<%BANNER%>

Ontology-Based Approach to Simulation with Application to Citrus Water and Nutrient Management

Permanent Link: http://ufdc.ufl.edu/UFE0022204/00001

Material Information

Title: Ontology-Based Approach to Simulation with Application to Citrus Water and Nutrient Management
Physical Description: 1 online resource (75 p.)
Language: english
Creator: Jung, Yunchul
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

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

Notes

Abstract: Simulation in agriculture and natural resource management is a popular methodology for studying environmental and agricultural system problems. Traditionally, building a simulation is treated as a software engineering problem, and simulations are implemented through manual coding in a particular programming language. Problems of implementing a model and developing a simulation system include difficulties in managing and reusing existing models and simulation system because it is hard to understand the detailed specification of the system model when it is written in a specific program language. Also, model specification may be lost during the programming process, and it is difficult to maintain documentation describing the system because documents are external to the programming process. Visual simulation environments reduce the burden of programming, but there are still problems related to sharing knowledge about the system. An ontology is an explicit specification of a conceptualization, which can be used to create a formal representation describing and categorizing concepts and relationships among the concepts in a particular domain. Ontologies enable sharing through a common understanding of the structure of information in a domain, enhance reuse of domain knowledge, and make domain assumptions explicit by separating domain knowledge from operational knowledge. While ontologies have been used in many domains as a way to represent generic domain knowledge, an ontology-based approach to modeling and simulation in the domain of agriculture and natural resources has not been well explored. In this thesis, ontology-based modeling and simulation methodologies and tools are developed which can be used by modeler and researcher to build mathematical models and simulations, and in the process provide a better way of representing knowledge about models, improve sharing and reusability, and provide a new basis for analysis of models and model elements. These tools are applied to develop CWMS (Citrus Water Management System) model as a way of evaluating the effectiveness of the proposed approach. An ontology for CWMS was developed using the Lyra ontology management system. Tools that were developed for building ontology-based models and simulations include the SimulationEditor, which is a high level modeling environment for designing a system structure based on a graphic interface, and the EquationEditor, which is a tool for designing a model in equation form and representing knowledge of each equation and symbol by using the underlying ontology. The main contribution of this thesis is the application of ontology-based techniques to modeling and simulation in agriculture and natural resource domains through the development of these tools and their application to a particular problem.
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 Yunchul Jung.
Thesis: Thesis (M.E.)--University of Florida, 2008.
Local: Adviser: Beck, Howard W.
Local: Co-adviser: Morgan, Kelly Tindel.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022204:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022204/00001

Material Information

Title: Ontology-Based Approach to Simulation with Application to Citrus Water and Nutrient Management
Physical Description: 1 online resource (75 p.)
Language: english
Creator: Jung, Yunchul
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

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

Notes

Abstract: Simulation in agriculture and natural resource management is a popular methodology for studying environmental and agricultural system problems. Traditionally, building a simulation is treated as a software engineering problem, and simulations are implemented through manual coding in a particular programming language. Problems of implementing a model and developing a simulation system include difficulties in managing and reusing existing models and simulation system because it is hard to understand the detailed specification of the system model when it is written in a specific program language. Also, model specification may be lost during the programming process, and it is difficult to maintain documentation describing the system because documents are external to the programming process. Visual simulation environments reduce the burden of programming, but there are still problems related to sharing knowledge about the system. An ontology is an explicit specification of a conceptualization, which can be used to create a formal representation describing and categorizing concepts and relationships among the concepts in a particular domain. Ontologies enable sharing through a common understanding of the structure of information in a domain, enhance reuse of domain knowledge, and make domain assumptions explicit by separating domain knowledge from operational knowledge. While ontologies have been used in many domains as a way to represent generic domain knowledge, an ontology-based approach to modeling and simulation in the domain of agriculture and natural resources has not been well explored. In this thesis, ontology-based modeling and simulation methodologies and tools are developed which can be used by modeler and researcher to build mathematical models and simulations, and in the process provide a better way of representing knowledge about models, improve sharing and reusability, and provide a new basis for analysis of models and model elements. These tools are applied to develop CWMS (Citrus Water Management System) model as a way of evaluating the effectiveness of the proposed approach. An ontology for CWMS was developed using the Lyra ontology management system. Tools that were developed for building ontology-based models and simulations include the SimulationEditor, which is a high level modeling environment for designing a system structure based on a graphic interface, and the EquationEditor, which is a tool for designing a model in equation form and representing knowledge of each equation and symbol by using the underlying ontology. The main contribution of this thesis is the application of ontology-based techniques to modeling and simulation in agriculture and natural resource domains through the development of these tools and their application to a particular problem.
