QnD:BIR) using the conceptual diagram which shows the integrated ecological and economic factors at the ranch-scale. QnD:BIR uses elements of the Standardized Pe rformance Analysis (SPA ) method applied to BIR to simulate elements of beef cattle pr oduction and economic dynamics. QnD:BIR uses simplified water and phosphorous dynamics at a m onthly time step generated from the long term research from southern Florid a beef and dairy cattle research QnD:BIR utilizes existing geographic information systems (GIS) coverage s and monitoring data available from the MAERC/BIR facility. QnD:BIR wa s tested on environmental data from BIR for the period of 2000 2003 for sixteen experiment al pastures including both imp roved and native pastures. Specifically, QnD:BIR simulation results of monthly runoff, phosphorus load and forage production were compared with comparable fieldscale data. Given the coarse monthly time

PAGE 13

13 step, simulations of these factors were gene rally acceptable for us e in the whole ranch simulations. Given potential climate data for the ar ea, specific scenarios were constructed to test different management scenarios in terms of P loading and cattle production metrics. The development of QnD: BIR provides a useful a nd modular system, capable of running various scenarios depending on the setup for simulating bo th environmental and enterprise functions, within an easy to use graphical in terface with the ability to move cows and manage the enterprise hands-on. Further model development and simulati on could be expanded to allow more detail in cattle response to temperature and surface water availability. Qnd: BIR is a simple model that uses empiri cal relations with accepta ble levels of accuracy (and a Nash-Sutcliffe coefficient of at least 0.5 mostly). The model also takes into account rainfall, water table depth, temperature, and so il characteristics for its hydrology and phosphorus cycle. However, since the model uses empirical relati ons, it cannot be applie d in conditions that differ vastly from the conditions present in BIR. Also, due to its simple nature it does not take into account factors such as light for ET, or drainage within and across pastures in the ranch and into the canals. Considering the significant posi tive qualities and certain limitations of the model, it can be said that the model is to be us ed more as a guideline to point the manager in the right direction for decision making than as a tool to provide ex act values or measures for runoff or phosphorus load in the long term.

PAGE 14

14 CHAPTER 1 INTRODUCTION Management of agricultural enterprises often oc curs within the context of complex environmental and societal challenges includ ing elements of economic, management and political viewpoints as well as th e often-explored technical perspe ctives. The quality of these agro-ecological systems is significantly affected by the rapid growth in the states population over the last three and a ha lf decades and concerns over non-point source pollution (2006 Integrated Water Quality Assessment ReportFDEP, 2006 ). As an example, beef cattle operations in sout h central Florida are c oncerned with long-term sustainability and viability under increasing re gulatory pressures. A dding to this existing challenge is the uncertainty of climate and envi ronmental drivers to ag roecosystems. Decision Support Systems (DSS) linking water resources a nd agriculture should be cognizant of the intersecting and sometimes conf licting goals of profitability, non-point source pollution effects and adaptive management. Recent advances in the climate sciences, including improved capabilities to forecast seasonal climate, have provided increased capabilities for developing useful decision support tools in service of specific ag riculture, forestry, and water resources management (SECC, 2007). These decision tools should communicate simulation results to decision-makers or stake-holders in their own language and metr ics whenever possible. Three fundamental questions arise when investigating water resource management within beef cattle enterprises. 1. What decisions are currently made that e ffect runoff, water qua lity on beef cattle ranches? 2. How accurate do forecasts have to be to be useful for beef cattle operations and environmental regulators?

PAGE 15

15 This research addresses some of these quest ions by exploring the role of managementfocused, agro-ecosystem models. The objectives of this research proj ect are the following: 1. Design and construct a decision s upport/scenario-based model of a beef cattle ranch using the QnD model system to simulate selected beef cattle and range management operations on a southern Florida beef cattle operation. 2. Test and calibrate the model using climate, hydrology, soil and forage monitoring data from representative pastures. 3. Explore the impact of scen arios on management decisions w ith respect to water resource factors such as runoff, nutrient loading, calf production and basic revenue/cost dynamics. The MacArthur Agro-Ecology Research Cent er (MAERC) located on the Buck Island Ranch (BIR), Lake Placid Florida provides a unique setting for production-related, agroecological research. MAERC/BIR combines a res earch facility with a commercial-scale, beef cattle enterprise (10,300 acr es) to explore the role of long-term ecological and social dynamics within sub-tropical grazing systems (www.maerc.or g). Recent multi-disciplinary research efforts at Buck Island Ranch (Swain et al. 2007; Arthington et al. 2007; Tanner and McSorely, 2007; Capece et al. 2007) have provided a useful dataset of climate, hydrol ogical, nutrient, vegetation, herbivore and production/economic data for fu rther integration and model development. Organization of this Thesis In addition to the introduction, this thesis research is divi ded into four chapters and a technical appendix. Chapter 2 provides a review of several items relevant to the research: an overview of the Buck Island Ranch monitoring effo rt; a brief review of the existing hydrological and nutrient modeling approaches that have been used in similar environmental systems; a detailed look into some of the related object oriented and decision support concepts and an overview of the QnD modeling system. In Ch apter 3, a detailed design of the QnD Modeling effort within Buck Island Ranch has been described along with the methodology for model

PAGE 16

16 calibration and testing. An anal ysis of the models structure and the methodology used to model the given system is presented, along with the results of m odel calibration and testing. The following chapter (Chapter 4) pres ents the results of QnD model testing on specific pastures (improved and native) within BIR. Chapter 5 provides an overall summary and lessons learnt from the modeling effort as well as potential next steps. The tec hnical appendix provides additional results and background fo r greater clarification in term s of object designs and model performance on specific pastures.

PAGE 17

17 CHAPTER 2 LITERATURE REVIEW This chapter provides a review of the con cepts and background used in developing a decision support system for a b eef cattle enterprise. An overv iew of Buck Island ranch, the research site is given in order to help put things into perspective, along with a brief history of a sample set models that have been developed to date. Following th is review, a section is provided that helps the reader to unde rstand the need for a more object oriented model and the object orientation concepts, which are discussed in de tail. The object oriented review section is followed by a detailed analysis of the QnD model and its overall structure, hence providing for an overall foundation of th e research project. Study Site: Buck Island Ranch Non-point source pollution is a major cause of concern in the south Florida water systems. Lake Okeechobee is one of the largest and most im portant water bodies of this region and a very popular site for research studies analyzing the effect of non-po int source pollution on a typical beef cattle agro-ecosystem (Martinez 2006, Yang 2006, Pandey 2007). According to Pandey (2007), the MacArthur Agro-ecology Research Ce nter, Buck Island Ranch (BIR, 4168, ha), Lake Placid, Florida, USA (27O 09N, 81O 12W), shown in Figure 2-1, was chosen for a research site since it is representative of th e subtropical, wet-prairie agro-ecosystem that exists in the Okeechobee watershed. For the proposed objective of this research too, this site would be well suited. Buck Island ranch presents a diverse and rich ecological setting, due to its coupling of sparse forests and wetlands in the midst of a commercial cattle ranch owned by the John D. and Catherine T. MacArthur Foundati on (Arthington et al ., 2007; Swain et al., 2007). The ranch area itself is at the center of a trib utary basin of Lake Okeechobee, the Indian Prairie/Harney Pond Basin, which has been drained for better past ure over several years (Arthington et al., 2007;

PAGE 18

18 Swain et al., 2007, Pandey 2007). This ranch enterp rise has been the center of research for almost a decade due to the willi ngness and interest of the ranchers themselves to limit the effects of non-point source pollution on the Lake Okeechobee watershed. A majority of the current and past research projects at Buck Isla nd ranch have focused on the experimental pastures (Summer 18, Winter 1-8) and to evaluate the effectiveness of the Best Management Practices (BMPs) to c ontrol the non-point pollution effects(Pandey 2007, Yang 2006). However, the BMPs mainly focus on reducing phosphorus loads in the region, and since BIR is primarily a commercial cattle ranch, one of the main management goals is to optimize the amount of beef production per unit area of land, or in other words improve the productivity of the enterprise. This trade-off bala nce tends to be the responsibility of the ranch managers to maintain. Cattle rotation is one of the practices that are co mmonly done at the ranch in order to maintain optimal feed for the cattle. The rotations are designed such that a cattle herd spends the majority of its time on the improve d summer pastures (May October), where the forage available is Bahiagrass ( Paspalum notatum ), which has a higher forage quality. Pandey (2007) explains the movement of cattle to be done due to two reasons : firstly, summer pastures are fertilized (NH4NO3 56 Kg N/ ha) (Arthington et al., 2007) in spring and, therefore, have better forage quantity and quality compared to winter pastures wh ich have never been fertilized (Swain et al., 2007). Secondly, wint er pastures are less intensivel y drained and as a result they are regularly flooded during the rainy season in summer. For the stocking rate experiment of 1998 2003 (Swain et al, 2007), the summer and winter experimental pastures were chosen to gauge the effect of different cattle stocking on the environment. Experimental Pastures The pastures in the ranch are divided into seasonal pastures depending upon where the cows are placed seasonally. The seasonally demar cated pastures fall mainly into two categories:

PAGE 19

19 summer pastures which tend to be improved, and wi nter pastures which tend to be native grass pastures. The summer pastures are eight approximat ely 20 ha (range = 19.0 to 22.1 ha) pastures where in the cattle are stocked during the summer months. These pastures are well drained and the major forage growth in this region is Bahiagrass ( Paspalum notatum ). Bahiagrass is generally considered to be higher in nutritional value than the native species of gras s that grows the winter or unimproved pastures (Kunkle 2001). The winter pastures similarl y are eight experimental past ures of approximately 32.2 ha (range = 30.3 to 34.1 ha). The forage on these past ures is not regulated and hence Bahiagrass is not the major forage species in the region. Compared to the summer pastures the winter pastures are not as well drained and retain water during large rain events. A network of shallow surface canals and drains carry all of the runoff water into the Harney Pond Canal and then onward into Lake Okeechobee. As a result, monitoring these pastures is essential to control th e addition of phosphorus to the lake. Overview of Past and Current Models Buck Island ranch has been the object of multiple research studies involved in hydrological and nutrient model development. Ma ny scientists and students from the University of Florida and other institutions have conducted se veral research studies at the site. Stocking rate experiments conducted during 1998 2003 have lead to several studies being conducted based of the data collected from the experiment. One such study involving the integration of Ranch Forage Production, Cattle Performance, and Economics in Ranch Management Systems by Arthington et al, ( 2007) indicated that stocking rates had a large effect on total produc tion and profitability. However, stocking rates had minimal to no effect on forage utilization or cattle performance. With no offsets in improved

PAGE 20

20 calving percentages, weaning weights or other meas ures of livestock performance, the inevitable outcome of lower stocking rate is impaired prof it potential (Arthingt on et al., 2007). Overall changes in stocking rate has a direct, one-to-one relationship w ith ranch revenues. If stocking rate effects surface water quality, there is a tradeoff between water quality improvement and profits from breeding cows. Moreover, an analysis of the soil phosphorus, cattle stocking rates, and water quality for the region (Capece et.a l., 2007) indicated better effec tiveness of approaches focused on decreasing phosphorus inputs and decreasing movement of accumulated soil phosphorus into surface runoff would be more effective than appr oaches focused cattle management for reducing P loads in surface runoff from cattle pastures. It also showed that th e stocking rate had no measurable effect on nutrients in surface runoff during 5 years of stocking treatments (Capece et.al., 2007). A Brief Summary of Hydrological and Nutrient Models This section provides a history of agricultural modeling and pr esents an overview of a few of the past and current models th at have been developed for vari ous landscapes in an attempt to help familiarize readers with the history of modeling and put into better perspective this modeling effort. Over the last few decades a va riety of ecological and biological models have been developed. The objective of this research entails the study of hydrology, nutrient movement and forage growth, along with beef cattle en terprise management, a nd a few modeling efforts focused on some of these fields are looked at in this section. The Hydrological models can be classified on a scale ranging from distributed physical based to the lumped conceptual models. In th e early 1970s the U.S. Environmental Protection Agency (EPA) began sponsoring a series of water quality models in response to the Clean Water Act, hence a majority of hydrological models us ed presently were developed during this time.

PAGE 21

21 Early conceptual hydrological models used a representation of basic laws of hydrology using differential equations and empirical algebr aic equations for modeli ng different processes (Yang, 2006). Some of the more popular models include Stanford Watershed Model (Crawford and Linsley, 1966), the SSARR (Streamflow Sy nthesis and Reservoir Regulation) model (Rockwood et al., 1972), Sacramen to Soil Moisture Accounting M odel (Burnash 1973), the HBV model (Bergstrom 1976), the tank model (Sugawara et al., 1976), the Xinanjiang model (Zhao et al. 1980), HEC-1 (Hydrologic Engineering Ce nter, 1981) and the HYMO (Williams and Hann, 1983), CREAMS (Chemicals, Runoff, and Erosion from Agricultural Management Systems) (Knisel, 1980) and the CREAMS derived GLEAMS (Groundwater Loading Effects of Agricultural Management Systems) (Leonard et al., 1987). Recently, the dynamic changes in the modeled areas are being explained by study in conceptual models of soil depletion, redistribution and moisture replenishment. (Arnold and Fohrer, 2005, Yang 2006). The next class of models is more physically-based than th e lumped conceptual models, which enable the more detailed representati on of the physical watershed and hence require simpler structure and fewer parameters. Some examples of semi-distributed physically based models include SWAT (Soil and Water Assessment Tool) (Arnold et al., 1993) SWIM (Soil and Water Integrated Model) (Krysanova et al., 1998, 2005), AGNPS (Agricultural NonPoint Source pollution model) (Young et al., 1989), TO PMODEL (a TOPography based hydrological MODEL) (Beven and Kirkby, 1979) and ANSWE RS (Areal Nonpoint Source Watershed Environment Response Simulation) (Beasley and Huggins, 1980), SDSM model (Singh 2001) with their application varying from examining wa ter quality to assessing the effectiveness of BMPs on runoff and nutrient lo ads of the given watershed.

PAGE 22

22 A third class of models is the completely physically-based distributed models. Some examples of physically based models are MODFLOW (McDonald and Harbaugh, 1988), MIKE SHE (Refsgaard and Storm, 1995) and GSSHA (Ogden 2001). Overview of Forage Models Forage models have been developed through ti me with varying complexities, ranging from a simple empirical model like th e Miami model (Leith 1975) to the more complex physiological models like the DSSAT (Jones 2003). The model developed mainly depends on the location for which it was developed and the types of forage it was dealing with. Ya ng (2006) has reviewed the structure of additional forage models su ch as PAPRAN (Seligman 1981), CENTURY model (Parton et al. 1987) ELM (Innis, 1978), ER HYM-II (White, 1987), GEM (Hunt et al., 1991), CCGRASS (Verberne, 1992) and GEMT (Chen and Cougheour, 1994) and the Hurley Pasture Model (Thornley and Cannell, 1997). All the models discussed in the above section were used extensively in their time and are applicable today. However, as the depth and computational requirement s of modeling complex ecosystems increases, the technical compet ence required by the model also compounds exponentially. This raises the need for a complete rewrite of the program code for any and every small change in the conditions or a change in th e site on which it is being applied. A requirement arises for a system that is capable of adapti ng to highly complex processes that might change from one location to the other wi thout having to rewrite the model, i.e. the issue of portability. Object Oriented Systems Development Object oriented systems are beginning to deve lop as a practical solution to the issue of model reusability. In recent times, the comple x and highly computationa l model development is turning to object orientation conc epts to better develop the model and at the same time keep the design simplistic for example the ACRU 2000 (Kiker et al. 2006). The most recent version of the

PAGE 23

23 ACRU (Agricultural Catchments Research Unit) (Schulze et. al, 1989) was reconstructed to be an object oriented model that described its la ndscape in terms of Components, Processes and Data. The ACRU 2000 (Kiker et al. 2006) is developed in Java and has proven to be highly extendable. Several modeling efforts have been conducted using this mode l in the south Florida region and the research site of Buck Is land Ranch. (Martinez 2006; Yang 2006; Pandey 2007). The following discussion highlights the vari ous facets of object orientation and its application in the world of mode ling. The use of object orientati on in modeling increased in the early 1990s with several models being developed using the concep t (Matsinos et al 1994, Mooij, 1996). Object Oriented Programming (OOP) originated as a development platform for physical modeling in Simula-67 programmi ng language. However, in the mid 1990s, it developed as the dominant programming methodology, largely due to th e influence of C++. In the past decade, with the rise in popularity of Java programming language, the use of OOP concepts has become more common, perhaps more importantly because of its implementation using a virtual machine that is intended to run code unchanged on many diffe rent platforms. This feature of portability of code is also being introduced now by Microsoft into the .Net framework. The magic quarks (Armstrong, 2006) of object orientation exists in the main components of it namely, inheritance, encapsulation, polym orphism and abstraction along with objects, instances and classes. The following sections provide some additional detail of these concepts in an attempt to understand their application wi thin environmental decision support systems. A class is the basic unit in OOP Classes are real world groups, which interact with each other through relationships. A class can be a species, a group, any unit that has common

PAGE 24

24 properties and implements common proce sses (Robson 1981, Rosson 1990). It can also be defined as a set of objects th at share a common structure a nd common behavior (Booch 1994). An object, on the other hand, is an instance of a class (Booch 1994) and can be anything from fish to cows to grass to anything that is being modeled. It is importa nt to understand that an object refers to the individua l but not the whole group. It is further explained by Armstrong (2006) as an individual, identifiable item, either re al or abstract, which co ntains data about itself and descriptions of its manipulations of the data. An object of a class has a ll the properties of that class and all of its parent classes. For exampl e, an individual cow ha s all the properties and processes of the cow species class as well as th e mammal class which would be the parent class. This concept of parent and child classes brings us to the first property of OOP, inheritance. Inheritance was introduced as a part of the development of OOP in 1967 in the Simula programming language. (Dershem, 1995). Some liter ature also inclines towards the idea that inheritance is the only unique feature introdu ced by OOP (Henderson-Se llers 1992). Inheritance has been defined as a mechanism by which objec t implementations can be organized to share descriptions (Wirfs-Brock, 1990) and also by Ar mstrong (2006) as a mechanism that allows the data and behavior of one class to be included in or used as the ba sis for another class. Inheritance signifies the property that any gi ven class can be derived from a nother class, and it in turn inherits all of the properties and the processes of that parent cl ass. This property continues all the way to the highest level of the hierarchy (Budd, 1991; Silvert, 1993). Such a hierarchical structure ensures that only very specific properties need to be specified for each individual object and it inherits most of the hi gher level properties from its pa rent class. This reduces the complexity of the code and increases the simp licity of the design itse lf (Moooij and Boersma, 1996).

PAGE 25

25 Another important reason for the simplicity of design of object orie nted models is the encapsulation property of OOP. It is described as a process used to package data with the functions that act on the data or more commonly as a property that hides the details of the objects implementation so that clients access the object only via its defined external interface (Wirfs-Brock 1990). Encapsulation is the containm ent of all of the processes and properties required and performed by the object of a class, with in that class or its pa rents. This reduces the coding overhead on the object. For example, a cow or a fish having all of its properties and processes within itself and allo ws another object to make it pe rform a given process at a given time. Encapsulation property makes such simplicity possible. Polymorphism was used in software devel opment and originates from it (Armstrong 2006). She also goes on to indicate that the literature appears to inconsistently apply the concept of polymorphism with some likening polymorphism to late binding or dynamic binding (Byard et al, 1990). Bringing together these conceptualiz ations, Armstrong (2006) defines polymorphism as the ability of different cla sses to respond to the same message and each implement the method appropriately. In modeling terms, polymorphism as be explained by the use of an example of feeding, which signifies grass to a cow but hunting game for a li on. The same feed process can be used to initiate both pr ocesses with the respective cl asses executing the corresponding processes of their species. Data abstraction originated in the 1950s a nd is commonly defined as the property of OOP to simplify complex real life situations by s uppressing irrelevant deta ils (Henderson-Sellers 1992, Ledgard 1996, Yourdon 1995). With the knowledge of OOP also comes the need to know the advantages and disadvantages of it. The object or ientation has several ups and downs discusse d in detail in the

PAGE 26

26 literature (Johnson 2000), but the mo st important fact about object orientation is the ease of model development. Moreover, the portability, reusability and the extensibility of the code are the most appealing facet of obj ect orientation to the modeling world (Johnson 2000). The ability to apply and use the model developed for one eco system and specific conditions and to easily change the model components to adap t to a totally different set of conditions is the ideal scenario for model design and code development. However, intended users of this model are th e ranchers who are responsible for making the decisions and managing the operations of the beef cattle enterprise. This further enhances the requirement for a link between the output of the complex hydrological and nutrient models and the decision making capability of the ranchers. The need is for a decision support tool that processes the data outpu tted by the models to a form which is easily interpreted by the decision makers. Adding a spatial, geographical informa tion systems module to it would further enhance the authenticity and the confiden ce of the model, and also improve its ability to accurately model the spatial variance of the landscape. Decision Support Systems Hydrologic models have served as a valuable tool for water resources management for many years (Greene and Cruise, 1995). The pressure to develop better and more accurate models requires the ability to better describe the lands cape spatially. This can be achieved by the use of geographical information system (GIS) within the model. A loose coupling of the simulation engine and the GIS alone is not sufficient to assist the decision-makers/ stake holders to efficiently make critical decisions. The necessary linkage is provided by the decision support system, which processes the data outputted by th e simulation engine and routes it to the GIS module to represent it in a form that would be an effective assistant to a stake-holder. Reitsma (1996) defines a decision support system (DSS) for water resources application as a computer-based

PAGE 27

27 system, which integrates state information, dynamic or process information, and plan evaluation tools into a single software implementation. In this definition, state information refers to data that represent the systems state at any point of time, process information represents the first principles governing resource behavior, and evaluation tools refers to software used to transform raw data into information used for decision making (Satti, 2002) A decision support system extends the scope of the simulation engine of any model to not only include fixed scenarios that are pre-determined and preset, but allows a more complete view of the various possible outcomes and options available to the decision-maker. In other words, it not only looks into what would happen, but also what option could be available to change it. The wide range of applications of DSS techniques for the study of water resources problems includes surface runoff, river basin management, urban storm water mana gement, groundwater contamination, have been discussed in literature (Dunn et al., 1996; Jamieson and Fedra, 1996; Ito et al., 2001; Sample et al., 2001). The current modeling effort is undertaken using one such decision support system that supports a GIS module and the above section is presented in order to help better understand the use and advantages of a GIS integrated model. Introduction to Questions and Decisions QnD The Q uestions a n d D ecisions (QnD) is a generic environmental modeling system that has been developed using Java based object oriented programming according to Kiker et al. (2006). The ideology behind this simple modeling structur e and approach is to present the model as a game (Figure 3-1) which will involve both managers and scientists. The game has appeal to both the communities due to its ability to output data and interpret it according to the need of the user. The managers or decision maker can use the simple user interface to assist them in making management decisions without having to proc ess numerical model output. The model uses

PAGE 28

28 several useful and easy to understand methods to appeal to manager using warning light, tabbed pane graphs indicating the trend of several important decision para meters and also a management toolbar that contains management option that can be applied fo r the next time-step. The user interface is developed using Java swing whic h provides a major advantage of platform independence. The model also integrates a geographical info rmation system (GIS) module into the user interface and has the ability to load several layers of shapefil e to provide a better understanding of the ecology and the landscape of the region. Th e GIS module is coupled to the Java model using GeoTools-Lite, an open source GIS appli cation programmable interface (API) for Java. The GIS module allows the user of the model to do several operations on the spatial units such as select, pan and zoom through a toolba r at the bottom of the screen. Th is allows the user to select one or more of the pastures and apply some specific management action to those selected pastures/spatial units, which accounts for an in teractive and dynamic m odeling experience. For the scientists and the number crunchers, the m odel has also has a more conventional form of output in the form of comma separa tes value files (.csv) that can be loaded into Microsoft Excel. An object oriented approach appeals to the world of modeling as the design of object, classes and methods is easier to reflect from thei r real world equivalents (Kiker et al., 2006). The benefits of an object oriented approach have already been discussed in earlier sections. The elemental objects in programming QnD are Com ponents, Processes and Data (Kiker et al., 2006). In QnD, all of the components are given a C prefix, the processes, P and the data a D prefix. So essentially the bu ilding blocks of QnD are the CComponents, PProcesses and the DData. Components are objects of interest (Kiker et al., 2006); a component can be any of the important physical players within the ecosystem, for example fish, grass, forest cover etc. The

PAGE 29

29 Components describe the constituents and entitie s within a spatial area or spatial unit. Every component has a set of data, sub-compone nts and processes associated with it. Processes are actions that i nvolve Components, and Data ar e descriptive objects assigned to components according to Kiker et al. (2006) and Kiker and Linkov (2006), in other words, processes are the tasks that any component perf orms to interact in some way with both the environment in general and other components. Pr ocesses use data to perform operations on the components to show the interacti on with other components. The da ta signifies the properties of the components which are modified by the pro cesses. Figure 2-2 describes a simplistic UML look at the relationship between components, processes and data. The relationship that is shown in the UML can be better described as, processe s and data are elements of a component, and a component can have one or more of each of these. A component can also contain subcomponents which can have their own processes a nd data associated with them. The processes too can have sub-processes contai ned within it. When the QnD: BIR model is discussed in the next chapter, a place of the sub-processes emer ge as the major placeholder for all of the calculations. In the UML use case diagrams, Figur e 2-3 to Figure 2-6, the role of different actors/players, the coders, model developers and the players, within the QnD system is made clear. Figure 2-3 describes the role of the coder or the code developer. Coders mainly interact with the Java code in the model. The design, development and maintenance of the model code itself, is the coders responsibility. The UML desi gn and overview of the system is also designed and maintained by the coder. A coder can intera ct with the players, the developers and the outside world overall through the QnD website with ideas and comme nts about the model