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 Yunchul Jung.
Thesis: Thesis (M.E.)--University of Florida, 2008.
Local: Adviser: Beck, Howard W.
Local: Co-adviser: Morgan, Kelly Tindel.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022204:00001


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 E20101106_AAAAEU INGEST_TIME 2010-11-06T18:13:04Z PACKAGE UFE0022204_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 6571 DFID F20101106_AABBKR ORIGIN DEPOSITOR PATH jung_y_Page_35thm.jpg GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
7776ba18aac8cf60628d404753561cbc
SHA-1
6f5201b60700ffd84a00385bc7f46d0f57fe91d4
41541 F20101106_AABBFT jung_y_Page_31.pro
41459510f509b753380092a8ab5e3e59
704aa37df4d8c723bc4f0806a0212df6a8f803b8
89206 F20101106_AABAYW jung_y_Page_14.jpg
7459be9b4d964d70fafe5681c1f8a1cb
9cac6f1476011a0d1917a96308f23ff943b03aff
406199 F20101106_AABBAW jung_y_Page_16.jp2
898030ab203f3b0c9b73d10752ce2a52
be25617a6f3c9d36da5de47b211017ce925d3503
246634 F20101106_AABATZ jung_y_Page_06.jp2
a74dcd11939d1add733a396f411d7560
ae73bb44b8120c59711ab71f800444fab61334b3
6740 F20101106_AABBKS jung_y_Page_36thm.jpg
7d39ca08036228ea364c02366e52ef5e
ee93e17b1585a2a8987453b1aa606cbf74e5fd7d
48891 F20101106_AABBFU jung_y_Page_33.pro
155d5e5b5a78fdf05bc9e316faeb595e
e7685b9b04e651029ecc9a07f4e47a29ad60adb4
86929 F20101106_AABAYX jung_y_Page_15.jpg
75dfc38c70bdd91f886d5e05a7714999
2a81c71445d0d43286248957531f4257f5e57f23
1051964 F20101106_AABBAX jung_y_Page_17.jp2
6f1ec0cd5b1572ccdd4ed29d3bb49412
61446a1a26f6e39b0577569339e2edc00c88c4b8
6391 F20101106_AABBKT jung_y_Page_37thm.jpg
0886f2189892b86823fc45e157aec54b
97717613957684b6ad38ebebddb4d1ee156c22b2
43834 F20101106_AABBFV jung_y_Page_34.pro
9cbb9ab596593cb77747b9fcd73e932d
e6aad640f1f49dfc25087a506dceb1b3106a657b
2408 F20101106_AABAWA jung_y_Page_59.txt
28ce4d565250da6b1c5d81979b1c58db
f1581f408ce992b7bb45abb30acd47f5ef3d7450
31947 F20101106_AABAYY jung_y_Page_16.jpg
45c9d1df3045ae3936b9d148970989ff
148d11295fc1251209115c7cec2f408d0a23f893
1051975 F20101106_AABBAY jung_y_Page_18.jp2
2a262f35d51a0c4c83b285132e0440a9
b10dfa99906f83e0da6af912681d12ebc9a66537
31592 F20101106_AABBFW jung_y_Page_35.pro
8e015cb11933186826154499fe133e9f
5804daa0172395c1428034a3f8153ae10c80df4f
80955 F20101106_AABAYZ jung_y_Page_17.jpg
01c531e016d7c0db3db774b355930259
47f889893ac997d4fa342907532d4c2adcfe9188
1051905 F20101106_AABBAZ jung_y_Page_19.jp2
7a93e82eb7c3c1a9bd603f71ee86bc82
7c9cfd9adf0ccb1886af53beadbcd402011c627c
23135 F20101106_AABBKU jung_y_Page_38.QC.jpg
fdd29bb4331f3d775eaf834bce9c0ac1
ffab01450de42630f4e946208cc896e776430fee
25624 F20101106_AABBFX jung_y_Page_36.pro
1387d80ca173967cb4063883e8f5c4d2
fc534423737ff9fee90b1a5f7ceda4f073e0c881
6310 F20101106_AABAWB jung_y_Page_30thm.jpg
63391ab65f256119495fd2bf638cf26e
48d96d07000f9edd65fda58d46c3e77f675dd6e1
16042 F20101106_AABBKV jung_y_Page_39.QC.jpg
d5f5525b02c6a487feb4957ab1ad5c79
18bbc2c654f8fdb0496b896a199df2faea5b30ae
38549 F20101106_AABBFY jung_y_Page_39.pro
d873bd5e177236a6484237ee4f6ddbc0
584572ed48dfb0182e9c595260394720e1dfc2b4
80746 F20101106_AABAWC jung_y_Page_28.jpg
4a6c7884e2fd3ce52cbfcc249a261337
a6e032a7a79ea5a2d43063009ab88997b8734a1c
25271604 F20101106_AABBDA jung_y_Page_11.tif
0fe53969430201894b16132f37583f4d
d11e33b5cc6a61059ac1f06e6815ab8ae70914d7
4070 F20101106_AABBKW jung_y_Page_39thm.jpg
3f33563eb3032a75f9052296b91f360c
90df20ebc182cf27bc5e2a8c352a110762ecfc20
48526 F20101106_AABBFZ jung_y_Page_40.pro
fa0c4a0b8530bdcad777e5295ddc635e
161f03bd2b5c6f2de799bd3fa72490f64ec44cbb
F20101106_AABAWD jung_y_Page_32.tif
d6310cb5e1f6b4267e91c687865eb7fe
9b50bb4d6e58291501d18589d9044f68c12f0a69
F20101106_AABBDB jung_y_Page_12.tif
0a496f43fe1b9f7a2278e8d633a36b03
8d20d8b0530d1d5d60855bbcea3895052ee6a2ba
26044 F20101106_AABBKX jung_y_Page_40.QC.jpg
42e6d3abc6e9bd8194d8906d3f8ab3d1
690b865ed9b802fc5e8a45066c0daecb1d628586
906416 F20101106_AABAWE jung_y_Page_47.jp2
69ed37aba5ec633df2cac4c0cfd03e97
5d71d21adad9fb349fe3f7b26a0716d1c21f6d7b
F20101106_AABBDC jung_y_Page_13.tif
eadaa49ccf780b57f1cf9cd59b9334c1
4d3028b3e1c52da184c5db23f6d8a13b60436b92
6188 F20101106_AABBKY jung_y_Page_40thm.jpg
f35f712de2631c401fd5623bb4dd6fae
ef05747aa91fdacf4b11829012d705511a6eb8c4
1913 F20101106_AABBIA jung_y_Page_48.txt
7664a4e7034fca9acb4ce84b6495c667
0cc5b4c789bc0f327087a3ae14ecaf2e3b6f0508
54679 F20101106_AABAWF jung_y_Page_69.pro
f9d80abc18d15549d6c846b24ce13ff9
9b07fe1384338b650e25c2d3fea1682c813fabc3
F20101106_AABBDD jung_y_Page_14.tif
7ca8976c7212285c6696919d51892f18
ec689602e59a3bbc39ca663e048dbde87e599ce5
5217 F20101106_AABBKZ jung_y_Page_41thm.jpg
1a0fac9bd0c9adf258259e30418addde
b642e2baaf344ce69f6a9d3a89446d2dece33ba7
2786 F20101106_AABBIB jung_y_Page_53.txt
9c135d15e7d2bdcffc2122f6194ae4b8
5993dd6b6b1c8c761b994347ba7bfea13ffc11b3
83480 F20101106_AABAWG jung_y_Page_43.jpg
5da421a6f6d4ff8a78f70fdf16708a33
c455b57e5d8516ebe5f6db9a04dcd7587af560a6
F20101106_AABBDE jung_y_Page_15.tif
656357afddd942b2c1c6a3a31c5be088
2deee0f183e10464a450ac1077007cba7c975452
3059 F20101106_AABBIC jung_y_Page_54.txt
da834625262ea9902bf97319d67ff83a
f4badc277e8014f75759d76479fbdcc1966679c0
6284 F20101106_AABAWH jung_y_Page_12thm.jpg
0faaa5952f281f021dd69c12fc10dce4
93398446ca20a930c913043fceee8d7f66076afb
F20101106_AABBDF jung_y_Page_16.tif
32a491c49a9b2de378c4a42de839be76
be56332cfd39b866281cd0067919b41a5e3cccd5
3047 F20101106_AABBID jung_y_Page_55.txt
4f1412d4a0a387867306609fbfa7e08f
1080537c12bae6fa0c99f6e49fc55ad6a0cbd0ef
52513 F20101106_AABAWI jung_y_Page_58.pro
9211182c4aa6c52bef52a6d0fcf2f97b
c82ee4fc3aabb717cc05979d3a508e98cc25618d
F20101106_AABBDG jung_y_Page_17.tif
ebcef80ca5741ccf65562e2d03cabae6
09ac76c2730b888da7df33965db7b83b848ee9f6
2624 F20101106_AABBIE jung_y_Page_57.txt
168802ad4b97162d921081e1bfd28087
cb287dd470b060c573d3cee8fbd1ee27a0c6afa1
22946 F20101106_AABAWJ jung_y_Page_11.QC.jpg
aeb84c8e2b897fe76d38757f7e7db41b
7b23dfad2ca2637733e883bed55b8989635553c1
F20101106_AABBDH jung_y_Page_18.tif
eb148367ce905cbf9bc4d6b3ce78c9ae
a0b664e9981b6da74bdedf3fecbc08388df92eed
2637 F20101106_AABBIF jung_y_Page_58.txt
e8e2b0843bf19eddc3f1a504f064b1a8
95b57d4ceacc4ba2dbb0272ffac70e64241701e0
108260 F20101106_AABAWK jung_y_Page_72.jpg
4944b3f0e2ba4cd27de0ab95abf3d28d
8f7b872545383d8c78665cfe3ec50e1bbbea2d12
F20101106_AABBDI jung_y_Page_19.tif
9e67624e2a88332232eb4795be7dc596
87b4e2f1cac2bfbffef3260dd5512c1121b30088
1967 F20101106_AABBIG jung_y_Page_60.txt
ce5f92badbe926a196a2d7a68b54ea5b
227647c1950fb6cb449ee0cd2ef982fa50017338
F20101106_AABBDJ jung_y_Page_20.tif
49bfd3bd3fcd9febd6babbfe6120f889
f63af31ff93c5805f098de613b688dd78d32c966
2165 F20101106_AABAWL jung_y_Page_18.txt
6e98f05044eaea8497bfebb0fa4b1886
84cca5eb93b2a9220719be4e6cb80c696411a168
1464 F20101106_AABBIH jung_y_Page_61.txt
8099819c0b35728ee2f825c2071d28ca
326a9d957cf393acffc9888db9e56f47635186af
F20101106_AABBDK jung_y_Page_21.tif
aa757f10f904f21e197e010bcb37feb4
092af13e95316387296db8f61c5e92fe30080533
53893 F20101106_AABAWM jung_y_Page_66.pro
74b6b21badf94fb3817f129ffbe90ac8
8f26064966eda0d564f139c1c9418022002f1c24
1979 F20101106_AABBII jung_y_Page_62.txt
e6a3067bad513b845d6b29106a9e2722
e93c752ac4afe696cc3ae288de3b205e3b0ce836
F20101106_AABBDL jung_y_Page_22.tif
fbf6ba40817c6fa0dc436a5f42703cbb
a46c23dea9a81ca56a81176cdad314124a29b126
F20101106_AABAWN jung_y_Page_63.tif
8b6b8f0ea099dc9734a7d3ad2701bd6e
0bb5d91bef91ca77c0a37d87fd832218bbc251d8
2200 F20101106_AABBIJ jung_y_Page_64.txt
554dc16f701998350d84840e65a4cd69
8dfcf94e5736d95ceb183f8805287b140c78c75b
F20101106_AABBDM jung_y_Page_28.tif
6e4d11f6b199de6bee7b01b507f1a71a
3ce3400147fb53fd8cd3ede679860a49b15d682e
86076 F20101106_AABAWO jung_y_Page_65.jpg
a0cc5cd32860dcef164dd3254e183fce
ee206e622f1b6362e3656211f4e8f9f78badd967
2346 F20101106_AABBIK jung_y_Page_65.txt
9e79e236c9c6bf5b8defd8a3866eb277
eca0b7e56f0746193d3a0b9ad2a00643f36f9f4a
25424 F20101106_AABAWP jung_y_Page_43.QC.jpg
e9a9ce19dcf3888eaaa89ac75ea711f2
4d6ca6ebcd7dad43449e23673eda20d28beaafd6
2122 F20101106_AABBIL jung_y_Page_66.txt
929eb26b65f302b4ae278b6a3e13f077
8d55caff6c4d1220ea5069f03b66769d02a62625
F20101106_AABBDN jung_y_Page_29.tif
e98dc83e56367ae909b9bac8400010d2
e8559d3813e178a6875bfa856ca9915eb441ea46
1937 F20101106_AABAWQ jung_y_Page_44.txt
a02d189d49bef8d747b17dbf13a8f2a5
21c18b71e956c6204f6099a2882b7f50991698e2
2406 F20101106_AABBIM jung_y_Page_67.txt
5ed2b756895f841b18fde89b3f464169
a7ff91bf55da99c4774f5ddecebce02cb2a5ae7f
F20101106_AABBDO jung_y_Page_30.tif
c7520216971896c2e5a0c5001dcc81f8
8965a7daf735fac270ec04362ae3350d6160603c
55304 F20101106_AABAWR jung_y_Page_56.pro
4b5000998b83b4debba0b04be0631ff4
5377c4f6b7b55c747988b439245f1aed222a41e8
2154 F20101106_AABBIN jung_y_Page_69.txt
ce0d6212a4e69f1623b5ca11665eb0ba
49a30b9e812fa39c44e6ad936c8576dfa90bf91d
F20101106_AABBDP jung_y_Page_31.tif
c3217f70cd393c6a3bbb394ef712c84c
005d829af30d431cc4ad4d1fbd3fcf28fab47879
100397 F20101106_AABAWS jung_y_Page_05.pro
ff13c04a8cb7dcb6b0d9de86a58c09c5
19f76ce7139a73f0ddad59c0b50ffda99ddc6de9
1233 F20101106_AABBIO jung_y_Page_70.txt
85153dad8df0a805e91e6a198a708733
529b86e98262b48313f3769858900d9d59e2cc32
F20101106_AABBDQ jung_y_Page_33.tif
13c9882a97a3579b0453bbbdbf0b2c30
b06970c3e2c36ab3097d60caff7b1fe1a58c474a
26143 F20101106_AABAWT jung_y_Page_46.pro
41eb11622feebe0a93f391585ec9daa3
13b3a5cc5d53ff6c7f05dc6fe11b2bacfe16fa53
2256 F20101106_AABBIP jung_y_Page_71.txt
0f4b23e79f5c8843f0db32bb39288f06
4d182323b2de64e8c9d07c1de1ece8a48c56027f
F20101106_AABBDR jung_y_Page_35.tif
184118d0b1a06ae9a61f9b5be7561496
fec14bca617ee9a297940eafb2d2d7bb2063c6dc
20815 F20101106_AABAWU jung_y_Page_46.QC.jpg
88db64a3f2d26a1ee784cab80056b792
b0eb8c5f216bd31ca58a00ea1514935b2a40aada
2493 F20101106_AABBIQ jung_y_Page_72.txt
baaa0c4eb27d7c4fcd66a58a690b66ec
686715f9139e31de73750275bbaaedceb01513d3
F20101106_AABBDS jung_y_Page_36.tif
1caf9a60fb78ddfca3146eaffd15e12d
8139f2bfcd0d48d85a1a7507e7e17078f6e962ad
2269 F20101106_AABAWV jung_y_Page_73.txt
468c4335ebacd1d3e6076fa237f81165
110d9efd43ad00ab5304fd2585955f2e168b0d1d
1359 F20101106_AABBIR jung_y_Page_74.txt
647845776b35563b9240ed4474805793
bf16f6c5b128eee85ab93c4f88a4fcfe17e00346
F20101106_AABBDT jung_y_Page_37.tif
97e4d4e5bb1ec3e1048ceb0121f79e93
bec6b47ec81493c7cdcb3126584a69dfc7698423
54810 F20101106_AABAWW jung_y_Page_65.pro
30f9db074da5f0da4e1801d615ebf7d2
159d29770520c314637f835433c32ab03f82eaf6
F20101106_AABBDU jung_y_Page_38.tif
c8d5fa3bd98043642a6c31b76f61a887
b9eded8e5a0ccda1faad7bef0b6eae3a7ce2283e
1291 F20101106_AABAWX jung_y_Page_03.QC.jpg
8e5edf522bfb2049b31c147d4b9359d6
4297878ddef8c69052fad7c621286168c6239fcf
440 F20101106_AABBIS jung_y_Page_75.txt
c7819d529d5c2f23727c9de7fead618c
ed90ba0af76c08396ac49cca792c8714a1880311
F20101106_AABBDV jung_y_Page_39.tif
89ade260c16b94639a16eed2ea599c80
2371131638e8368a9bc7cdecfc2235cbd4109f22
70570 F20101106_AABAWY jung_y_Page_48.jpg
5673f3c98919045c066d19d8bf129ebf
0baa13f9e82dbdb2716df5fc5afb00200d810099
3444643 F20101106_AABBIT jung_y.pdf
0ad39f687268d63d434e851d6e7c1f48
1965a5e72897132607ecdda39fba3b2db7506699
F20101106_AABBDW jung_y_Page_40.tif
260da152e597e6c75c19029eacd92571
d2fdade65cfdd6305ff8774d54e0054983a46599
989256 F20101106_AABAUA jung_y_Page_11.jp2
3f1c18a1b4593094c53889bce8966626
a018c0ee706e53997dcef2ff2f0e7e430a1bbd6e
26979 F20101106_AABAWZ jung_y_Page_63.QC.jpg
3a6d8c4f65ace9ac7f71548b19dd63ed
d3e6120aeb60c6dc032b252081ca04245050f105
6913 F20101106_AABBIU jung_y_Page_01.QC.jpg
feb4943574359d140a3a42813ce32d24
7f3b5364444344255e8edd6269d3f27d7d350daf
F20101106_AABBDX jung_y_Page_41.tif
9019c601273e96ddee4c8131990a4267
2b87d9e373262c2c99a87f9848fc54b6b44e6f7b
2018 F20101106_AABAUB jung_y_Page_33.txt
64e1562a0aa1e825c5b3937d3423311d
bca3521de2894e29f5d2778ecf2f6a20f78dbea4
446 F20101106_AABBIV jung_y_Page_02thm.jpg
45632931012928147d84d92ab66be344
cc32a6a991d69cc1f2f3f21651f8a71aa3fd4634
91769 F20101106_AABAZA jung_y_Page_18.jpg
524f76d9fcd248ba4bdfed4bacd54963
00ba6de1cb8541e0359535b052f13f99c3e20374
1051957 F20101106_AABBBA jung_y_Page_20.jp2
69c83cc8b871cab6ea6ffe010df43067
141ee87b3161883744481565d69a38b73463711a
F20101106_AABBDY jung_y_Page_42.tif
fb681f1005e506e7a7cec6ebffec855f
b754ff3661e52cb8233378c02ee956cd75dadcaa
741333 F20101106_AABAUC jung_y_Page_39.jp2
8baac458e40397bdb8cba13f909c6887
1d15855606421ea3158e799bb6cdafc04a476fae
522 F20101106_AABBIW jung_y_Page_03thm.jpg
1323cb863a845ec71d92c9dd5129c277
b92bb564dee1ea6ef704aa16e82fcaa42ad4ad77
83014 F20101106_AABAZB jung_y_Page_20.jpg
613b30a76b43ac1d9a3d8539d49be5db
3d76b9a89840ccd6b65f59eb6a94cec4018e5063
513063 F20101106_AABBBB jung_y_Page_22.jp2
bd645d431d2c16032006a9f72f314374
90c1ca7c6ae842dfb827d68067e20fe60c270b53
F20101106_AABBDZ jung_y_Page_43.tif
6ec8b365c50580d099300092db0c42e7
27964f1bd36748fa0f69e36a5aded7b301e6dc5e
8423998 F20101106_AABAUD jung_y_Page_25.tif
2f26931ab80e06003d9fd03020c3593f
a8103650c3e8ac52d9f5677a85fb07151c5657e8
10260 F20101106_AABBIX jung_y_Page_04.QC.jpg
97d61bf3df790f5dc129c4c064560b32
f029ab14242de59b02c892d603189e66806ecb06
90266 F20101106_AABAZC jung_y_Page_21.jpg
cdf3ffd3e86f311adff8885231f48b62
01fcbb2f308067b250dc4c979e022f6c9fefa40f
1051916 F20101106_AABBBC jung_y_Page_23.jp2
9a8e1b70f5ef0cbe26eeb12a1e505f46
b54dca565368c08e5dd618f4beacc1b9ef6b883e
1051978 F20101106_AABAUE jung_y_Page_21.jp2
93cd053e52c0a995d57f6663e4f4dfbb
a80346ef39b173b31c1326db92a4d039ff2145bd