PAGE 30

30 design. The model deployment is th e final step of the developmen t. QnD is deployed using Java Network Language Protocol (JNLP) as described in the following sections. The developers role is depicted in figure 2-4. The model developer doesnt interact as much with the code itself but primarily with th e XML input files. The XM L input file provide a powerful and generic way to setup the QnD mode l, enabling the model developer to apply QnD to different sites without having to change the java code at a ll. Through the XML the developer can describe the modeled site, the components, processes and da ta discussed earlier. Moreover, the user interface is also descri bed in the XML by the model developer, all of the warning lights, charts, graphs, GIS images and management opti ons etc are designed and described in the XML by the model developer. The developer interacts w ith both the players of the system, who are the stakeholders, and the coder to best formulate the model development pr ocess. Once the model has been developed, the calibra tion and validation of the mode l is the model developers responsibility. The players or the users of the model consti tute the third class of people who interact with the model. Figure 2-5 describes the interacti on of the players to be minimal with the java code or the XML files. They mainly interact w ith the user interface of the model and the actual output files of the model. Their role as the user is to use th e model to better understand the implications of the decisions they are require d to make. Players sel ect the scenarios, the management options and essentially run the model. It is the interaction among these three actors that constitutes the major design and development of the model, by generating a coll aborative dialogue among st the users and the model developers, acquiring technical data and di scussing both informal rules of thumb and technical implications of manage ment decisions (Kiker et al., 2006). QnD allows both hard data,

PAGE 31

31 such as field-measured experiments, and soft data, such as experien tial learning or general impressions to be valid model inputs (Kiker et al., 2006). The result of the dialogue is conveyed to the coder in cases where there is a re quirement for code changes to the model. Kiker et al., (2006) describes model development methodology as iterative and interactive, involving alternative discussions w ith the stakeholders and model development, in order to best understand the requirement of the ma nagers and tailor the model to it. The object oriented nature of the model coupled with the expert knowle dge gained from the discussion with the stakeholders provides the b ackbone of the QnD model. Once an initial (prototype) version QnD has b een developed, it can be used as a game to stimulate further discussions between managers, sc ientists and stakeholders to try out different management alternatives and investigate possible repercussions of those decisions (Kiker et al, 2006; Kiker and Linkov, 2006). Once the development of the model is comp leted and the model has been put through validation, it can be deployed online as a webbased model-game using Java Network Language Protocol. Some of the earlier versions of QnD have been de ployed as a game making it a good resource for teaching and learning about the environment. Overall the object oriented QnD model proves to be a powerful and easy to use decision support tool, which couples an interactive design environment with a quick and efficient model development and deployment cycle. It establishes an essential link between the research oriented, complex hydrologic model and a simplistic user interface driver decision support tool used and preferred by the managers, and hence is id eal for use in the current research study.

PAGE 32

32 Figure 2-1.The figure shows the Buck Island ranc h with its summer and winter experimental pastures. (Kiker et al, 2006)

PAGE 33

33 Figure 2-2. A simplistic UML look at the different components of QnD model (Kiker et al. 2006). Figure 2-3. Role of the code r or the code developer

PAGE 34

34 Figure 2-4. The developers role Figure 2-5. The interaction of the players is minimal with the java code or the XML files

PAGE 35

35 Figure 2-6. The class diagram of a typical QnD system

PAGE 36

36 CHAPTER 3 METHODOLOGY The objective of this chapter is to provide a detailed account of the current version of the QnD model applied at the Buck Island Ranch (Q nD:BIR). The actual design and development of QnD:BIR is discussed, including the major processes involved in the model. Some of the actual objects (components, processes and data) incl uded in the model development are further explored in this chapter, followed by an account of some of the model calibration, validation, error quantization and data represen tation techniques commonly used. Design of an Enterprise-Level Model for Buck Island Ranch QnD:BIR The modeling effort for the QnD: Buck Island Ra nch was designed to be simplistic and based on literature-derived concepts, empirical data and expert knowledge. Most of the relationships in this version of QnD are either empi rical, calculated from the data re corded at the research site or is based on the basic laws of hydrology. Figure 3-1 shows a screen shot of the model with the cow icons and the GIS coverage of BIR. The basic spatial setup of the model divide s the whole area of BIR into 68 spatial units (CSpatialUnit), each representing the 68 pastures on the ranch. Each of these spatial units contain a CHabitat, which the main holde r of the other local components. In QnD:BIR, the habitat is considered to be default and is contained in all spatial units. Each of the CHabitats contains several local components includi ng CBahiagrass, CNativegrass, CUplandSoil, CWetlandSoil, and a CHerd wherever a herd is present. This is further clarified with the help of the UML class diagram in Figure 3-2. The grasse s and soil have a percent area DData, which signifies whether the selected spatial unit is an upland or a wetla nd, improved or native pasture. Moreover, another DData (DImprovedPasture) property is defined to signify whether the current spatial unit is an improved or a native pasture. There are severa l other DData objects and processes which are

PAGE 37

37 global to the model which signif y values used for calculation (DZero etc) or are input/output variables of the model. The main focus of this model is to simulate five major aspects of the Buck Island ranch ecosystem and ranch management operations: QnD:BIR Hydrology The hydrology of this region has been previously modeled by in-depth models like ACRU2000 (Campbell et al., 2001; Clark et al., 2001; Kiker and Clark, 2001; Martinez, 2006; Yang, 2006; Pandey, 2007) and WAM (SWET, 2002), wh ich have enabled hi ghly detailed subdaily modeling. On the other hand, the QnD Bu ck Island Ranch model has followed a very simplistic, deterministic method to model th e hydrology of this region by using simple relationships between the Ground Wa ter level data and the amount of available storage in the soil. This enables the developmen t of an effective hydr ological model in a short period of time which models the actual data on a monthly sc ale within acceptable levels of accuracy. The model structure is defined by a number of equations and relationships. One of the major inputs for modeling the hydrology of any regi on is the rainfall da ta. For the QnD Buck Island Ranch model, the Average monthly rainfall da ta is provided as input to the model. This model uses Ground Water Level as an input factor for calculati ng the runoff. Ground water table data is read in as input to the model. The hei ght of the possible available column of storage is calculated as the difference between the mean he ight of Buck Island Ranch from the sea level and the height of the gr ound water table. i.e. Possible Available Column of Storage (in mete rs) = Mean Height of Buck Island Ranch (in meters above Sea Level) Mean Height of Ground Water Table (in meters above Sea Level) (3-1)

PAGE 38

38 The available water storage is dependent on the soil type. Th is value, deduced from the Possible Available Column of Stor age, is based on the porosity and the field capacity of the soil. This factor accounts for the plant availabl e storage and the grav itational storage i.e. Available Water Storage (in mm) = Possible Available Column of Storage (in mm) (Porosity Field Capacity) (3-2) Moreover, an evapotranspiration (ET) factor is also added to this plant available storage which is calculate as a function of average temperature of the area during the month. There are other factors that affect ET, but since temperature is the major factor, it is taken into account in the model. The value of the possible runoff is calculated as the difference between the rainfall and the actual available storage. Runoff (mm) = Total monthly rainfall(in mm ) (Available Water Storage (mm) ET factor) (3-3) Finally, the total runoff volume is calculated for each pasture, taking into effect its area and the amount of runoff as a result of rainfall, i.e. the total runoff volume is calculated for every pasture using the equation: Runoff Volume (in L) = Area of the Pasture (square m) Runoff (mm) (3-4) To visualize the model using the object orient ed approach used by QnD, each of the values being calculated represents a DData, and each re lationship/equation is described by one or more PProcesses. The hydrology component is governe d by the soil component (CUplandSoil or CWetlandSoil), which contains the PProcesses. The inputs are again DD ata values which are either local to the spatial unit (DPercentAvailableStor age) or globally exis t (DMonthlyRainfall). In QnD terminology, the whole of hydrology is large a part of a single PProcess, PCalculateRunoff, that governs it. This PProc ess has includes several sub-processes which

PAGE 39

39 perform the calculations based on the relations hips described earlier. The detailed list of processes supported by QnD and th eir explanations are part of appendix B. These sub-processes systematically perform the calculations in the order they were setup, and the calculated results of total runoff volume is stored in the DData, DRunof fVolume. It is important to note that QnD only performs the processes at every time-step, if the component is present in specific spatial unit that is being updated. QnD:BIR Nutrients The nutrient movement of this region has b een previously modeled by in-depth models like ACRU2000 (Campbell et al., 2001; Clark et al., 2001; Kiker a nd Clark, 2001) and WAM (SWET, 2002) which have enabled highly-deta iled modeling. On the other hand, the QnD Buck Island Ranch model has followed a very simplistic deterministic method to model the nutrients of this region by using simple relationships. Th e simplistic approach enables the modeling of the complete ranch, on a broad scale, which helps to better understand ranch dynamics and the effect of the hydrology and nutrient cycle on ranch mana gement and profitability. In this model, nutrients in the region are divided into ex tractible phosphorus and stable phosphorus as perceived from the standpoint of the model. The extractible phosphor us is responsible for all the nutrient movement from the soil to the cows and in the hydrology. The stable phosphorus, as the name suggests, is considered to be relatively st able and is always present in the soil. However, every time step, there is a nutrient movement from the extractible phosphorus to the stable phosphorus and vice versa at different rates. This movement is gove rned by transfer coefficients that are used to transfer extrac tible to stable phosphorus and vice versa. This process of transfer can be shown by a series of equations as following: Extractible to Stable Phosphorus Transfer Amount (in kg) = Extr actible to Stable Transfer Coefficient Extractible Phosphorus (in kg). (3-5)

PAGE 40

40 Stable to Extractible Phosphorus Transfer Amount (in kg) = Stab le to Extractible Transfer Coefficient Stable Phosphorus (in kg). (3-6) The runoff event triggers the movement of a fr action of the Extractible Phosphorus with the runoff. This amount of phosphorus or in other words, the runoff phosphorus load (measured in kilograms) is calculated in the model using an empirical linear relationship that is derived directly from the measured BIR data. This linear relationship is based upon th e measured average runoff volume to the measured average phosphorus load for the set of native and improved pastures. Figure 3-3 shows the trend line characteristic of this linear relationship. The object oriented interpretation of these c onditions and components, which is used by QnD, describes the pools of phosphorus as a pr operty of the soil component (CUplandSoil or CWetlandSoil) of the sp atial unit. Each of these compone nts contains DData values which signify the presence and amount of extracta ble (DExtractableP) a nd stable (DStableP). PProcesses, PExtractableToStableTransfer a nd PStableToExtractableTransfer interpret the phosphorus transfer between the two pools, using the PTransfer process type, to the object oriented QnD model, incorporating the e quations discussed earli er in the section. The actual phosphorus load coming off of the spatial unit is governed by the PExportInRunoffImproved and PExportInRunoffNa tive processes which correspond to the Runoff vs. PLoad relationships described earl ier for the improved and native pastures respectively. Both of these processes affect the total phosphorus load DData (DPLoad), updating its value at every timestep. QnD:BIR Forage Growth The forage model component of QnD:BIR is again designed with the similar simplistic approach as the other components of QnD. Fora ge is a very important component of the BIR

PAGE 41

41 ecosystem and the sustenance of the ranch depend s on the cows having enough forage to feed on. Lack of forage growth also causes the cow/ calf condition to worsen and thereby result in unhealthy cows and additional expenditure in bu ying supplemental feed for the cow. QnD:BIR mainly considers there are two f actors that affect the growth of forage, the relative rainfall and the seasonal effect. For the effect of relative rainfall, the av erage monthly rainfall for the whole period is calculated for every month. Average monthly rainfall (month = Jan) (mm) = ( All years) Monthly rainfall (mm) / N (3-7) The average monthly rainfall is then used by th e model to calculate the relative rainfall for the current month. The relative monthly rainfall affects the forage growth as an empiri cal linear relationship. The values of this relationship are calibrated to best suit the conditions present at BIR, with within the confines of the acceptable results. Forage growth rate = (relative monthly rain fall) (using the linear relationship). (3-8) Total grass biomass (Total Forage) = Forage growth rate Total grass biomass (3-9) The seasonal effect on forage is more complex within the model than rainfall effect. Moreover, it also accounts for the wilting of the grass during dr y season. This is governed by an empirical curve, which is calibrated to suit the site being modeled i.e. the Buck Island ranch. The seasonal effect is governed by si milar equations as mentioned above: Forage growth rate = (current month) (using the curve). (3-10) Total grass biomass (Total Forage) (in 1000 kg) = Forage growth rate Total grass biomass (in 1000 kg) (3-11)