2581 F20101106_AABBIY jung_y_Page_04thm.jpg
a9a018e177e0b9b90e7432582af4b314
6330da17de8cdc39e76762a03b32c30997150169
39989 F20101106_AABAZD jung_y_Page_22.jpg
167e48467da9f2c065ee333e5ca314c6
971ea7c581c14d6b0848494037d9824773d65eb1
1051960 F20101106_AABBBD jung_y_Page_24.jp2
81d96b40253de7551f7c4e99b77da7d7
a00019707eeee637c17d1752ba3081dc0bce046c
79700 F20101106_AABAUF jung_y_Page_33.jpg
be80e9202a2d9b449552323620c7348d
7f4ba52bb84b2189004b0fc6b176b1d7299a65c8
27370 F20101106_AABBGA jung_y_Page_42.pro
148bf68c80338f0cb02b97235dad1010
15a472e152709047487c37bdaf0702946fe0533a
20583 F20101106_AABBIZ jung_y_Page_05.QC.jpg
750e544b4188ff4615a474eaa27b1939
3a42a11f65c74b4c4fd0907c5d18c4879e7aaa7c
867233 F20101106_AABBBE jung_y_Page_25.jp2
84b290156749887bfbfb72e77232388d
dce28590bc91d5a45fc4a338b217fb44b232d8cf
70239 F20101106_AABAUG jung_y_Page_62.jpg
ab2c585b50fd74750804c207fc78daec
b46925137e8949d94f87622dcab36a217e5ee1c6
50768 F20101106_AABBGB jung_y_Page_43.pro
3cdfd768ce960da43c29b9ab50f16ce7
3d81cd1fc051af3e3f2c8fd3b5b433101a8040de
84624 F20101106_AABAZE jung_y_Page_23.jpg
c0a6076fb3d1acfa5677a79b842900f9
a811e5532644b3357881377650657b612b3b7536
1051962 F20101106_AABBBF jung_y_Page_26.jp2
54303772f6f56bb294a04ba152f64321
34584471dceda9dd39a3540aa5dfd704868b46cc
69209 F20101106_AABAUH jung_y_Page_41.jpg
8f4699374c0a5dd79ca9a1aa664c1e58
1bd24f250763e24d6983c6b1bcc1194be8f80365
48895 F20101106_AABBGC jung_y_Page_44.pro
74baf5507ef52c7d7ed45a1782f89510
e3374967c372a014ab637b783f10d03f6d8ae2ee
21261 F20101106_AABBLA jung_y_Page_42.QC.jpg
c6395e3cbf3bdd41ee19e5cdad36c1a5
ed0b4dd163331fc1909da7e5bc091c5cb574cfc1
61822 F20101106_AABAZF jung_y_Page_25.jpg
5d7ee0879d40ed5607986bc5b35e3d58
0850ff771fe0d2cd5deb7308ad97bd1538edc580
1035241 F20101106_AABBBG jung_y_Page_27.jp2
6cebc27a4460f8509103924f22fa671b
b75bbff96ed870a96244f2ab7b9fda99754410b2
892 F20101106_AABAUI jung_y_Page_22.txt
0e6730b14b765260d312eb940040fe80
499701ffa5820640d4a529a62e15c1f818b3b308
39677 F20101106_AABBGD jung_y_Page_47.pro
9701762b18c5dba7d5f4128ca39dbfae
584407e21db8da9a0d9152e3d7502efa6f0c10b9
25476 F20101106_AABBLB jung_y_Page_44.QC.jpg
149bf50f959ee711f5aef0d28dadf332
eaead57b0dac49ecca5d37b578e899865d019dbd
89113 F20101106_AABAZG jung_y_Page_26.jpg
7d91ce21ebdba3017f3842d2ff7efdd9
b4672a32d8b484c8b8a220192ca9af6036c7c336
1051979 F20101106_AABBBH jung_y_Page_29.jp2
1b0d794edf89a881f2a6091bae4d9d92
4737de47c92354782978e4f9d304fafd88252f7b
55141 F20101106_AABAUJ jung_y_Page_18.pro
513825f64a785629c97efa6254b77039
faa68169fd382f0a604bfd5d51cdc6a8350dc769
45095 F20101106_AABBGE jung_y_Page_48.pro
a1320893a533602578df44cfa6abb845
d4602aa08ebf17d14e42931af85632c9672dd549
6275 F20101106_AABBLC jung_y_Page_44thm.jpg
08b4c3070fdac821ef801e315437ae02
5b36d4b4dde256d5d4eee087fe80ba0cae6d9b6b
1051981 F20101106_AABBBI jung_y_Page_30.jp2
387363ffb969bb8fd6980138fec37782
e442fe359224deb8381cf231d939ff3640c560ab
5713 F20101106_AABAUK jung_y_Page_42thm.jpg
3c9b7ca83afd0d71307c912b47a2d6fb
4a3dc0b0882b5eae614d976dc992cd5cd2656b4e
53238 F20101106_AABBGF jung_y_Page_49.pro
424ad51dc2c3b0cfdb72994a075e9272
94f617c29a81b6de8818e6867772c02d93b05d43
70640 F20101106_AABAZH jung_y_Page_27.jpg
fb6ee98d4a22cd73fc0dcf373b8c368b
d34067979b580db0c6b69f4eb1b834bffbca133a
26412 F20101106_AABBLD jung_y_Page_45.QC.jpg
d4e9fef59aab37a8ac728ef9ab05ca6d
1fa04dfb132c455e831be1e9fb94f9c01fb44c26
994440 F20101106_AABBBJ jung_y_Page_31.jp2
ebd60d444d28e70f29e9e59574505628
510a372eb4fdabb7e3a1b5fde3cf6480d6011f54
28879 F20101106_AABAUL jung_y_Page_73.QC.jpg
836c402e985615906e3a4bea0fdf75ff
ae3e790b85349cc64a2b03038685bb2ce04d29fe
29542 F20101106_AABBGG jung_y_Page_50.pro
ce25f984bd1b5b4e3e9c4954efc08369
12ae85c627a8e01de1eaa8f3971be5d5f1cd8a03
76346 F20101106_AABAZI jung_y_Page_29.jpg
ca94e5739861b6d34cd7c070c00df9dc
de3780206d1face35ad563166e7fb7b4f9f0d43c
6995 F20101106_AABBLE jung_y_Page_45thm.jpg
f85eca8173d0055c23bd00fc00030ad9
8bab37eb229741fb0172a0c15527840ce67e742f
1051976 F20101106_AABBBK jung_y_Page_32.jp2
0cde458984f9b9edeaa3670c925442d2
36025d324ca17e42a3a00efb61ddd49c952bb708
41424 F20101106_AABAUM jung_y_Page_68.pro
7b217f935fd712b588a3fc07fba73c88
2cb8cf97b5d654e62e66f43a5ab3d761d1103e8c
47733 F20101106_AABBGH jung_y_Page_51.pro
952e3559f56cb7ba3ff14e4c47ad20a6
8ebe0a1c93d8450ac0675484871a19f089e9840f
87217 F20101106_AABAZJ jung_y_Page_30.jpg
a9fd974c4eea1a9b929ec06ab61e1935
9f32bec765acf11702b4c5706f0b83b599468c5e
21443 F20101106_AABBLF jung_y_Page_47.QC.jpg
fd710ad14e7fcafad19b2893c0335ac1
4b78f50445807216e8832e4d2a0d0a748fa4e6e7
48790 F20101106_AABAUN jung_y_Page_20.pro
a800f43f32b18a07e89096beec9cefc9
d7d67e2ab8a294b8143c7e970a5615c350ee38f0
41296 F20101106_AABBGI jung_y_Page_52.pro
da2c14a6a7c646663352ccc55b4f0d5b
d0a424bea0553f1f48db8e82452a464543bf051f
69448 F20101106_AABAZK jung_y_Page_31.jpg
0baa2a49a115546d3629b2517de82ae0
5e2d1f90be12b972aca563d16ca85fb6bcc1b9e9
5607 F20101106_AABBLG jung_y_Page_48thm.jpg
cbf9ba6c7c5473957452f207a60a1f95
d2a203c33c4e32dcc59a7183e841aa14338766b3
976682 F20101106_AABBBL jung_y_Page_34.jp2
766da94efe23778d234edd9a1ac6c0d7
c1c76876627dd372eee87c55f34f14480785439a
70444 F20101106_AABAUO jung_y_Page_55.pro
3080be30df6aead16e3d0ad8817e4dc9
37685269d8838d121baafbe2ae4ede41f4ba415b
65424 F20101106_AABBGJ jung_y_Page_54.pro
75ad115044ff3e393fdcedfb07417c03
e630eab1c7216b709d777f1f0c8787ad0123d0ea
72689 F20101106_AABAZL jung_y_Page_32.jpg
98cc01b545968df19a1b1aa12fc5844c
6a2380b09c1567a551b5873f975fe24db444f440
28143 F20101106_AABBLH jung_y_Page_49.QC.jpg
921f4a3b73894fa9e68672086f5b3fa0
dc383fa7871d589161057435fb08b42527dfd97c
1051965 F20101106_AABBBM jung_y_Page_35.jp2
2fedfc0371a0ae145f69c3e35fa8c3a7
22d2e8373b7bc9cb8b0b2ed659f8b63e7461da06
1694 F20101106_AABAUP jung_y_Page_11.txt
e1b8c529a9be1388be595e3c3b434676
f6004a0cdabcd6f2bfc19ee6de12a28c8130b024
49693 F20101106_AABBGK jung_y_Page_59.pro
35126ceeb3f6c8f6d2e1dfa612ab8e3f
ebb429006038e85bf64e18f033130ed5f398b349
80744 F20101106_AABAZM jung_y_Page_35.jpg
1638c5e10d2d975a2707af2949cadcd9
bd6d6e1947200f5ce086e0cd0b4a86e8f5cd70a9
6495 F20101106_AABBLI jung_y_Page_49thm.jpg
355d14737229d2b55199b93ba4742ff5
345a6514bd64d51339c79b51d5c3d4668d5ea838
1051966 F20101106_AABBBN jung_y_Page_36.jp2
715ac7baa2ac36a2e1949c0d71bc8300
75db7767047f0fbda84b91c611019d23c6b320db
52718 F20101106_AABAUQ jung_y_Page_19.pro
ff9b5508717fc1e12734727b40905d70
1ed49f1156abd568d4569b30abac1b9e4e0685d8
48589 F20101106_AABBGL jung_y_Page_60.pro
4f7decccb5bf54e2739d343110e658ea
383532202a20f3591a4e51381bd90d81c2530722
88131 F20101106_AABAZN jung_y_Page_37.jpg
37f56725d6dae643ce85ee034fed0b83
eeb6669fbb4d52e2d652f7263009bd53621fa2dd
16676 F20101106_AABBLJ jung_y_Page_50.QC.jpg
5d84babc04953ceceb7cb010c53c8065
ac0bab8ec23b5c37a37d874fa616611b59f67821
36023 F20101106_AABAUR jung_y_Page_32.pro
16c5bc6516c32993f0d186c8b4c1c8ac
face93170599a4f8aa32b414eb52cdba590a416a
34159 F20101106_AABBGM jung_y_Page_62.pro
5b060d7237ca83839861e89e4857e41b
a2f3047a987435f57f1f90cf72beab8e318b0c5e
83579 F20101106_AABAZO jung_y_Page_40.jpg
d42a1484c66e8cf4f6df9ba143575382
a991c7453d0579dd455318767aeea757bacd7269
F20101106_AABBBO jung_y_Page_37.jp2
cd8be0914063cb6405438885abd13d46
aab09cef06d761253e1972196fc24e308ff02faa
5034 F20101106_AABBLK jung_y_Page_50thm.jpg
fe9a3172d1b93b190eea2b6de5e0657f
89296aea2c027b39222aa9a5a1608304cddb67f0
58318 F20101106_AABAUS jung_y_Page_73.pro
ffa274f72036b1bfb74fb7da591537b0
b6d32fe7a7ccf5c58723326f596afaf47aa90d04
51622 F20101106_AABBGN jung_y_Page_63.pro
e34a75f54ae569d4c59957fb3560fb0f
25d21a57f23dca04f623f3b44ce6b4b8a739427a
82200 F20101106_AABAZP jung_y_Page_44.jpg
5569041971c8c876db9696d8cf9508b4
6c8b9cd34f97e42d8d166d9ea342696f58153500
999911 F20101106_AABBBP jung_y_Page_38.jp2
8c9bf010b2f41394775bf7637ea26ce5
6fc912d2c76b900cb4818ebaccf495a58bcc0ec3
25182 F20101106_AABBLL jung_y_Page_51.QC.jpg
8502b9a575908807238738132c8f0601
348df03a452920482009ff28bcd5031a43661f2b
52668 F20101106_AABAUT jung_y_Page_15.pro
a60cfefd1aeae9037cdc70ed80e9dbf1
7fd726d8b4c74929a52c841cdaed801f1543687c
54362 F20101106_AABBGO jung_y_Page_64.pro
304b2be4938ecc392ad6da406d4e69bb
cae70479ff75409d8d4528e4c9d9284ea126c5ae
81394 F20101106_AABAZQ jung_y_Page_45.jpg
2bd079c2a4447619ccaf242e123a3351
8a6bfd2f2c51611b917e58d1fd3c465aa4237436
1051982 F20101106_AABBBQ jung_y_Page_40.jp2
4fcda628b0b40fec61c3b3bf87553172
c65d8ff22a42c0c2a4c8e430eece8709f9c55b70
5834 F20101106_AABBLM jung_y_Page_51thm.jpg
38b6889f8d6bf9ad26cc6b54a0f08751
93c6794eba750c640546093a4e0a53b2a81456bf
33925 F20101106_AABAUU jung_y_Page_04.jpg
cea6c4bc0513f1668ab81a99494d85cc
955bd95f5cc2693b32221f11b62350cfdadc40c1
31145 F20101106_AABBGP jung_y_Page_70.pro
2dbdbe8d8cdd9c8c3341d27c12bde20b
b3e894a24f5873a08e4f74ba72a8df084a01247f
64425 F20101106_AABAZR jung_y_Page_46.jpg
c68074607d51cc929d61dbe8ced305aa
c8be1f0ecff4fd4ae8234cc58958e40bade1e3e4
916758 F20101106_AABBBR jung_y_Page_41.jp2
9198a746176ea80fba76f1eb0217ee89
9e65fe921f02db61cc581a6ecddf09a9a86a346a
21514 F20101106_AABBLN jung_y_Page_52.QC.jpg
40abe196bbfcbd13a028cffee932472d
7c703a75180b8ed0aa8fe2d461b65d8986bec8a6
F20101106_AABAUV jung_y_Page_09.tif
2a88cb30afdedd9f92424e4410f4aae3
7d3eec06dfdbfa6d04238ae1c290c066e5380833
68513 F20101106_AABAZS jung_y_Page_47.jpg
4b71ebc06fed64c9d61a42e889d5adcc
90ce8ff0e490d5d981ccd3f6802373b87232826e
F20101106_AABBBS jung_y_Page_43.jp2
4246050e335ad9bf562f392f75e0029a
cbdef8fc63d241053666e9399048e2e07daf7c74
5828 F20101106_AABBLO jung_y_Page_52thm.jpg
82c72422b4de8459b09aa6c6456d2091
ccffae6088d52bbdadaf7e25e7a029fbd046bf08
70496 F20101106_AABAUW jung_y_Page_68.jpg
64dfe40d26ceeb2aea230c85bd410717
b9edbf6e253322308faed2098b40595d89b04653
64349 F20101106_AABBGQ jung_y_Page_72.pro
861620399a315677639125f6e1b9c705
72fc846ea97f6a43720ee576c9f354cb0cb44838
87704 F20101106_AABAZT jung_y_Page_49.jpg
57c0378a3de934b9a9474f7c63d3d3ae
c751e066e45f398d3e8662c094d37f404f7cab8e
F20101106_AABBBT jung_y_Page_44.jp2
94658f611bf7e019d3884d94b9d4064e
72730794111cdaf4c2887c7cc868593f31e25380
6351 F20101106_AABBLP jung_y_Page_53thm.jpg
cdb9f297833e8daaa9c88f639830112c
5a5bf1c6215cd8f0baf04331a63969b4b8bf586d
2115 F20101106_AABAUX jung_y_Page_26.txt
201eb6231afb32d960d9010a1a6e3990
00f826ac19a9612a4a723876ae9c37491c372381
34574 F20101106_AABBGR jung_y_Page_74.pro
55ae26a47f73f6c361264e39daea1535
432c1997700ae19cb88024928e0afad4479f8a5b
50110 F20101106_AABAZU jung_y_Page_50.jpg
31a017fa2be972ba32bf5c65e9430738
b135ff18b6149f306319a4d93fa06f4cd3140da7
972030 F20101106_AABBBU jung_y_Page_46.jp2
764f64235f67c1280fa6eff5c89941d8
a2a1293d9d7404671bc3584b3a4820a353618e50
19306 F20101106_AABBLQ jung_y_Page_54.QC.jpg
3be98bd1746cd3c98d046d0f69f550bf
370488d10c042ab964b81f1fd92adb5f2f45e49f
23113 F20101106_AABAUY jung_y_Page_34.QC.jpg
586d3edae926283539c1872b316ef655
01ef25fa967c1ff3aa7c95075c7f91ffb3bbf38b
486 F20101106_AABBGS jung_y_Page_01.txt
77768fb4e4beecac46309d5ce04e98b3
dc95863c7cf9a68fc1b6ccc178c5441f0760fd13
69621 F20101106_AABAZV jung_y_Page_52.jpg
c1caf960df5cccc55cb826a675b11c28
6501ebf7d211662c2466ce08fcb761edf8a16f00
1051970 F20101106_AABBBV jung_y_Page_48.jp2
4ce94a8d3b1b55e92f518590e3fe7f1f
4d3b76dcc9c7e429b50a5f69335b3efd25eeac93
5069 F20101106_AABBLR jung_y_Page_54thm.jpg
fe40e9a68b94565db0e7ef8ef31daae4
4878fd228c22bf3535835cfa3cff1237a18ee6ff
10094 F20101106_AABAUZ jung_y_Page_75.pro
deff00c3d7b070590503c0f0891b8c3e
a0bd12e93d555d8dd68e76e30044f3d664cece35
84 F20101106_AABBGT jung_y_Page_02.txt
258bad75b763b6ad0eccef6608bb8c7e
5287466d39265011d8e9e8c1868af8ba99311882
81436 F20101106_AABAZW jung_y_Page_53.jpg
58d3dad5e7db10ab98b05d2bbd01b397
f44b8fc38738049ec80c1beca1e2da05fb08a5fb
1051967 F20101106_AABBBW jung_y_Page_49.jp2
f09dec7e83707a32a1c00eec6708b9ba
c53444271502e2e424dd43045533fcc42259e6c0
6343 F20101106_AABBLS jung_y_Page_55thm.jpg
a1cf6206a745507eca759e29fab44d5c
bc87db589a72cedd6c11f8f6a758c4ee068f16ad
111 F20101106_AABBGU jung_y_Page_03.txt
a7537cdb780d0493fbc07426d45216a4
a80d6d4044a187b38c17865cc9586923e372fd82
61813 F20101106_AABAZX jung_y_Page_54.jpg
60c75cad7ea7a49de6c94a3a9f3d922f
4c7cc3f92eff7e0289606a10ced179c3c05af962
679839 F20101106_AABBBX jung_y_Page_50.jp2
2ff58d81ce377063d258cd169af24da5
8595a1fa63f336664984c8bdf638e9b6968d492d
28850 F20101106_AABBLT jung_y_Page_56.QC.jpg
e4ced4e306db5536edfbc130b794ac3f
64a5db00fbd02f0726d5bbba9d43f8b2d9cec908
764 F20101106_AABBGV jung_y_Page_04.txt
1a7bd0d9e44a4632be0902b8ecb21b3f
b8b607d217ae300778fe1a05a91a1543ba25c52c
1948 F20101106_AABAXA jung_y_Page_68.txt
8004fd4a7381b46bce41a9e61fe9fee0
827415971991b5c8e2f3351deb0510bfb54b0a1c
83920 F20101106_AABAZY jung_y_Page_55.jpg
15fcb9de42eac90174bdc26f6ca73372
d7888f9765d59c642eb5ea213d58013a4a186676
1051938 F20101106_AABBBY jung_y_Page_52.jp2
c675721d283f01aee207ee68d621a7b6
d72b0cf3670bbc4b35ed2c5581292d35e1b30c0d
16534 F20101106_AABBLU jung_y_Page_57.QC.jpg
071f13844974cdf9b9f781ae2d4d967f
997f15b3b999d6f37923104f3f6b97d10eed50a2
4112 F20101106_AABBGW jung_y_Page_05.txt
ff034ab0146b35b88c4a994577b9e91a
c8c9cbd278a62d16dbc7ff16b5a1ec0409ec37eb
53683 F20101106_AABAXB jung_y_Page_39.jpg
ea263403ca78c3af8d4bd4768e085147
4cff2e7676ed214032447c707d4290ca0cc6c39d
91188 F20101106_AABAZZ jung_y_Page_56.jpg
7d1bf7b37b2ab189585ec4ad10f08607
d1a7f9de293a0e2be20db0b8ffde94105b8245d4
1051955 F20101106_AABBBZ jung_y_Page_53.jp2
a811eb5db83a2f0bc8f566b51c955ac4
be1482e7a44d8f6a8db6d9b25da34b38da2a710e
470 F20101106_AABBGX jung_y_Page_07.txt
b025412b902da482b4a6cbc5d6d0d253