PAGE 42

42 Similar relations are used to model both na tive and improved pastures, where the herds are present and the herds consume the grass at a constant rate per day per cow. The specific value and its background are mentioned the next few subsections. With the coupling of the three factors, QnD:BIR overall presen ts a good and simplistic model design based off of expert opinion, literature and the ranche rs view of their ecosystem. Forage growth relations are incorporated in to the model using two simple PRelationship sub-processes within the PCalcula teForageGrowth process. These two PRelationships signify the rainfall effect and the seasonal effect on fora ge, defined earlier in this section. The forage relationships are specified w ithin the CBahiaGrass or the CN ativeGrass components of the model, which are to the most part similar. QnD:BIR Beef Cattle Management The cows are the most integral part of th e Buck Island Ranch ecosystem. As mentioned earlier, the ranch is primarily a beef cattle ranch and cows are a major asset and a very important player in the ecological balance at the ranch. The model is primarily a management tool, with its primary focus as the stakeholders and their interest s, which, in this case, are the ranchers at the Buck Island Ranch. Figure 3-4 shows a timeline fo r the ranch management operations, the model is primarily based on these timelines to eff ectively simulate the beef cattle enterprise. The modeling of the cows at the Buck Is land Ranch has been developed based upon the figure shown above. The model portrays three asp ects of the life cycle of the calves, i) the impregnation of the cows (breedi ng), ii) the birth of the calves, and iii) the time when the calves are weaned and sold. Breeding The calves born on the ranch that can be furt her divided into three categories on the basis of the part of the breeding season that the cows are impregnated in, namely: the calves born of

PAGE 43

43 cows that are impregnated early in the season, wh ich form the early coho rt; the calves born of cows that are impregnated in the middle of the season, which form the middle cohort; and the calves born of cows that are impregnated late in the season, which form the late cohort. This breeding season ranges from January to the end of April. The rate of impregnation during this season depends upon the climatic conditions, i.e., te mperature, rainfall, etc. and the condition of the cow. From the analysis of the ranch Sta ndardized Performance An alysis (SPA) data, the average impregnation rate of the cows all through the breeding season in the ranch is between 75% and 80%. The model assumes an almost e qual rate of impregnation throughout the breeding season, only varying due to the climatic conditio ns. During the entirety of the breeding season, all the herds of cows on the ranch are exposed to bulls for a time period ranging from 90-120 days, and the ratio of the number of bulls to cows is 1:25 (Source: Patrick Bolen). Calving The average gestation period of the cows is 9 months (Source: Patrick Bolen), hence the first calves are born around November from th e cows that were impregnated early in the breeding season. The calving continues all the way th rough to the end of the following February. The cows impregnated early in the breeding season give birth at the onse t of the calving periods, around the month of November, and the calves so born form the early cohort. Similarly, the middleand the Late Cohorts are born all th e way through February. The population of the cohorts can be calculated by the following equations: Early Cohort Population (cow units) = Early Pregnancy Rate Cow Population (3-12) Middle Cohort Population (cow units) = Midd le Pregnancy Rate Cow Population (313) Late Cohort Population (cow units) = Late Pr egnancy Rate Cow Population (3-14) The pregnancy rates for early, middle and late cohor ts are dependent on the condition of the cow.

PAGE 44

44 Weaning and selling From the time of birth, the calves gain 1.4 pounds a day, on an average (Kunkle et al., 2001). The calves are weaned around the month of May. At this stage, it is ensured that the calves reach their endured target-selling weight of about 492 pounds on an average.(SPA 2005) The calves that fall short of the required mark c ould be fattened by using one or more techniques such as supplemental feed or/ hormones. This is reflected in the model as a management decision that the user can make while running the model. On ce this is done, the calves are sold in cohorts. The selling of the calves is also a management d ecision which is part of the user interface in the model. This gives the user/stakeholder the option to sell any/all cohorts at the time he thinks is right for the ranch. Cow intake and waste Intake The average intake per day per cow is about 25.9 pounds of dry matter (Kunkle, 2002). Of all the food ingested by the cattl e, the utilization rate is about 55%, i.e. the nutrition level of the food ingested by the cattle (K unkle 2002). This utilization rate is higher for Bahiagrass as compared to the other varieties of forage (K unkle 2002). This percentage of utilization of the forage by the cows is calculated from the amount of grass/forage available on the pasture. After the forage is ingested by the cattle, the total gra ss biomass is also calculated and updated. Waste The cattle waste on the ranch is modeled primar ily for the phosphorus and nitrogen content present in it. The amount of phosphorus presen t in the cattle waste is about 0.044 kg/cow and about 0.04 kg/calf and the amount of nitrogen is about 0.019 kg/cow and 0.017 kg/ calf (ASABE Standards 2007). Of the total amount of phosphorus and nitrogen present in the cattle waste, a

PAGE 45

45 certain percentage is extractible. This value of extractib le phosphorus and nitrogen is accordingly updated from the phosphorus and nitrogen lo ads that are dropped on the soil. The beef cattle management is more complex to visualize in object oriented terms due to the variety of factors, parameters a nd processes involved. To start wit h, the cattle herd can be looked at as a component (object) and all of the QnD:BIR beef cattle management is contained within this local component. Several PPr ocesses are included within this component each signifying the various major aspect of ranch management discu ssed earlier in this section (breeding, calving and selling). It is important to note that, each of these b eef cattle management sections are implemented for each of the three cohorts that the calves are divided into. This further complicated the design of the QnD:BIR beef cattle management. An attempt has been made to retain the simplicity of the design of the XMLs to assist any of the stak eholders, who might be in terested in the design of the model, to understand it easily. QnD:BIR Ranch Incomes and Expenditures Consideration of the financial details in the m odel starts with the pr evious years ending balance as the current years be ginning balance. This amount is considered to be a fixed value for the purpose of this model. During the course of the management cycle, the ranch may incur a variety of both revenue as well as expenses. The ma jor source of revenue for the ranch is the sale of the calves, which are sold at an aver age price of about $1.10/pound of calf (Source: Standardized Performance Anal ysis, MAERC). Other sources of revenue may include SOD and sale of pregnant cows or bulls. In our model, we look mainly at the sale of cows and the lifting of SOD, both of which are management options avai lable to the user. At the end of each monthly time step, we calculate the total revenue gained by the ranch.

PAGE 46

46 The ranch also incurs a number of manageme nt and maintenance e xpenses that include feeding and grooming of the cattle. One of th e major expenses of the ranch is providing supplementary feed during the preconditioning peri od of the calves. To improve the total yearly yield, the ranch also buys impregnated cows. Anothe r source of expenditure at the ranch is the hormones injected into the cows, which is an al most regular practise (Source: Patrick Bolen). Apart from the abovementioned expenses, there are a number of other conditional expenses that may be incurred by the ranch, one of which includ es pumping water. During the years that the ground water level is low, additi onal water is pumped into the canal and may cause electricity overhead. To analyze the various combinations of conditions and their corresponding financial repercussions, the model is designed with the idea that the user of the tool, i.e., the rancher, gains an overall management and financial perspective a nd can vary the conditions to study the effects. This takes into consideration that within the model, all these options are mainly management options which the user can set while he is runnin g the model, the idea being that the financial decisions should always remain in contro l of the rancher who is using the tool. The object oriented model interpretation of incomes and expenditures has far-reaching applications, outside the fi eld of hydrological and ranch operation modeling. QnD:BIR establishes a link, on a simplistic level, between the world of economics and the object oriented programming model. The model design handles the ranchs month to month economics by tracking on a simplistic level, the incomes and expenditure of the ranc h through the cow/calf operation. The two DData values of DTotalIn come and DTotalExpens es govern these two values. And the different between them is consid ered to be the operating balance of the ranch (DTotalOperatingAmount). The incomes and expens es are usually a result of the management

PAGE 47

47 options which the PTotalExpenses and PTotal Income aggregate to update the values of DTotalIncome and DTotalExpenses. Incomes and e xpenditures can be a resu lt of more than just cow/calf operations, for example, pumping exces s water etc. can also result in the ranch incurring expenses. Hence this module of QnD:BIR is controlled at the global level, and does not belong to any local component. However, as each of the local components c ontribute to the total incomes and expenditures, the va lues are updated and the total ope rating amount is calculated at the end of every timestep. Overall, this module can be looked at as a set of processes or operations in the object oriented sense, which is affect by individual objects or component s through their respective local processes. A similar simplistic design can be a starting point for the development of other applications that need to translate financia l or economic processes into object oriented programming. Addition of New Features into the QnD Model Initial design and versions of QnD:BIR were reviewed by ranch management and scientists who requested that a significant new feature would need to be added to the user interface; moveable icons that repr esent cattle herds. The request ed features were not present in any of the previous models of QnD. Thus, an additional feature in QnD Buck Island Ranch is the capability to add icons representing cows on the spatial units representing pastures. These icons can be placed on the spatial units if the da ta suggests the presence of a cow herd on the corresponding pasture. The cow icons can be move d as one moves the cows from one pasture to another as part of a management decision. This ca pability to move cow herds is a very important part of the management cycle of the operation on the ranch. The technological challenge asso ciated with the development of such a management option and dynamic icon movement is what makes in a special feature of QnD:BIR. The dynamic

PAGE 48

48 placement of the cow icons on the pastures to indi cate the presence of a herd involves placing an additional GIS layer, a marker point layer, on t op of the existing GIS maps that are loaded at startup. The GeoTools-Lite is generally a useful API for GIS, but the documentation of it is still sparse which further complicates the task. This point layer is generates by checking each of the spatial units that are being loaded from the XML f iles for the presence of a herd or in QnD terms the presence of a CHerd component. If a CHerd is found in a spa tial unit (CSpatialUnit), a new point is added at the (x, y) location of the cen troid on the new marker layer. The procedure is replicated for all of th e spatial units and the result is the pl acement of the cow icons on the user interface GIS map of QnD:BIR. Once the icons are placed onto the GIS map, moving the cows as a management option is the next technical difficulty. Qn D supports the movement of a component from one spatial unit to another, wherein, all of th e processes and data linkages are moved along with the component to the destination spatial unit and the links ar e reestablished. In order to facilitate the management option of moving cows, Java Swing obj ects were used to create the user interface extensions and QnD component movement support was used to move the cows in the model setup. When the cows are moved, instead of havi ng to redraw the whole marker layer, QnD:BIR simply loads the marker layer and deleted th at one point corresponding to the CHerd being moved and adds a new point at the destination spatial unit. The map is then refreshed in order to reflect the changes made to the layers. This management option provides future m odel developers, the option to move their components during the course of a model run as a part of a management decision, which in turn further expands the flexibility of the model itself.

PAGE 49

49 The simplistic model design of QnD:BIR enab les it to cover varied modules, on a ranch scale. Moreover, the design also accommodates a ny future improvements to the model structure and design relationships, hence making it highly extendable. Model Calibration, Validation and Data Representation The efficient development and working of a model requires checking the accuracy of the results and increasing the robustness of the mode l. Model Testing is used to improve the performance of the model by detecting the desi gn shortcomings in the model algorithms and using procedures like calibration and validation to improve the hard iness of the model as well as its accuracy and performance. The following sect ions expand on the procedures used for model calibration and validation: Model Calibration Model calibration is the methodol ogy used to tune or update the model settings to suit the site that it is being used for. Model calibrati on process involves identifying parameters that within the model, that could allow the possibility of an error factor, namely parameters that have been used from a general national average or deri ved from other studies at different sites where the conditions might not be exactly the same as th e chosen research site. Once these parameters have been identified, their valu es are changed to suit the condi tions being modeled. Moreover, for the calibration period chosen, these paramete rs values are corrected within the allowable limits to best capture/follow the observed data trends. Model calibration is done in order to improve the model s accuracy, by adjusting the parameters to best suit the hi storical data observed. However, the more complex the model, the more parameters that can be changed, making the calibration process to be more and more complex.

PAGE 50

50 Model Validation Model calibration prepares the model to best suit the conditions at the current research site. A model is then validated by running it for the validation period without any further change in parameters or setup after it has been calibrated. The results of these runs are then compared to the observed/measures values of the parameters th at are being modeled. This is done in order to gauge the effectiveness of the model and its setup. During the valid ation period, several statistical and other methods are used to calculate and quantify th e error that is present in the model. The following sections describe some of the commonly used statistical error quantification methods. Model Evaluation Graphical representation of th e model results does quantify th e model and the general trend between the observed and the model results. Howeve r, visual evaluation of the results alone is not sufficient to gauge the effec tiveness and accuracy of the model. Statistical analysis methods are used to quantify the visual evaluation of th e result by using several methodologies to quantify the amount of error that is exhi bited by the model resu lts. The following secti on attempts to give an overview of a few commonly us ed model evaluation techniques. Statistical Representation Quantitative methodologies of analysis are requir ed to evaluate the results of a calibrated model. This section looks at some of the common ly used methodologies that are used to validate the results of this modeling effort, namely, R oot Mean Square Error (RMSE), Pearson productmoment correlation coefficient (R2), and Nash-Sutcliffe (NS) Co efficient (Nash and Sutcliffe, 1970). In all the following equations, iP is the observed value, iO is the model-simulated value, and N is the number of observations.

PAGE 51

51 The Root Mean Square Error: RMSE is essentially the overall sum of squares errors normalized to the number of observations (Hes sion et al., 1994)(Yang 2006). The RMSE is calculated in the same units as the analyzing qu antity. The following equation is used to calculate the RMSE: N i i iO P N RMSE1 2) ( 1 RMSE 0 (3-15) This value can be interpreted in term of th e units of the modeled parameter. Due to the presence of a quadratic term in the equation, a large error value has a gr eater effect and on the other end, smaller values indicate bette r model performance (Evans et al., 2003) Pearson product-moment correlation coefficient (R2), is the measure of linearity between two variables. R2 is probably the most popular measure of fit in statistical m odeling. The values of R2 can range between 0 and 1, with 1 being the perfect match of measured and predicted. The equation used to calculate the value of the coefficient (R2) is given by: 2 2 2 1 2) ( ) ( ) ( ) ( R P P O O P P O Oi i i N i i 1 02 R (3-16) The Nash-Sutcliffe (NS) coefficient (Nas h and Sutcliffe, 1970) is one of the more effective ways to indicate a goodness of fit. Th is method is also recommended by the American Society of Civil Engineers (ASCE, 1993) as an e ffective instrument of mo del validation. An NS value of 1 indicates a perfect f it and alternately, as the value approaches zero, the lesser the accuracy of prediction of the model. The NS can be computed by using the following:

PAGE 52

52 N i i N i i iO O P O NS1 2 1 2) ( ) ( 1 1 NS (3-17) The NS is most effective when the coefficient of variation for the observed data set is large (ASCE, 1993). The modified form of Ceff was developed by Krause et al. (2005) to reduce the sensitivity of Ceff to large values: Ceff m 1 Oi Pij i 1 NOi O j i 1 N with j =1 (3-18) The overall model design is simplistic but the knowledge based iterative approach strengthens the modeling effort. The strength of the model is measures through the process of testing and validating the model, which is described in the following chapter.

PAGE 53

53 CHAPTER 4 MODEL TESTING, RESULTS AND DISCUSSION This chapter focuses on the results and the testing of the model. Model results from the calibration to the validation stage are represente d using a variety of gr aphical methods and an analysis of the most and least favorable result s is performed to inde ntify the strengths and weaknesses of the model. The QnD:BIR decision support tool as discussed in Chapter 3, is developed using Java and object orientation for the south Florida beef cattle agro-ecosystem. This model uses simple relationships based on measured data, the laws of hydrology and results derived from consultation with the ranch managers. The model was developed to be applied on a whole farm for a variety of processes and events bei ng modeled, ranging from hydrology and nutrient movement, which is the major validation modules of the model, to cow/ranch management and the income-expenditure cycle. The enterprise ma nagement and the income-expenditure cycle run different scenarios to assist the decision makers to interpre t the output of the hydrology and nutrient engine and its effect on enterprise mana gement and profitability. The time-step for the version 1.0 of QnD Buck Island Ranch is set to be monthly to better suit it to the decision timescale at which the ranch is being managed. Model Inputs QnD is capable of reading time series inputs fr om a file in comma-separated format. These files have to be declared in the XML input files of the model which are read in and stored in hash tables. For QnD: Buck Island Ranch, the inputs provided to the model include average monthly rainfall and ground water table dept h, which are being read from time series files. For the GIS module, a GIS shapefile describing the modeled area is also a part of th e input, which contains

PAGE 54

54 the area and perimeter of each pasture, which is also read in as input. All of the input that is read is stored into DDriverData objects. As a part of the discussion about the inputs of the model, an overview of the input data and its implications is warranted. Figure 4-1 is a graph of the total m onthly rainfall and the ground water levels of BIR. From the graph we notice that the time period that we are applying the model is a combination of wet and dry pe riods. Early 2000 and 2001 were relatively dry periods, with low ground water table. June Se ptember 2001 is a wet period with high rainfall and as a result we notice a rise in the water table depth. This is again foll owed by a relatively dry period between October 2001 and June 2002. Hence the water table too drops slowly. The point to be noted here is that the wate r table is supported to some exte nt by the presence of the HarneyPond canal which is maintained at an almost cons tant height, and which c ontributes to the water table. But this effect of the canal is alrea dy considered by the model by taking the water table depths directly as input. The following year, late 2002 2003 is a wet year with rainfall events sparred throughout the year. This results in the water table stayi ng relatively high thr oughout this period, which presents two different scenarios. Firstly, late 2002 period, wherein, the rainfall is relatively low, but the level of the water table does not drop, in fact there is a slight increase observed during this period. Alternatively, the period of mid to late 2003 presen ts a different scenario of high water table and large amount of ra infall. Each of these scenarios could affect the accuracy of the model, as QnD:BIR directly relates rainfall and ground water depth with runoff. This implies that a high water table would result in even a slight amount of rainfall causing heavy runoff, which might lead the model to over-predict in this period.

PAGE 55

55 This overview is designed to understand the inpu t conditions that the model is being tested upon and their implications and effects on the model. Model Outputs QnD outputs data in two modes, one of them is through the graphica l user interface (GUI) which would be graphs and indicator lights main ly intended for the decision makers to assist them in managing the enterprise. These outputs can be customized to suit the requirement of the enterprise being managed. For QnD: BIR, GUI outputs are the cow and calf population and the operating-amount remaining. The other, more conventional, mode of output is the comma separated file (.csv) with the numerical values of the output parameters. For QnD: BIR, runoff volume, phosphorus load and grass biomass cons titute the major outputs required for validation of the model. Model Calibration The model was calibrated for the dry period of September 2000January 2001 and a wet period of July 2001 and August 2001, for the experi mental pastures summer pastures 1-8 and winter pastures 1-8. For the hydrological model, the parameter used for calibrating the model is the percent available storage, which combines the effect of plant av ailable storage and the gravitational storage. Within the forage model, th e rainfall effect and seasonal effect parameters are the parameters that the model was calibration on. Hydrology During calibration, the hydrology of the m odel was calibrated based on the percent available storage parameter. The calibrated value of this parameter was determined to be 0.238. This number is large enough to account for the plan t available storage and in also the effect of evapotranspiration on the rainfall, which is not ta ken into consideration by the model separately. The model overall was predicting the runoff amo unts on the higher than the measured values.

PAGE 56

56 Figure 4-2 shows a typical graph for one of the summer and one of the winter pastures during calibration. This covers the tre nd shown in more detail in appe ndix A, which contains all the results from all of the experime ntal pastures. Since part of th e calibration period is one of the driest times for the region, the model calibration should have helped it predict any future dry patches with accuracy. Nutrients Calibration period model runs for nutrients fo llowed a similar trend at the hydrology, with the measured values being less than the model pr edicted. Figure 4-3 shows a typical graph for one of the summer and one of the winter pastures during calibrati on. This covers the trend shown in more detail in appendix A, which contains all the results from all of the experimental pastures. Forage The forage model calibration was done for the period of 2000-2001. The Buck Island ranch forage data set does provide forage yield data for January, March, and December of 2000. Moreover, the forage yield for the each pastur e varies over a range for any given month. QnD:BIR was calibrated within the range of values measured at BIR. Hence, the model calibrated for the available ranges in the absence of continuous data, for the calibration period. Model Validation/Testing Model testing for remaining time series data is de scribed in this section. In order to give a general idea of the trends of the model over the sixteen pastures, the highest and lowest performance levels of the model for both the improved (summer) and native (winter) pastures. Additional results are provided in appendix A. Hydrology Considering the overall predictions of the mode l, it can be seen that the overall model tends to marginally overpredict the values of ru noff, given the scale of the values, while missing

PAGE 57

57 some of the smaller runoff events. Analysis of Su mmer 5 (Figure 4-2) Past ure indicated that the model predicted the values quite closely with a Nash-Sutcliffe co efficient (NS-Ceff) of 0.6169, a Root Mean Square Error (RMSE) of approximately 5.025 million liters, and a Normalized Mean Square Error (nMSE) of 0.3831. During th e validation period be tween January 2001 and December 2003 excluding the calibration months, th e higher peak events occurred during the months of May and September of 2001, 2002 a nd 2003, and January of 2003 of which the model predicted most months with an acceptable degr ee of accuracy except those of September 2000, October 2001, June 2002, and January 2003 where it missed some of the more significant events. However, some of the smaller runoff events occu rring during the early mont hs of the year each year were missed by the model. The cumulative Runoffs, however, are generally close to the measured values, indicating that the overall amount of runoff from the Summer 5 pasture (Figure 4-3) is being predicted with better levels of accuracy. This indicates that though the model misses a few small events, it is able to predict the total volume of runoff over a period of time. In the graph of the cumulative runoffs, the curve of the measured values follows the curve of the predicted values very closely, which, along with the high Nash-Sutcliffe co efficient value, makes Summer 5 the best performance of the model for summer pastures. On analyzing the measured vs predicted scatte r graphs (Figure 4-4), it can be seen that Summer 5 has one of the best prediction trend among all the summer pastures for runoffs. The overall model does tend to marginally under-predict the values; however, th e error value is not very high. The measured values and the pr edicted values are also highly proportional. Considering the least favorable of the mode ls performances, analysis of the Summer 4 pasture (Figure 4-5) indicated that the model did not perform as well. The graph indicates a

PAGE 58

58 greater degree of over-prediction than other pastures with a NS-Ceff value of 0.5100, an nMSE of 0.4900 and a RMSE of approximately 4.113 million liters. During the validation period between January 2001 and December 2003, the model predicted the higher peaks that occurred during the months of May and September of 2001, 2002 and 2003, and January of 2003, with a lesser degree of accuracy than the best perfor mance of the model. Moreover, the model missed predicting a few of the events that occurred during the months of September 2000, October 2001, December, January and February 2003. The predicted cumulative runoffs for this past ure (Figure 4-6) are higher than the model measured cumulative runoff values. The values be gin together at the start of the validation period, but move apart from September 2001, wh en the model missed a runoff event. Following this period, the distance between the two curves increases since the model overpredicts the peaks in 2002, except for certain points where they come marginally closer. The measured vs predicted scatter graphs (Figure 4-7) for Summe r 4 indicates that a percentage of the values are over-predicted. The NS-Ceff value is also calculated to be the minimum among the coefficient values of all the other summer pastures, wh ich indicates that the model has not proven as effective in modeling Summer 4. Though the overall values were less accurate than the models best performance, the model still managed to capture the general tren d of the peaks and the lows of the predicted values quite accurately. Moving to the native pastures, among all the model predictions for the native or winter pastures, the best performance of the model is s hown in Winter 4 (W4) (Figure 4-8), with a NSCeff of 0.809, which is considered to be quite accurate, an nMSE of 0.1908, and an RMSE of approximately 5.608 million liters. The runoff gr aph for W4 shows a close match between the

PAGE 59

59 measured values and the model predicted values giving the model a highly acceptable degree of accuracy. The graph shows accurately predicted high peaks for the months of of May and September of 2001, 2002 and 2003, and January of 2003. However, the model still misses a few of the smaller events, e.g., in the months of September 2000, May 2001, January, May and August 2003. The cumulative graph for the runoffs in W4 (Figure 4-9) gives a very clear indication of the accuracy of the model in the abovementioned pasture. The curve for the measured values follows the curve for the model predicted values very closely as seen in the graph. The values are very close almost throughout th e validation period, except for an instance in November 2002 and one in September 2003, where the measured valu es differ from the predicted values with a slightly greater margin. The measured vs predicted scatter graph for W4 (Figure 4-10) indicates clearly the accuracy of the model with respect to pastur e W4. Very few values on the graph are underpredicted, and the high proportionality also indicates the high accuracy of th e values. The best fit trend line also matches quite closely with th e 1-1 trend line, overall implying a good model performance for W4. Among the native pastures, the model shows the least accuracy of performance on the Winter 3 (W3) pasture (Figure 4-11). The re sult graphs indicate the presence of some discrepancy between the measured and the predic ted values for the month of July and October 2001, June, August, December 2002, and June, September and October 2003. For certain other months like September 2000, November 2001, Fe bruary, March, April and November 2003, some less significant runoff events are missed by the model.