d08006fe93979e39e2dd63bdf1db0e80bddcff37
4388 F20101106_AABBLV jung_y_Page_57thm.jpg
2ab60dbba0e1a09a066ef0c702aa396f
89dd6b55a39d95ad440ab5b9e9171986e15c6276
F20101106_AABBEA jung_y_Page_44.tif
f032a27d3264f567a92e89f266dc5186
27f58dacd2f031a3c9e9599844b7b9464e250e0d
2843 F20101106_AABBGY jung_y_Page_08.txt
64f3f30a287c964228c8bb9b81f5cd5e
48171b58565350b64245873ada7ec47544e48643
F20101106_AABAXC jung_y_Page_05.tif
8a338516f9db5bd70047347bfb52c8ff
e06b6b2fe6f726d74ab96ba745da70ad05cfb354
20668 F20101106_AABBLW jung_y_Page_58.QC.jpg
28c82e4c39f8ddd626b7e00476c8ac88
6cf819540f254be93e8cca3001606921f9e5ea47
F20101106_AABBEB jung_y_Page_45.tif
4e494a4e50fb03c93ca2945fbb948863
dfff81fac1a25bfe75c6af8b338892cc5635a2df
860 F20101106_AABBGZ jung_y_Page_09.txt
a75b98b630c818f0287e1c84ba37f385
d16ca76779c7b48b4034f2183eb5616899ac5a4f
90034 F20101106_AABAXD jung_y_Page_19.jpg
a5174254f4b589a784966a84476d8cd1
463602c42f225ad83da2f2deb5c524cb916b2ff5
26857 F20101106_AABBLX jung_y_Page_59.QC.jpg
55b7a2f4123a41cb64bf12a196da4202
ad7c347d0484b2d3fdb37f4107f6742bd8c3b67d
F20101106_AABBEC jung_y_Page_46.tif
7bfb1c1155a0f3339427276b19b89ca8
412963d25157c2ec7f475ab827ed764385f32c50
18287 F20101106_AABAXE jung_y_Page_45.pro
8c93520f9c670e30d8866eeb13c8d0cf
ba40d035fb1b3333d0e4a16e47aa902a095300f1
6766 F20101106_AABBLY jung_y_Page_59thm.jpg
54a30cb8eb3027d57d208511094973cb
7501b4e2adc330b74bfe2fa986a990ca98b057a2
4966 F20101106_AABBJA jung_y_Page_05thm.jpg
8e9a6c186404510d1033db9b90c7e67c
727d6cd4f44b838ce285f94b0b5c861313f08139
F20101106_AABBED jung_y_Page_47.tif
3885d3e2840c8fedd0a0673367e8f7e9
83b91cea6801cd8a6c63634fd89b95bebebe3336
F20101106_AABAXF jung_y_Page_24.tif
b30c6af17458ee186d743a29d3f34e53
45771d7e8580a5e660f0a8996202b112d994a32e
25010 F20101106_AABBLZ jung_y_Page_60.QC.jpg
56514e0fa4bad40a4ae2438318b68c7b
27d3c2a90c53c5dd6a2318d92710304282fdc558
5627 F20101106_AABBJB jung_y_Page_07.QC.jpg
73f2198f0e56904cf33af1b2c237638d
ce06ff10b65f9fda8b4bcd1949764d09a25f107b
F20101106_AABBEE jung_y_Page_49.tif
70276036626dbed11463c1828b9bb064
4f22b30c87769c2440904f6b33e32bf2a2b49f41
F20101106_AABAXG jung_y_Page_73.jp2
948eb73fb51b0694ac140590bd522e46
dcbfb50354d0e5edfe8487f2c6145e3e6a107dd0
24695 F20101106_AABBJC jung_y_Page_08.QC.jpg
7dc41399b36e50791cc99436c0f2d522
3b8c2b87d461b8151272363eeac276810f658d5a
F20101106_AABBEF jung_y_Page_52.tif
de778399b0db95315e4080a84af27233
031142ad464f137d4f03b1ae08b86e9d94648ed6
2016 F20101106_AABAXH jung_y_Page_09thm.jpg
2f66f020ccff687239b881d0b5181409
d1f73b2038094df1bea3c38dca6df4695a40bb95
6098 F20101106_AABBJD jung_y_Page_08thm.jpg
2bb1879a22771b7bfdd26f759bdfa3c5
ae4969508e461e7d2e964d40e7190834497e8328
F20101106_AABBEG jung_y_Page_53.tif
7b63c67748c57aaa467855630309f721
c17d1522169d66e33b4b6ae29c7eb4097129361b
61367 F20101106_AABAXI jung_y_Page_53.pro
53b2e543025bc557c4c38f9c1f342e5d
fa0f93bb6b5b738ef115d6aed1e7097ad04138e1
7092 F20101106_AABBJE jung_y_Page_09.QC.jpg
cbac68e4b9921188d4917a3dd15f3f85
8d0263e5b7f0f67edda2406b05917db3df41bd3a
F20101106_AABBEH jung_y_Page_55.tif
b96cf7bd29d84a391ecd2a3576a19a63
04097c7f101c26d48bf824dbc71a674021bfed3c
1075 F20101106_AABAXJ jung_y_Page_46.txt
afe41a74faa04e7794fe5ffea04052cd
b2f55faf2c6a856275f34f78c648eb2604e34e26
1900 F20101106_AABASM jung_y_Page_51.txt
a2bcf25e87594ceeff4e9659a31dc79f
96b80958f24659a53a7b01bafbc5871daecc87ee
22033 F20101106_AABBJF jung_y_Page_10.QC.jpg
6124397dded63c5003beb0f6299331fa
560189c04d85a6f64d595dcd021c91e0529ff5d0
F20101106_AABBEI jung_y_Page_56.tif
3d046953bbad058bfed731cb0b2b1f9a
9a2f5eb3833347c55da4221213fd2c9d09ad8492
F20101106_AABAXK jung_y_Page_48.tif
8c8d6a5a50045201e1332a0ed4d96a5c
56d86d86968c40eb0a0108512318a72953922ba5
52775 F20101106_AABASN jung_y_Page_13.pro
e0098ec293b4325cc2b05dd3f80d59d7
7328d7287fc56c682c2bfe8134c16c4d2e486214
5456 F20101106_AABBJG jung_y_Page_11thm.jpg
6e3530de17b703567270c1a0a9aff22e
6316cd186a03f9e83196417acd3aea93eb33400a
F20101106_AABBEJ jung_y_Page_57.tif
26a2baa58bf06eb95cdc274787f8c2fe
f14b97af7974e2bdd721a81af8ef156028420171
100670 F20101106_AABAXL jung_y_Page_71.jpg
9a8bf65fcab38073b222d99c39dd7ed7
c759355068f972d2f6ab3490fa45d562ec4a768d
25525 F20101106_AABASO jung_y_Page_55.QC.jpg
5e9cc52f390b6e82bbeb246d6e9ac630
6afd120819fa928f158f1955d8afd315de97853e
27191 F20101106_AABBJH jung_y_Page_12.QC.jpg
419d8d5ce6ed0ae3b8dd6cffdb5bf3aa
b66c24d93799c7e19ef85f4a199a9940de0ec148
F20101106_AABBEK jung_y_Page_58.tif
f78e15bb1ded376f9b54f981acc94e2c
28a5fcde79a6fe92915431f0f58fb152d8613c73
F20101106_AABAXM jung_y_Page_62.tif
3868c527a97f69393bba19a906da65e3
0e6b61b7dff49dc0c48f379f4721da9262c61fea
F20101106_AABASP jung_y_Page_28.jp2
b41f9131c2a56cb3482cb338a7caefa4
80d0c7869a8454ccdd59eb210d26b883a263a82a
27243 F20101106_AABBJI jung_y_Page_13.QC.jpg
3a97e5b6212bc95551b5fd0845c0c5fa
48df7d3bdc632b32791b720987a303d54aa2bde8
1545 F20101106_AABAXN jung_y_Page_75thm.jpg
3463616471a02f39d658b1c29b0a1e73
785b08ae559b85940f9274ef7e549d0e85f67d91
27330 F20101106_AABASQ jung_y_Page_15.QC.jpg
e6155052b72ad53a1f9f1b5459b23f57
6d472c15ecbc182f3bd78308d91b43f8555856ca
F20101106_AABBEL jung_y_Page_59.tif
99f679a2d2ea1c4c73e2cd9b29fbaf79
a128a0c85b86530a05a997b4159c0b2f2052fa49
6549 F20101106_AABBJJ jung_y_Page_13thm.jpg
0014bf80284c97d628ad68c51443f005
44a70068401782d62eb701bb66be8764f8107e42
419469 F20101106_AABAXO jung_y_Page_04.jp2
82b622edf0b6d7f776866dde8714cc85
15484727a13367472a6b76ac3ed60f185d1584ba
1550 F20101106_AABASR jung_y_Page_32.txt
b72854d54a9e0fdc85152ed0df1ce208
0e246c74da5ce0c984c5ee7f0c36b3b2611b2815
F20101106_AABBEM jung_y_Page_60.tif
2050bc1fa5b187b425a69556b0173e7d
a4752177cc5ce4cf2eec6af36c2042dbcfd7aeed
27546 F20101106_AABBJK jung_y_Page_14.QC.jpg
93c4dbf926232f95a9cbcfbe8ce451de
78a0786cc350bb0140cc80c5cfe24edb4de04b0c
F20101106_AABAXP jung_y_Page_46thm.jpg
6bd5cd78cd14264e31de18164ca4cbbf
c60457c21a7bd7fc5afa87d7f3ea520f4db7baa2
2175 F20101106_AABASS jung_y_Page_56.txt
f150d3f6b3c57ad7f145398a86f12b5f
641d90c26259ce378abc649ecb988fd7f62e8119
F20101106_AABBEN jung_y_Page_61.tif
da9f9e84166437e94604b47b245423de
1157c973addd0dfbeac0009a13d6de2560f1cf64
6584 F20101106_AABBJL jung_y_Page_14thm.jpg
274d2d86708a2e79a79479db412a4f7b
6bf5db12d6734e86e2383bd8c8e8807dfbf265fe
5553 F20101106_AABAXQ jung_y_Page_58thm.jpg
d9fbf15d36518f4924112f5247df9be5
6ccff6af8138d43774ef548f7ceaac8419da0a84
F20101106_AABAST jung_y_Page_27.tif
182f4ec53f41f105ab688e478ccd4155
c8f1db7942ebaaa2a7221a4be84ccfaf7b6b32a1
6350 F20101106_AABBJM jung_y_Page_15thm.jpg
9f201bac5ba176dff76370da2cd49ceb
8d9947e2e4de2ce59633c4aeae33dbc257a0d58b
697 F20101106_AABAXR jung_y_Page_16.txt
af7f300e4b450f41cb06cc2fb6b39dda
c0787d06f753b56ad69f0284c4ca45cc8a78f12b
52948 F20101106_AABASU jung_y_Page_37.pro
9dd831ac206b544c09b2f75a364ebade