PAGE 60

60 The cumulative runoff for W3 (Figure 4-12) is correspondingly refl ective of the lower accuracy of the model on this pasture. The curv es follow closely until May 2001. However, after this period the model over-predicts the runoff on three crucial occasions in September of 2001, 2002 and 2003 which causes the cumulative curves to sparse out rapidly. The measured vs predicted scatter graph fo r the W3 (Figure 4-13) pasture indicates an overprediction of some of the values. It can be seen from the graph that the best fit trend line does not perfectly match the 1-1 trend line as it does in W4, which is the best performance of the model. Phosphorus Load The analysis of the model perfor mance in the various pastures shows that the model tends to slightly overpredict the values of the peaks and misses a few smaller events, similar to the performance of the model on different pastures for runoffs. However, over NS-Ceff values overall are higher for the phosphorus load simu lations, indicating the overall better performance of the model in predicting phosphorus loads Among the model performances in the various summer pastures, th e pasture with the most accurate model predictions is Summer 8 (S 8) (Figure 4-14), with a NS-Ceff of 0.5960, an nMSE of 0.4040 and an RMSE of 5.35 kg. The gra ph depicting the S8 nutri ent load indicates the close match between the measured and the predicted values. For this pasture, the most significant events take place during the months of Ma y and September 2001, 2002 and 2003, and December 2002. Though the model mirrors the measured va lues with an acceptabl e degree of accuracy, there remain a few values that the model misse s, as shown for the months of September 2000, May 2001, May 2002, and September 2003. The cumulative load graph (Figure 4-15) also indicates a good level of accuracy in model performance. During the initial stages of the validation period, the pr edicted values closely

PAGE 61

61 mirror the measured values. However, around th e period of September 2001 and 2002, there is an increase in the discrepancy between the values This discrepancy varies until the end of the validation period. The measured vs predicted scatter graph (Figure 4-16) for S8 shows a good prediction performance. Though a few of the values are under-pre dicted in the graph, th e best fit trend line matches the 1-1 trend line tolerably well and is al so indicative of the accu racy of the model in this regard. Considering the pastures for which the performance of the model was less than completely satisfactory, the pasture where there appears to be maximum discrepancy between predicted and measured value is Summer 3 (S3) (Figure 4-17) with a low value of NS-Ceff (0.1484), with 0.8516 and 5.205 Kgs respective valu es of the nMSE and RMSE calculated. In this case, the model matched the high peak values for the months of May and September 2001, 2002 and 2003, and December 2002, though without the accuracy displayed by the model in other pastures. The model also missed the va lues in may 2001, 2002 and September 2003. The overall model however, catches th e general trends of peaks and lows similar to the hydrology module. The graph for the cumulative nutrient load (Figure 4-18) indicates the lower level of accuracy in the performance of the model for this particular pasture. Th e match or relationship between the parameters remains close until May-September 2002, after which, the model overpredicts a few values causing a gap betw een the measured and predicted curves. The measured vs predicted scatter graph for S3 (Figure 4-19) indicates some underprediction. A number of the data values are overpredicted, but the best fit trend line is below the 1-1 trend line and is not very accurate.

PAGE 62

62 Overall, for the summer pastures, the model cap tures most of the significant events and the general trends of peaks and lows This is noticed across all of the eight summer pastures. The detailed graphs of these are included in appendix A. On comparing the model performance for th e various native pastures with regards to nutrient loads, the pasture with the most accurate model performance is Winter 4 (W4) (Figure 420), with a NS-Ceff value of 0.6946, an nM SE value of 0.3054 and an RMSE value of approximately 0.8899 Kgs. In the graphs associated with this pasture, the values predicted values closely match the data measured at BIR. The model predicts values with very marginal discrepancy for the high load months of July and October of 2001 and 2002. In spite of missing a few of the smaller events, the model still mainta ins a highly acceptable level of accuracy and W4 is therefore among the best performances of the model in native pa stures for nutrients. The cumulative load graph for W4 (Figure 4-21) shows a close relationship between the predicted values and the measured values. In the middle of 2002, th ere is a slight discrepancy in the values, which can also be seen in some of the other months, namely, the end of 2001, 2003 and the beginning of 2002. However, overall, the cu rve for the predicted values closely matches the curve for measured values, thus indicat ing a good degree of accuracy in the model. The measured vs predicted scatter graph fo r W4 (Figure 4-22) indi cates a slight under prediction with regards to the data points on the gr aph. However, the best fit trend line indicates a slight under-fitting with resp ect to the 1-1 trend line. Among the model performances for all of th e native pastures, the pasture where the model was the least effective is Winter 8 (Fig ure 4-23) with a NS-Ceff value of -0.1755, an nMSE value of 1.1755 and an RMSE of approxima tely 1.0439 Kgs. According to the data shown in the graphs, the model closely predicts the values for the months of May 2001, May 2002, May

PAGE 63

63 and January 2003. The model ove rpredicts the values for September 2001, 2002 and 2003 and December 2002. However, it still manages to main tain an acceptable degr ee of accuracy as can be seen from the graphs. The cumulative graph of the nutrient loads fo r W8 (Figure 4-24) i ndicates a discrepancy between the predicted and the measured values. The curve for the predicted values follows the curve for the measured values for the initial fe w months of the valida tion period. From August 2001, the curves start drifting apart as the discrepa ncy in the values increases. The measured vs predicted scatter graph for W8 (F igure 4-25) indicates a higher de gree of overprediction than is seen on the graphs for the other native pastures. The best-fit line seems to be above the 1-1 trend line. Forage Growth Model testing was done for the period of 2001 2003. As mentioned earlier, the forage data recorded at BIR is not continuous and each of the readings for every month has range of values, and hence, the possibility of erroneous values. The model was tested only for the improved controlled pastures (Summer 1 and Summer 8) (Swain et al, 2007), which were maintained at a zero cattle stocking rate. Mo reover, the cattle moveme nt is a part of a management decision in QnD:BIR and for the testing period of the model, no management options were used during the simulation to effi ciently gauge the accuracy of the model without any external influences. The te sting procedure however is not as comprehensive as it was for hydrology and nutrients due to the discontinuous and varied nature of the measured data. The overall performance of the model was ga uged by comparing the si mulated results with the range of measured values for the chosen pa stures. Analysis of Summ er 1 pasture (Figure 426) indicated that the model wa s acceptably accurate and within the measured data range for most of the testing period. It can be noticed th at the model over-predicted the forage yield for a

PAGE 64

64 few months and missed one increas e in yield growth over the testing period in September 2003. The NSCeff was calculated to be 0.2440 which is acceptable but does not reflect the variable nature of the measured data. Analysis of Summer 8 pastur e (Figure 4-27), revealed a si milar trend with the models overall performance was within acceptable limit s. The NS-Ceff value was calculated as 0.2930. An important observation of the models design indicated that th e simplistic empirical approach followed by the model, is unable to account for th e different factors that influence the forage growth. However, the model overa ll was effective in simulating the growth trends of Bahia grass. Summary of Model Testing The model was tested against the BIR meas ured data for the period of 2001 2003. The graphs obtained are included in the appendix A. Moreover, several error estimation methods were used to gauge the performance of the mode l over the 16 experimental pastures which were explained in detail in chapter 3. After analyz ing the output of all of the methods and graphs, though the model seems to perform good and bad ove r all of the pastures, largely, the model captures the trends of the peaks and lows, occasionally missing/over-predicting a few peaks/lows. The RMSE values do not vary larg ely over the pastures, and the cumulative and monthly graphs indicate that th e model performance is within th e acceptable lim its of accuracy. Collectively, the model performed well, with er ror estimations within the acceptable limits of modeling standards. Enterprise Wide Simulations and Scenario Analysis The model development and testing have been di scussed in the previous sections, which is a requirement to validate the authenticity of any model. This section is the actual application of the model for the whole beef cattle enterprise an d the 68 pastures in the farm. The model was run

PAGE 65

65 for the period of 2000 2003 for the whole farm and the total runoff, phosphorus load, grass biomass and the monthly operational expenses of the model were measured under different simulated rainfall conditions apart from the measured rainfall which is provided as input. The model was run for the whole farm simu lation following the time schedule discussed earlier in chapter 3 with the move cows options being enforced every summer and winter to move the cows from summer to winter pastures and vice-versa. Moreover, apart from the actual rainfall, the conditions of less than regular rainfall and more than regular rainfall were also run as separate scenarios. Scenario 1: Measured Rainfall The first scenario is the regular measured c onditions of rainfall and temperature. The cows are alternated between the Summer 1-8 pastures during the summer and th e Winter 1-8 pastures during the winter for the period of the run. The out put values that are taken up for this analysis included total monthly runoff, total phosphorus lo ad, total grass bioma ss and total operating amount available with the ranchers per month, with the assumption that BIR started out with $1000000 as capital when the simulation period starte d. The operating income is not only an indicator of the ranch operating costs and incomes, but also serves as a pointer to the cow-calf ranch management operations, as they are the major influence in deciding the profitability of the ranch. Scenario 2: Low Rainfall The second simulation involved reducing the am ount of rainfall over the period of the run by calibrating the measured values of the rainfall input to 15 percent less than that of a regular period, hence causing more severe drought conditi ons. All other input para meters were kept the same as rainfall is the major driver of QnD:BIR, it was chosen as the parameter to run different

PAGE 66

66 scenarios for. Also the cow movement was diffe red and other non-experime ntal pastures were used to move the cows during the winter namely, South marsh west and South marsh center. Scenario 3: High Rainfall The other end of the spectrum to lower rainfall is the higher rainfall year. The idea is to see if more rainfall necessarily has an effect on the profitability of the ente rprise and on the general hydrology and nutrient loads of the region. Howeve r the cow movement in this case too is similar to the other scenarios, wherein, the co ws are moved from summer or improved pastures to the unimproved pastures in the winter. All of these scenario runs were run to provi de an understanding of how the model can be useful both to researchers and the rancher to pr ovide a starter for future predictions based on climate prediction data which can be continued as a part of this research. Results and Discussion The results of the whole farm simulation runs provided results which follow similar trends to those seen in the experimental pastures during the validation period, as can be seen from the graphs in Figure 4-28. The operating amount values are interesting due to the sinusoidal nature, which further implies that profitability over the longer term required more than just cow movement and calf sales, without provided ex cess support for the calves/cows to grow and flourish, Though some months make good profits, overa ll, in order to maintain sustainability, the conditions do need to be favorab le or the operating costs setup within this model, must be reduced.

PAGE 67

67 Table 4-1. List of values of Nash-Sutcliffe coefficient (Ceff) Normalized Mean Square Error and Root Mean Square Error (in million liters) for runoff in summer pastures. Summer1 Summer2 Summer3Summer4 Summer5Summer6 Summer7 Summer8 NSCeff 0.583 0.587 0.613 0.510 0.617 0.561 0.536 0.572 nMSE 0.417 0.413 0.387 0.490 0.383 0.439 0.464 0.428 RMSE 4.179 4.368 4.346 4. 113 5.025 5.417 5.164 4.809 Ceff_m 0.57 0.5940 0.6210 0. 5910 0.6260 0.5820 0.5380 0.5780 Table 4-2. List of values of Nash-Sutcliffe coefficient (Ceff) Normalized Mean Square Error and Root Mean Square Error (in million liters) for runoff in winter pastures. Winter1 Winter2 Wint er3 Winter4 Winter5 Wi nter6 Winter7 Winter8 NSCeff 0.729 0.674 0.382 0.809 0.695 0.738 0.715 0.705 nMSE 0.271 0.326 0.618 0.191 0.305 0.262 0.285 0.295 RMSE 5.902 7.231 6.977 5. 608 8.806 6.687 8.215 7.151 Ceff_m 0.656 0.604 0.491 0.663 0.616 0.633 0.649 0.618

PAGE 68

68 Table 4-3. List of values of Nash-Sutcliffe coefficient (Ceff) Normalized Mean Square Error and Root Mean Square Error (in Kgs) for load in summer pastures. Summer1 Summer2 Summer3Summer4 Summer5Summer6 Summer7 Summer8 Ceff 0.539 0.436 0.148 0.464 0.368 0.337 0.444 0.596 nMSE 0.460 0.564 0.851 0.535 0.631 0.662 0.555 0.404 RMSE 5.680 4.623 5.205 5. 233 6.479 5.537 6.940 5.359 Ceff_m 0.563 0.518 0.403 0.524 0.513 0.471 0.567 0.587 Table 4-4. List of values of Nash-Sutcliffe coefficient (Ceff) Normalized Mean Square Error and Root Mean Square Error (in Kg s) for load in winter pastures. Winter1 Winter2 Wint er3 Winter4 Winter5 Wi nter6 Winter7 Winter8 Ceff 0.411 0.534 0.504 0.694 0.287 0.586 0.461 -0.175 Nmse 0.588 0.465 0.495 0. 305 0.712 0.413 0.538 1.175 RMSE 1.321 1.436 1.511 0. 889 3.106 1.552 2.239 1.0439 Ceff_m 0.383 0.511 0.436 0.567 0.476 0.572 0.529 0.303

PAGE 69

69 Figure 4-2.Monthly runoff in the Summer5 pasture Figure 4-3.Cumulative runoff in the Summer5 pasture Figure 4-4. A measured vs predicted scat tergraph for the Summer5 pasture runoffs Ceff = 0.617

PAGE 70

70 Figure 4-5.Monthly runoff in the Summer4 pasture Figure 4-6. Cumulative runo ff in the Summer4 pasture Figure 4-7. A measured vs predicted scat tergraph for the Summer4 pasture runoffs Ceff = 0.510

PAGE 71

71 Figure 4-8.Monthly runoff in the Winter4 pasture Figure 4-9.Cumulative runoff in the Winter4 pasture Figure 4-10. A measured vs predicted scat tergraph for the Winter4 pasture runoffs Ceff = 0.809

PAGE 72

72 Figure 4-11.Monthly runoff in the Winter3 pasture Figure 4-12.Cumulative runoff in the Winter3 pasture Figure 4-13. A measured vs predicted scat tergraph for the Winter3 pasture runoffs Ceff = 0.382

PAGE 73

73 Figure 4-14.Monthly phosphorus lo ad in the Summer8 pasture Figure 4-15.Cumulative phosphorus lo ad in the Summer8 pasture Figure 4-16. A measured vs predicted scatte rgraph for the Summer8 pasture Ph loads Ceff = 0.596

PAGE 74

74 Figure 4-17.Monthly phosphorus lo ad in the Summer3 pasture Figure 4-18.Cumulative phosphorus lo ad in the Summer3 pasture Figure 4-19. A measured vs predicted scatte rgraph for the Summer3 pasture Ph loads Ceff = 0.148

PAGE 75

75 Figure 4-20.Monthly phosphorus lo ad in the Winter4 pasture Figure 4-21.Cumulative phosphorus lo ad in the Winter4 pasture Figure 4-22. A measured vs predicted scatte rgraph for the Winter4 pasture Ph loads Ceff = 0.694

PAGE 76

76 Figure 4-23.Monthly phosphorus lo ad in the Winter8 pasture Figure 4-24.Cumulative phosphorus lo ad in the Winter8 pasture Figure 4-25. A measured vs predicted scatte rgraph for the Winter8 pasture Ph loads Ceff = -0.175

PAGE 77

77 Figure 4-26.Monthly forage yiel d for the Summer 1 pasture Figure 4-27.Monthly forage yiel d for the Summer 8 pasture

PAGE 78

78

PAGE 79

79 Figure 4-28. A) Total Forage Yield in the ranch B) Total monthly runoff in the ranch. C) Total phosphorus load in the ranch. D) To tal operating amount at the ranch.