5d89742a5008b6aed28c78f97eb2689803313f53
F20101106_AABBEO jung_y_Page_64.tif
10c93096c1b6e4e5c92af0714c757134
8c9a6a80946dcc2a257ab1183725949553fba258
10447 F20101106_AABBJN jung_y_Page_16.QC.jpg
e6b277d33cefe9f8a3209468ec99d725
6f82125a86f0a5b27c0ef7af14beb5dd05d10593
68975 F20101106_AABAXS jung_y_Page_42.jpg
364372f4da200c9c4f97c79c7b42dc9f
ac4d8e2c1a43be38409b10411d3aa8f521462dc6
87726 F20101106_AABASV jung_y_Page_24.jpg
3f900c8643cbd3b3120c5aff2292c94c
7d3ab9baa9b42b2276917a71f152c4ac0129aba6
F20101106_AABBEP jung_y_Page_65.tif
25acf368a87f55d98086bf82b33a707e
bcfff34e9a170bfd0ee0775fdf1f6d1422fa59b4
2500 F20101106_AABBJO jung_y_Page_16thm.jpg
1089b3d5f8353c06fd0e2cd6070431e9
12e1de3f61c10852df31845c49c298cb266be45f
74708 F20101106_AABAXT jung_y_Page_38.jpg
00ae4e5dfeb11d967c0ef221c59a437c
0fd698cae78d8fc2d22a7c70b9a75d246f0d4964
1514 F20101106_AABASW jung_y_Page_07thm.jpg
f5fdd3cc4143820916a68f9f59d14653
728201bcecccc36dd32ccdc213bfc3b9ca745ee0
F20101106_AABBEQ jung_y_Page_66.tif
49328f3c2ba09a6df4c14e9867fce310
5a2627d47562830ca47efd5c08aa1c4fa1cb4aba
24937 F20101106_AABBJP jung_y_Page_17.QC.jpg
9856804519085fb644eb3bd6c62be539
3c7f337ead38fc499aa9e6f988d2766896cb6f1d
91178 F20101106_AABAXU jung_y_Page_08.jpg
3cf38db8ce83a57cd589667fbd4dc26e
80e07895e765499a3aabcd8d5c16514c8d1c6222
1051929 F20101106_AABASX jung_y_Page_05.jp2
fb071c8cdaafc0f0b42a60ab0f9aa7c8
abc47ed5eaab6225317baa56d786c4789ec01958
F20101106_AABBER jung_y_Page_67.tif
f613faadb818ca21063c30fa7badf16c
e09fc3f17a1fec697cdbd8dc56ff3dc208fba1d9
5899 F20101106_AABBJQ jung_y_Page_17thm.jpg
0b27401ce1ea860e471fae8ea588ec32
917c4c7528d0c5a9b5a232aaf1d108ede8e5a46a
16779 F20101106_AABAXV jung_y_Page_70.QC.jpg
b6b3f419f29dc489cbd6d03436467565
8de4c9aac7e129ddc1eda55d444a76dc1355a88f
22027 F20101106_AABASY jung_y_Page_48.QC.jpg
944004b00fc58bd2294809af329f46e1
d3b42b0223b6123bdcaf46abdc24f327fd2a9077
F20101106_AABBES jung_y_Page_68.tif
c7cf66d384a21ebe2c770b43e086d7a6
f9aa2854fa4df596cfa87a3dbac8adfccd759e04
6774 F20101106_AABBJR jung_y_Page_18thm.jpg
5c50d4921665db619fc5691f93ec4941
edec27da65c0b54a24b080c340b10c40931a066d
1431 F20101106_AABAXW jung_y_Page_50.txt
0180c938e291f94e31a5d5f44489cc99
5f5b4926926189c40d03c0ac398e671687cad4a8
82260 F20101106_AABASZ jung_y_Page_59.jpg
d5fb123e070ab65222a2fccda65ca3f2
e831076b1c7b67896b145a344970b011ad2cb464
F20101106_AABBET jung_y_Page_69.tif
1f6d46fbf9438035005b642ab75a0667
6d9718e5db03d7ca644f5b8e5b65844662c65822
28070 F20101106_AABBJS jung_y_Page_19.QC.jpg
19371e615277afb9bc994dfd59d20b37
d6c979c68c65c56bc34b401439240f68a5c4248d
1800 F20101106_AABAXX jung_y_Page_52.txt
aa278199556425ed23a8b16c2542e82d
ff85c09e7ee39e8666bbc1d4ab1770b2d8fb01b5
F20101106_AABBEU jung_y_Page_70.tif
d5511217f30a2419d9d5732494d479ca
6265d9ed7c9538fb47f75df2d79d24d50bd132e5
17367 F20101106_AABAXY jung_y_Page_16.pro
536ad063aa41144709a101a98144181f
42a5b221923942d9de48235a97120fcae26c8287
F20101106_AABBEV jung_y_Page_71.tif
802ab362c5b4cf47057ad3115404c4b4
1a28fc60f8925d3ff3826ef1e34fabedebe91c3b
6383 F20101106_AABBJT jung_y_Page_19thm.jpg
3548d6d58cfeff47edbcaa983d607e39
f691f0ba3ea6d438a3a8ad2ed04f7c5a68d8bd73
28073 F20101106_AABAXZ jung_y_Page_66.QC.jpg
6703babc5e33a38a9c9cc2b2743fc0dc
ee62aa85245f69a1ab10f8d6b56bbb58bd38a9a6
F20101106_AABBEW jung_y_Page_72.tif
c3591dd76d7dd6042351568fba3abe18
9fda2368f7262576160f9ce634fceca2e1fc4cb0
F20101106_AABAVA jung_y_Page_50.tif
daaba5401dc486b58b3ffb3b3a017222
3afa4d37bb90e61a3d14741b1dbed6a80a451fbe
25577 F20101106_AABBJU jung_y_Page_20.QC.jpg
4277c5472e7d747477427c6f7caafae0
33d40c0544a38c6270d7eb7551bcc96f6f3af8d2
F20101106_AABBEX jung_y_Page_74.tif
1085b64a4d00f3fde04717ff79332af8
2957e71afb9f19728725892bedb69cac22b84b35
79100 F20101106_AABAVB jung_y_Page_51.jpg
4c6ba16eec2d5fa1a10b9526d2dd6d74
388c027867a3584443718630fe762ded00310c94
6234 F20101106_AABBJV jung_y_Page_20thm.jpg
cb084cbef95e072d21a3cf58482bd2f6
4e3b590af06b7311956e53c655859d4efaec59bd
982441 F20101106_AABBCA jung_y_Page_54.jp2
a8c88b3d339defe3bc5221793166700c
feefbe854a9c2c18f6bcc873cb837b1e9c508e7a
F20101106_AABBEY jung_y_Page_75.tif
56f4c54cd7ea06f034fa40ff67933179
badb7ac720df45d4ca4fd69cbb45b98113d02d77
19720 F20101106_AABAVC jung_y_Page_25.QC.jpg
ab25b4d6aa2badc83909c94a63567b14
6d54be4ccc20e976072d7ee77d81050e388854c2
28066 F20101106_AABBJW jung_y_Page_21.QC.jpg
70f2d3737d4f7c9ffdbec32fe77f3069
b71bdaf4453c2e0fb86811521e054145a0e08b8c
1051926 F20101106_AABBCB jung_y_Page_55.jp2
98dc364be97fcaaace745031ab655eec
9485588257df84cfa8f1a559dfa273bc64a3e794
8276 F20101106_AABBEZ jung_y_Page_01.pro
2d5b5a1c1466982703e88db90522d537
540d779c09e5e7435fc3f98857f1514d850ada1e
10923 F20101106_AABAVD jung_y_Page_07.pro
3c7f8af1e7a0d4d81992228bac82bf17
2c7aef78b28a09395c0783216da293e353840d85
6724 F20101106_AABBJX jung_y_Page_21thm.jpg
d8b5a21857aeb7fd645c6b4ace142084
dcaccd453309881998b312dfaca52de4e867ac78
1051972 F20101106_AABBCC jung_y_Page_56.jp2
41c8bf159f6e98209da13f708b8f6972
8ddb10d7953ff1824087fb95133f1de0613fe735
25477 F20101106_AABAVE jung_y_Page_36.QC.jpg
7ba47db46b55f57c98f3f042327559e4
40b12d031c2d985b3b30931c89f653b36dd24709
12344 F20101106_AABBJY jung_y_Page_22.QC.jpg
343c9986429053366d675193110bf872
f323ac8531f78125942ef1d24c2ff6f71ef521ca
1885 F20101106_AABBHA jung_y_Page_10.txt
701c614dfadec69a54e9818dc323de95
045c5bc9c520e65f296830752c56453af2319962
737606 F20101106_AABBCD jung_y_Page_57.jp2
5b1047643e11b5333e63741ec82e6ba7
c013042edcb3d7953744b4eb41689db49177ea31
F20101106_AABAVF jung_y_Page_54.tif
81a86a32bd00ce1d2292e38a9263f4b2
19b50ce3dc4a420efaa363213a33f5ab1d1cf2ac
3067 F20101106_AABBJZ jung_y_Page_22thm.jpg
23863832b65342c9db64c44366da3c88
ca9ea7f84ca896c69ceb4fd88131501065a0039c
2124 F20101106_AABBHB jung_y_Page_12.txt
f0f4aef8bc2927de82ebbfa6680d9870
c8da147b788cfe27691f6cd624178a374945dc72
986796 F20101106_AABBCE jung_y_Page_58.jp2
7aabd06bf69d36bcb2631e651e146c09
f45e8ab65a1d9d016b38354ba4f561092e3336bb
25222 F20101106_AABAVG jung_y_Page_33.QC.jpg
22b570600ed8387ce297e7de5e4da7ff
ed9fe7bc44a81fa1d689fec16fc7f190e9bc4430
2107 F20101106_AABBHC jung_y_Page_14.txt
952667232894368a2de1e74c6c77636d
7ecdf9fe07ca545bbd68b6b479a64eb755e0235d
1051944 F20101106_AABBCF jung_y_Page_59.jp2
6c589caf4fbe4b5ef6e0a11a7da9390a
ca745b410dc1db6cc086b456dfc8928dca53a71c
43065 F20101106_AABAVH jung_y_Page_38.pro
bd49fc51bbaa7459b7646ab655d19708
237309a808563ad2ec80e08bdce7b24b1be8a1c1
6006 F20101106_AABBMA jung_y_Page_60thm.jpg
59f852610803f318e74120604a63bc23
4220f308ca7600138db88f106a11e7dafdd5f182
2070 F20101106_AABBHD jung_y_Page_15.txt
7bc68366fe51d65cdf7036092e462fc7
ec8aa500d34746d3b097c8d8d8a6c6aa951ea787
F20101106_AABBCG jung_y_Page_60.jp2
685baa6160745d88795737150e588d26
d644cd8f275b84bcb70b77fd91498561cefe27cb