PAGE 80

80 CHAPTER 5 CONCLUSION AND FUTURE WORK Conclusions Beef cattle enterprises and their manageme nt face several complex management and political challenges in an already fragile ecosyste m of south Florida. On one end, there exists a struggle to maintain profitability and sustainabil ity, and on the other, the effort to conserve the fragile ecosystem of the region and hence the political pressure from the environmental protection agencies. The current need is for a sy stem that models the ecological issues of non point source pollution, and interp rets the results in a user friendly decision oriented format. The objective of this research was to design and develop one such decision support tool with the capability to model the ecological pro cesses and interpret them as well. QnD: BIR was intended as a model that can be used by bot h the research community and the ranchers themselves with the intention of assisting th e decision making process for managing the ranch. Moreover, the research site chosen, Buck Is land Ranch (BIR), provides a unique setting for production-related, agro-ecological research. MAERC/BIR combines a research facility with a commercial-scale, beef cattle enterprise (10, 300 acres) to explore the role of long-term ecological and social dynamics with in sub-tropical grazing systems. QnD: BIR is based on literature knowledge, actual laws of ecology, previous modeling efforts and expert wisdom from researchers and ranchers, the intended users of the model. The unique iterative model development methodology of QnD allows very close participation with the researchers and ranchers duri ng the whole process of model de velopment to cater it to their requirement. Moreover, the model is developed as an enterprise wide model, to be used on all of the pastures within MAERC/BIR. The scale is yet another aspect that makes QnD:BIR unique.

PAGE 81

81 Once the development stage was completed, QnD:BIR was tested on environmental data from BIR for the period of 2000 2003 for sixteen experiment al pastures including both improved and native pastures. Specifically, Qn D:BIR simulation results of monthly runoff, phosphorus load and forage yield were compared with comparable field-scale data. After analyzing the output of all of th e statistical error estimation me thods and graphs, largely, the model captures the trends of the peaks and lows, occasionally missing/over-predicting a few peaks/lows. The hydrology and nutrients simula tions generally follow the trend mentioned above. The forage yield simulations are also to th e most part accurate and within the range of the measured values. Moreover, the statistics of error estimation also indicate that the models performance is within the accepta ble limits, given the coarse mont hly time step. The model also has added modules to simulate ranch cow/cal f production and the incomes and expenditures management which result in making it a comple te enterprise wide decision support tool. Future Research Recommendations This modeling effort provided generally accep table results and a helpful decision support tool for researchers studying sust ainable ranching and ranchers to assist them in managing the ranch operations. However, as in any modeling effort, there is scope for future advancements within the model. QnD:BIR design methodology allo ws such advancements to be integrated easily into the model. The following are some recommendations for future work in this area. Integrate Future Climate Prediction s and Analyzing Different Scenarios Climate prediction data from the Southeast C limate Consortium (SECC) can be integrated into the model to simulate future ranch operation and ecological trends. The climate predictions of future rainfall and temperature can be used as external input to the model. Various management scenarios can be run on the model. Th e analysis of the output would assist the ranch managers to assess the future of the ranc h operation and help them prepare for it.

PAGE 82

82 Improvement of the Cattle Production Module The model currently has basic relationshi ps governing cow production and movement. More research could be done into various fact ors influencing the cow/ calf production operation ranging from buying of pregnant Heifers to cull ing of older cows. Moreover, detailed cow movement records are being collected and analyz ed recently at MAERC/BIR. This can be used to further improve the cow herd movement simulations under various scenarios. Integration of a More Complex Model into QnD The hydrology and nutrient systems of Qn D:BIR are developed using a simplistic approach. In order to more accurately model the hydrology of this region, more complex models like ACRU 2000 or the Century mode l can be integrated into QnD such that QnD:BIR uses the complex structure of these models and their outpu ts to better analyze th eir effect on the ranch operations of MAERC/BIR. Integration of a More Advanced GIS App lication Programmable Interface for Java During the course of development and testing of this model, it can be noted that the GIS interface that is used in QnD:BIR is an older ve rsion and does not allow the developer to use the complete range of GIS capabilities. GIS is a po werful tool in spatial analysis, and further exploration of some of the advan ced features of GIS can empower the model to be more spatially aware. This feature would help the model bett er gauge the topology of the region and hence better predict the hydrology of th e area. Moreover, a more advanced GIS link would also enable the use of more than one principal layer to select and run the mode l simulations and hence expand the modeling capacity of the model. Overall, QnD:BIR modeling effort is a synt hesis of both scientific data and expert knowledge to create an easy to use decision supp ort system with applications in education and future research in modeling.

PAGE 83

83 APPENDIX A MODEL RESULTS AND GRAPHS Figure A-1. A) Monthly runoff in the Summer1 pasture. B) Cumulative runoff in the Summer1 pasture. C) A measured vs predicted scattergraph for the Summer1 pasture runoffs

PAGE 84

84 Figure A-2. A) Monthly runoff in the Summer2 pasture. B) Cumulative runoff in the Summer2 pasture. C) A measured vs predicted scattergraph for the Summer2 pasture runoffs

PAGE 85

85 Figure A-3. A) Monthly runoff in the Summer3 pasture. B) Cumulative runoff in the Summer3 pasture. C) A measured vs predicted scattergraph for the Summer3 pasture runoffs

PAGE 86

86 Figure A-4. A) Monthly runoff in the Summer4 pasture. B) Cumulative runoff in the Summer4 pasture. C) A measured vs predicted scattergraph for the Summer4 pasture runoffs

PAGE 87

87 Figure A-5. A) Monthly runoff in the Summer5 pasture. B) Cumulative runoff in the Summer5 pasture. C) A measured vs predicted scattergraph for the Summer5 pasture runoffs

PAGE 88

88 Figure A-6. A) Monthly runoff in the Summer6 pasture. B) Cumulative runoff in the Summer6 pasture. C) A measured vs predicted scattergraph for the Summer6 pasture runoffs

PAGE 89

89 Figure A-7. A) Monthly runoff in the Summer7 pasture. B) Cumulative runoff in the Summer7 pasture. C) A measured vs predicted scattergraph for the Summer7 pasture runoffs

PAGE 90

90 Figure A-8. A) Monthly runoff in the Summer8 pasture. B) Cumulative runoff in the Summer8 pasture. C) A measured vs predicted scattergraph for the Summer8 pasture runoffs

PAGE 91

91 Figure A-9. A) Monthly runoff in the Winter1 pasture. B) Cumulative runoff in the Winter1 pasture. C) A measured vs predicted scattergraph for the Winter1 pasture runoffs

PAGE 92

92 Figure A-10. A) Monthly runoff in the Winter2 pasture. B) Cu mulative runoff in the Winter2 pasture. C) A measured vs predicted scattergraph for th e Winter2 pasture runoffs

PAGE 93

93 Figure A-11. A) Monthly runoff in the Winter3 pasture. B) Cu mulative runoff in the Winter3 pasture. C) A measured vs predicted scattergraph for th e Winter3 pasture runoffs

PAGE 94

94 Figure A-12. A) Monthly runoff in the Winter4 pasture. B) Cu mulative runoff in the Winter4 pasture. C) A measured vs predicted scattergraph for th e Winter4 pasture runoffs

PAGE 95

95 Figure A-13. A) Monthly runoff in the Winter5 pasture. B) Cu mulative runoff in the Winter5 pasture. C) A measured vs predicted scattergraph for th e Winter5 pasture runoffs

PAGE 96

96 Figure A-14. A) Monthly runoff in the Winter6 pasture. B) Cu mulative runoff in the Winter6 pasture. C) A measured vs predicted scattergraph for th e Winter6 pasture runoffs

PAGE 97

97 Figure A-15. A) Monthly runoff in the Winter7 pasture. B) Cu mulative runoff in the Winter7 pasture. C) A measured vs predicted scattergraph for th e Winter7 pasture runoffs

PAGE 98

98 Figure A-16. A) Monthly runoff in the Winter8 pasture. B) Cu mulative runoff in the Winter8 pasture. C) A measured vs predicted scattergraph for th e Winter8 pasture runoffs

PAGE 99

99 Figure A-17. A) Monthly phosphorus load in th e Summer1 pasture. B) Cumulative phosphorus load in the Summer1 pasture. C) A meas ured vs predicted scattergraph for the Summer1 pasture Ph loads

PAGE 100

100 Figure A-18. A) Monthly phosphorus load in th e Summer2 pasture. B) Cumulative phosphorus load in the Summer2 pasture. C) A meas ured vs predicted scattergraph for the Summer2 pasture Ph loads

PAGE 101

101 Figure A-19. A) Monthly phosphorus load in th e Summer3 pasture. B) Cumulative phosphorus load in the Summer3 pasture. C) A meas ured vs predicted scattergraph for the Summer3 pasture Ph loads

PAGE 102

102 Figure A-20. A) Monthly phosphorus load in th e Summer4 pasture. B) Cumulative phosphorus load in the Summer4 pasture. C) A meas ured vs predicted scattergraph for the Summer4 pasture Ph loads

PAGE 103

103 Figure A-21. A) Monthly phosphorus load in th e Summer5 pasture. B) Cumulative phosphorus load in the Summer5 pasture. C) A meas ured vs predicted scattergraph for the Summer5 pasture Ph loads

PAGE 104

104 Figure A-22. A) Monthly phosphorus load in th e Summer6 pasture. B) Cumulative phosphorus load in the Summer6 pasture. C) A meas ured vs predicted scattergraph for the Summer6 pasture Ph loads

PAGE 105

105 Figure A-23. A) Monthly phosphorus load in th e Summer7 pasture. B) Cumulative phosphorus load in the Summer7 pasture. C) A meas ured vs predicted scattergraph for the Summer7 pasture Ph loads

PAGE 106

106 Figure A-24. A) Monthly phosphorus load in th e Summer8 pasture. B) Cumulative phosphorus load in the Summer8 pasture. C) A meas ured vs predicted scattergraph for the Summer8 pasture Ph loads

PAGE 107

107 Figure A-25. A) Monthly phosphorus load in th e Winter1 pasture. B) Cumulative phosphorus load in the Winter1 pasture. C) A measured vs predicted scattergraph for the Winter1 pasture Ph loads

PAGE 108

108 Figure A-26. A) Monthly phosphorus load in th e Winter2 pasture. B) Cumulative phosphorus load in the Winter2 pasture. C) A measured vs predicted scattergraph for the Winter2 pasture Ph loads

PAGE 109

109 Figure A-27. A) Monthly phosphorus load in th e Winter3 pasture. B) Cumulative phosphorus load in the Winter3 pasture. C) A measured vs predicted scattergraph for the Winter3 pasture Ph loads

PAGE 110

110 Figure A-28. A) Monthly phosphorus load in th e Winter4 pasture. B) Cumulative phosphorus load in the Winter4 pasture. C) A measured vs predicted scattergraph for the Winter4 pasture Ph loads

PAGE 111

111 Figure A-29. A) Monthly phosphorus load in th e Winter5 pasture. B) Cumulative phosphorus load in the Winter5 pasture. C) A measured vs predicted scattergraph for the Winter5 pasture Ph loads

PAGE 112

112 Figure A-30. A) Monthly phosphorus load in th e Winter6 pasture. B) Cumulative phosphorus load in the Winter6 pasture. C)A measured vs predicted scattergraph for the Winter6 pasture Ph loads

PAGE 113

113 Figure A-31. A) Monthly phosphorus load in th e Winter7 pasture. B) Cumulative phosphorus load in the Winter7 pasture. C)A measured vs predicted scattergraph for the Winter7 pasture Ph loads

PAGE 114

114 Figure A-32. A) Monthly phosphorus load in th e Winter8 pasture. B) Cumulative phosphorus load in the Winter8 pasture. C)A measured vs predicted scattergraph for the Winter8 pasture Ph loads

PAGE 115

115 LIST OF REFERENCES American Society of Civil E ngineers (ASCE) Task Committee on Definition of Criteria for Evaluation of Watershed Models of the Wate rshed Management Committee. 1993. Criteria for evaluation of watershed models. Journal of Irrigation and Drainage Engineering 119(3): 429. Arnold JG, Allen PM, and Bernhardt G. 1993. A comprehensive surfacegroundwater flow model. Journal of Hydrology 142:47-69. Arnold JG and Fohrer N. 2005. SWAT2000: current capabilities and res earch opportunities in applied watershed modeling. H ydrological Processes 19:564-572. Arthington, JD, Roka, FM, Mullahey, JJ, Colema n, SW, Lollis, LO and Muchovej, RM (2005 in review). Integrating Ranch Forage Producti on, Cattle Performance and Economics in Ranch Management Systems. J. Range Manage. Arthington, JD, FM Roka, JJ Mullahey, SW Coleman, LO Lollis, and RM Muchovej. 2006. Integrating ranch forage production, cat tle performance and economics in ranch management systems. Rangeland Ecology and Mgmt. 60(1). Beasley DB and Huggins LF. 1980. ANSWERS (Area Non-point Source Watershed Environment Response Simulation), Users ma nual. Purdue University, West Lafayette. Beven KJ and Kirkby MJ. 1979. A physically-based variable contributi ng area model of basin hydrology. Hydrology Scien ce Bulletin 24(1):43-69. Booch, G. 1994. Object Orient ed Analysis and Design with A pplications. Benjamin/ Cummings, Redwood City, CA, Byard, C. Object-oriented technology a must for complex systems 1990. Computer Technology Review 10, 14, 15. Campbell KL, Kiker GA, and Clark DJ. 2001. De velopment and testing of a nitrogen and phosphorus process model for Southern Afri can water quality issues. 2001 ASAE Annual International meeting. Paper No 012085, St. Joseph, MI.: ASAE. Campbell, KL, Capece, JC and Tremwel, TK 1995. Surface/Subsurface Hydrology and Phosphorus Transport in the Kissimmee River Ba sin, Florida. Ecological Engineering Vol 5(2): 301-330 Capece, JC, Campbell, KL, Bohlen, PJ, Graetz, DA and Portier, K. (2007). Water Quality Impacts of Beef Cattle Ranches in the La ke Okeechobee Basin. Rangeland Ecology and Management. Capece, CJ, KL Campbell, PJ Bohlen, DA Graetz, and KM Portier. 2006. Soil Phosphorus, Cattle Stocking Rates and Water Quality in Subtropical Pastures in Florida. Rangeland Ecol. and Mgmt. 60(1).

PAGE 116

116 Chen DX, and Coughenour MB. 1994. GEMT a gene ral model for energy and mass transfer of land surfaces and its application at the Fife s ites. Agricultural and Forest Meteorology 68: 145. Clark DJ, Kiker GA, and Schulze RE. September 2001. Object-oriented restructuring of the ACRU agrohydrological modeling system, Tenth South African National Hydrology Symposium 26-28. Crawford NH and Linsley RS. 1966. Digital Simu lation in Hydrology: The Stanford Watershed Model IV. Technical Report no. 39, Department of Civil Engineering, Stanford University, Palo Alto, CA. Dershem, HL and Jipping, MJ 1995. Program ming Languages: Structures and Models. PWS Publishing Company, Boston, MA. Dunn, SM, Mackay, R., Adams, R., Oglethorpe DR, 1996. The hydrological component of the NELUP decision-support system: An apprai sal. Journal of Hydrology 177, 213-235. Evans BM, Sheeder SA, and Lehning DW. 2003. A spatial technique for estimating streambank erosion based on watershed characteristics. Journal of Spatial Hydrology, Vol.3, No.1. FDEP (Florida Department of Environmental Protection). 2006. Integrated Water Quality Assessment for Florida: 2006 305(b) Report and 303(d) List Update. Florida Department of Environmental Protection. Division of Water Resource Management, Bureau of Watershed Management, Tallahassee, Florida. Greene, RG, Cruise, JF, 1995. Urban watershed modeling using geographic information system. Journal of Water Resources Planni ng and Management 121(4), 318-325. Henderson-Sellers, B. 1992. A Book of Object -Oriented Knowledge. PrenticeHall, Englewood Cliffs, NJ. Hydrologic Engineering Center. 1981. HEC-1, Flood Hydrograph PackageUsers Manual. US Army Corps of Engineers: Davis, CA. Innis GS. 1978. Grassland simulation model. Ecological Studies 26. New York: SpringerVerlag. Ito, K., Xu, X., Jinno, K., Kojiri, T., Kawamu ra, A., 2001. Decision support system for surface water planning in river basins. Journal of Water Resources Planning and Management 127(4), 272-277. Jamieson, DG, Fedra, K., 1996. The WaterWa re decision-support system for river-basin planning. 1. Conceptual design. Journal of Hydrology 177, 163-175. Jones JW, Hoogenboom G, Porter CH, Boote KJ Batchelor WD, Hunt LA, Wilkens PW, Singh U, Gijsman AJ, and Ritchie JT. 2003. The DSSAT cropping system model. Europ. J. Agronomy 18: 235-265.

PAGE 117

117 Kiker GA and Clark DJ. 2001. The development of a Java-based, object-oriented modeling system for simulation of Southern African hydrology. 2001 ASAE Annual International Meeting. ASAE Paper no. 012030. St. Joseph, MI.:ASAE. Kiker, GA, Rivers-Moore, NA, Kiker, MK and Li nkov, I. 2006. QnD: A modeling game system for integrating environmental processes and pr actical management decisions. (Chapter in Morel, B. Linkov, I., (Eds) Environmental S ecurity and Environmental Management: The Role of Risk Assessment. Spri nger, Netherlands. Pp:151-185. Kiker, GA and Linkov, I. 2006. The QnD Model/Game System: Integrating Questions and Decisions for Multiple Stressors. (Chapter in Arapis, G., Gonc harova, N. and Baveye, P. (Eds) Ecotoxicology, Ecological Risk Asse ssment and Multiple Stressors Springer, Netherlands. Pp:203-225. Knisel WG (ed.). 1980. CREAMS: A field scale model for chemical s, runoff, and erosion from agricultural management systems. US DA Conservation Research Report 26: 643. Krysanova V, Mller-Wohlfeil DI, Becker A. 1998. Development and test of a spatially distributed hydrological/wat er quality model for mesoscale watersheds. Ecological Modelling 106: 261. Krysanova V, Hattermann F, and Wechsung F. 2005. Development of the ecohydrological model SWIM for regional impact studies and vulnera bility assessment. Hydrologic Processes 19: 763-783. Kunkle, WE, J. Fletcher, and D. Mayo. 2002. Florida cow-calf management, 2nd edition Feeding the cow herd. University of Florida/ IFAS Extension Electronic Data Information Service. Ledgard, H. 1996. The Little Book of Object-O riented Programming. Pr enticeHall, Upper Saddle River, NJ. Leith H. 1975a. Modeling the primary productivity of the world. In: Leith H, Whittalcher, RH. (Eds); Primary Productivity of the Biosphere. Springer-Verlag, New York. Leonard RA, Knisel WG, and Still DA. 1987. GLEAMS: Groundwater loading effects of agricultural management systems. Trans actions of the ASAE 30 (5): 1404-1418. Martinez, CJ 2006. Object oriented hydrologic and water quality model for high-water-table environments. PhD diss. Gainesville, F.L..: University of Florida, Department of Agricultural and Biological Engineering. McDonald MG and Harbaugh AW. 1988. A Modul ar Three-dimensional Finitedifference Ground-water Flow Model. US Geological Su rvey, Techniques of Water Resources Investigation Book 6, Chapter A1; 586 pp.

PAGE 118

118 Muoz-Carpena, R. A. Ritter and Y.C. Li. 2005. Dynamic factor analysis of groundwater quality trends in an agricultural area adjacen t to Everglades National Park. Journal of Contaminant Hydrology Ogden, FL, and A. Heilig, 2001, Two-dimensi onal watershed scale er osion modeling with CASC2D, in Landscape Erosion and Evolution Modeling, R. Harmon, and W.W. Doe III, eds., Kluwer Academic Press, New York, Pandey, V. 2007 Analysis And Modeling Of Cattle Distribution In Complex Agro-Ecosystems Of South Florida. PhD diss. Gainesville, F.L. : University of Flor ida, Department of Agricultural and Biological Engineering. Parton WJ, Schimel DS, Cole CV, and Ojima DS 1987. Analysis of factors controlling soil organic matter levels in Great Plains Gra sslands. Soil Society of America Journal 51: 1173. Refsgaard JC and Storm B. 1995. MIKE SHE. In Computer Models of Watersheds Hydrology, Singh V (ed.). Water Resources Publi cation: Highlands Ranch. CO; 809-846. Reitsma, RF, 1996. Structure and support of wa ter-resource management and decision-making. Journal of Hydrology 177(3-4), 253-268. Robson, D. Object-oriented softwa re systems 1981. Byte 6, 8, 74. Rosson, M. and Alpert, SR The cognitive conse quences of object-oriented design 1990. Human Computer Interaction 5, 4, 345. Rockwood DM, Davis ED, and Anderson JA. 1 972. User Manual for COSSARR Model. US Army Engineering Division, No rth Pacific: Portland, OR. Sample, DJ, Heaney, JP, Wright, LT, Koustas, R., 2001. Geographic information systems, decision support systems, and urban strom-wa ter management. Journal of Water Resources Planning and Manageme nt 127(3), 155-161. Satti S 2002, Gwrapps: A Gis-Based Decision Suppor t System For Agricultural Water Resources Management, MS thesis Gainesville, F.L.: University of Florida, Department of Agricultural and Biological Engineering Seligman NG and Van Keulen H. 1981. PAPR AN: A simulation model of annual pasture production limited by rainfall and nitrogen. p. 19 2-220. In M.J. Frissel and J.A. van Veen (ed.) Simulation of nitrogen behaviour of soil-plant systems. Pudoc. Wageningen, the Netherlands. Sugawara M, Ozaki E, Wantanabe I, and Katsuy ama Y. 1976. Tank model and its application to Bird Creek, Wollombi Brook, Bihin River, Sana ga River, and Nam Mune. Research Note 11, National Center for Disast er Prevention, Tokyo, Japan.

PAGE 119

119 Teague, W.R., Ansley, RJ, Pinchak, W.E. AND McGrann, J. 1995. A research-rancher partnership to achieve sustai nable use of rangeland. Texa s TechResearch Highlights Swain, H., PJ Bohlen, KL Campbell, LO Lollis and AD Steinman. 2006. Integrated Ecological and Economic Analysis of Ranch Manageme nt Systems. Rangeland Ecology Mgmt. 60(1). Tanner, GW and McSorley, R. (2007). Bioi ndicators in ranch management systems. Rangeland Ecology and Management. Thornley JHM and Cannell MGR. 1997. Temperate grassland response to climate change: an analysis using the Hurley Pasture M odel. Annals of Botany 80: 205221. Verberne ELJ. 1992. Simulation of the nitrogen a nd water balance in a system of grassland and soil. Nota 258. DLO-Institut voor Bodemvruchtbaarheid, Oosterweg 92, 9750 RA, Haren, Netherlands. White JR. 1987. ERHYM-II: model description a nd user guide for the BASIC version. US Department of Agriculture, Agricultural Rese arch Service, ARS-59, Washington, D. C. Williams JR and Hann RW. 1983. HYMO: Pr oblem-oriented language for hydrologic modelingUsers Manual. USDA: ARS-S-9. Wirfs-Brock, RJ and Johnson, R.E. 1990. Surveyi ng current research in object-oriented design. Commun. ACM 33, 9, 104. Yang, L. 2006. Coupled simulation modeling of flatwoods hydrology, nutr ients, and vegetation dynamics. PhD diss. Gainesville, F.L.: Univers ity of Florida, Depart ment of Agricultural and Biological Engineering. Young RA, Onstad CA, Bosch DD, and A nderson WP. 1989. AGN PS: A nonpoint source pollution model for evaluating agricultural watersheds. Journal of Soil and Water Conservation 44(2): 168. Yourdon, E., Whitehead, K., Thomman, J., Oppel, K. and Nevermann, P. 1995. Mainstream Objects: An Analysis and Design Approach for Business. Yourdon Press, Upper Saddle River, NJ. Zhang, J., CT Haan, TK Tremwel, and GA Ki ker. 1995. Evaluation of phosphorus loading models for south Florida. Transactio ns of the ASA E 38(3): 767-773.

PAGE 120

120 BIOGRAPHICAL SKETCH Sudarshan Jagannathan was born in Mumbai, Indi a, in 1984. He graduated with a Bachelor of Engineering in Computer Science and Engineer ing from Osmania University of Hyderabad. Shortly thereafter, he moved to Florida to pursue his graduate studies at the University of Florida, in 2005. He pursued a concurrent Mast er of Science degree speci alizing in Agriculture and Biological Engineering and in Computer Engineering.