6179 F20101106_AABAVI jung_y_Page_23thm.jpg
7aa08ae579d68bd748746a81442be84e
23874619d804a33db2f41a3e01cde98ea62585a9
17449 F20101106_AABBMB jung_y_Page_61.QC.jpg
889b6df066c36055ff3c55617d9e2fb2
440715e724c817f724dc96c5ffb1eb879a50c6a9
2102 F20101106_AABBHE jung_y_Page_19.txt
db24e7a39476859dcac8bfa70c60a52a
e1703459a12216889ff5c3a95f14a2dbfda8ef63
1013836 F20101106_AABBCH jung_y_Page_61.jp2
bae5f8010eec190287b8d60bd8d70e1f
2a7f645c7ec381cfa7dda059888cd06d7da28aac
27522 F20101106_AABAVJ jung_y_Page_37.QC.jpg
fab80f67f57c8609156c137519ab0c7a
aef343c57ddee775df158820af822df7acc17b15
4951 F20101106_AABBMC jung_y_Page_61thm.jpg
2b7fb6ce164c4694ecf23ce142a8c2cb
053ec5265b0c8af278898ef46320a5d177ba038c
2139 F20101106_AABBHF jung_y_Page_21.txt
68d652aaf94dfd338dc74903005caf0c
07a1b8d13ce6660b4311215d89cdab07b2d91fa8
F20101106_AABBCI jung_y_Page_63.jp2
0cda13d2d8b226f58e5bb2b856fdc7f1
e9462319bf0e263bf9e0e254d9d096e4c9023e2b
25659 F20101106_AABAVK jung_y_Page_35.QC.jpg
fbf9c5ffdfaf233545eea4213c8e3dcc
65dda0db30985f123a4758419a36ffadd2138ceb
21098 F20101106_AABBMD jung_y_Page_62.QC.jpg
b9630cec7ae40f0b36865bfeab017530
55733b0f316cb7141e9d21535e4b22805176cae8
2087 F20101106_AABBHG jung_y_Page_23.txt
2969e27fbfcec0fecc6e6e2e2a6da7e0
d0337cfe187bd4fb043b3d67aff579b67a8a03f8
F20101106_AABBCJ jung_y_Page_64.jp2
06e27e71def9aad0af459a6feed1af9e
66ce9f5fa3ba70db809644c3e3770f578c62b662
1929 F20101106_AABAVL jung_y_Page_20.txt
0687a2d071e53b6939dd3e5562967cc9
723310c522c71a664c377a30b32bb0cb10bbedff
5465 F20101106_AABBME jung_y_Page_62thm.jpg
fe9bee16a192a305cfb72affa78c78c8
3c2541fa1385db48e9d43618364668f7dc0170ba
1051980 F20101106_AABBCK jung_y_Page_65.jp2
3ba139b581d355409e518b9ca859dbc1
d38aacc377248b19b115d80349131530cc04230f
1963 F20101106_AABAVM jung_y_Page_17.txt
77f986ee006c4488c5020b44daf5cafa
edd2aec8bb5c3f609322605d3732e1b3e35ae4a1
2065 F20101106_AABBHH jung_y_Page_24.txt
0954d358d5195dfcaba5102b6b31c6bf
fbb215398839007a7fd85b896f5056e2bbc07617
6692 F20101106_AABBMF jung_y_Page_63thm.jpg
4a68dc5b08ea975426af923c49ae678e
67020785bc375938afe2df55df2eac8a89a0dbe4
1051977 F20101106_AABBCL jung_y_Page_66.jp2
6b5d0520564e1af5eeb91c7323898848
064d61f37a85cb13c2fb22e3c95d2caf971e9b4f
43180 F20101106_AABAVN jung_y_Page_41.pro
5f6f86f06f1e5c2737fd906220cff5be
88eb00dca833c52d671d21db2548df31f8f77b14
867 F20101106_AABBHI jung_y_Page_25.txt
76faf23b91ec37a1cfcd4f01e4aade4b
50a3e4c3e6f1ad7b1efc68df179bb411ceaee371
29264 F20101106_AABBMG jung_y_Page_64.QC.jpg
c10a152e8e3fa687690349f75b54be45
1dc244da75816febca51e47d73e6e2542bdaed1d
26362 F20101106_AABAVO jung_y_Page_23.QC.jpg
2f653b0e6ebffc9dd93699dba940d784
e8ce65dbe5033139d91cd3abd6ffe0ae4df21318
1707 F20101106_AABBHJ jung_y_Page_27.txt
477b8c4909ff219f46bcba1a23a84890
d3f5f279531a841d5ccbb76da86c0f8385c4cb8f
27015 F20101106_AABBMH jung_y_Page_65.QC.jpg
b14adeaaf0ae45021a407ddd82f65784
6943814f7326fd41bade1f4ea2781540b0a75c5b
1051910 F20101106_AABBCM jung_y_Page_67.jp2
3d3054b4920b783d945a10fe04b8dd85
0a1a7f37b17502a050d0eeba00c8676f20c5396f
562 F20101106_AABAVP jung_y_Page_06.txt
dd38aadf32151a190f2ad4bbf842db61
a3f07424f482391d5a5624443db124f427754d14
2536 F20101106_AABBHK jung_y_Page_28.txt
2cbbbb0fe9010664d9c98fec6c39aeee
4b74b7609ddf16b14817c8283993fba7d6d4e05b
6508 F20101106_AABBMI jung_y_Page_65thm.jpg
516c26576fc668221d9f1f98a5fb43d5
f7b3e0f479f238e81d7eae48cd64cda0df7b6350
946156 F20101106_AABBCN jung_y_Page_68.jp2
29fc0b8b85b7d3807cf4d3dedd62516c
437f8aae0a01e41c26d7f2826125274ea0c149b0
F20101106_AABAVQ jung_y_Page_26.tif
3cf5ad5a71ba81ded371d9efc2ae613d
842530cc1d351002588a68a46bdd05f6b242ddfe
2619 F20101106_AABBHL jung_y_Page_29.txt
95a433acc116a46792ce3fbc4ea64016
8f7e7986689171641e17c40858361af59ba872a2
6007 F20101106_AABBMJ jung_y_Page_67thm.jpg
4a37b9321f9d17bed1f6e71164dfbe5c
6181baadc3bf140c5b2d7d3bb8396615ea407a97
1051963 F20101106_AABBCO jung_y_Page_69.jp2
f901019c6db6a6509216d260cd703993
73730c2f87afa21ece86bce1fca1afeb30f8b24b
1051945 F20101106_AABAVR jung_y_Page_45.jp2
d005ea9422d2f0fb03ddbb1f4c1d861d
d4f1c058bd759fcfc78526c5c2d28379396dfd04
2316 F20101106_AABBHM jung_y_Page_30.txt
9b72a98f3b67d6fd70354c0e79e47c11
a9d8c7d7a54d3bd2fe79a5d57523158412e71939
21434 F20101106_AABBMK jung_y_Page_68.QC.jpg
0c1ca32c662e0d7df5eb3e5dc3cbad5d
2ec86c271e850f68eb721d69ffeb37ea27ec7c9c
705506 F20101106_AABBCP jung_y_Page_70.jp2
4be42e6dfca724a62c003494dfeb2cf0
ec5b019ee08c1affed8cb87215a5d8310e20e14e
53896 F20101106_AABAVS jung_y_Page_57.pro
c4e8260b1002d168f5f5510aa1dc10a8
3e3645475a36f8a65e1d0184f84b9448f61970f8
1957 F20101106_AABBHN jung_y_Page_31.txt
2be241b226329d31dab65e3a1a420a58
7285ab4df6f48714aa52b48a0d31b92c5ff094af
5248 F20101106_AABBML jung_y_Page_68thm.jpg
e8481c0d6af6c144cccf3d2a2b6d2f35
15ad0cd6bf89714a4f6644a8cfe1cc111a1b8b4a
F20101106_AABBCQ jung_y_Page_71.jp2
36cec7671082f5d9dacf9b6fb1534d22
80c5366d6ace69c4ed4eab4820fa2c05296a9b9d
F20101106_AABAVT jung_y_Page_51.tif
64eac3956b120fcfd9554ed31585db7d
94ecc62cb0d1d81ca45d58266985571e2e8b0ca1
1848 F20101106_AABBHO jung_y_Page_34.txt
72f33364b8659aafb5c9bdc467dad9d5
c85c429caa464cfac36a85d45052a547994dab68
6749 F20101106_AABBMM jung_y_Page_69thm.jpg
319ddf28bdd9041175d2015a7a792064
ed4646be79701de6c8b12f93d298db82d5d9fb9e
1051969 F20101106_AABBCR jung_y_Page_72.jp2
6e5c3280dc6db3a344dc5042242cde1d
a193123f41e29937b3532a59c842b6f264ff3a29
1662 F20101106_AABAVU jung_y_Page_01thm.jpg
00dd301a69112022690c3ca476c7f0da
b46a36cdf83b5ada43e3cb03760786aff4116dc6
1246 F20101106_AABBHP jung_y_Page_35.txt
2b68b374ec5159fa7e2efc013fdcb188
155f9ce18d1294a2f63e71cca86078b15568b63b
3969 F20101106_AABBMN jung_y_Page_70thm.jpg
a3e0d49d67bb45902220adc7c1ff4774
892cbc38ec840919c3efd388060f7792a0b52591
869549 F20101106_AABBCS jung_y_Page_74.jp2
c2b9f75acd4cba42c8613d557066b616
ab30946776adad19af3607524ff00d7b6e46eaa3
81704 F20101106_AABAVV jung_y_Page_67.jpg
12d262a0f03e4713517e835d37bdfb57
3f9b244f367012e9abe84eb5673c601223f0162e
1019 F20101106_AABBHQ jung_y_Page_36.txt
75f7d7b63f119a770d119a66e3898360
cb652a15d9b94dc0a2da96d4975fc93c8afcbb60
28460 F20101106_AABBMO jung_y_Page_71.QC.jpg
19ae2c95043387aa6443c3dbd501a0a9
6067d86bfe8d8746026d752fc9bcb9c235535dfd
242319 F20101106_AABBCT jung_y_Page_75.jp2
fac34c52da07eba16598f73eb3badb08
b338d8d87cdea1b9af1bb33eb5729839e6ef428d
25560 F20101106_AABAVW jung_y_Page_67.QC.jpg
f83027e4bf0d15a9e56694c8764af3e0
60c634ae49d3ed27c0c40ba472fe5894bbea836d
7001 F20101106_AABBMP jung_y_Page_71thm.jpg
61db75773962d7e629ea48157b24fad0
cc815b47862463384c11b702eb2ef636d3b1b8b4
F20101106_AABBCU jung_y_Page_01.tif
9e8bc1dcdff6a5fbd04203fc6c7469f5
7f49f0f50501a7a28261cb7981d4ac04c40504e5
2071 F20101106_AABAVX jung_y_Page_63.txt
d0327d447b1eb7e2bc48741e0b87bdbc
d2eadcabe0c4a87fac203d4ce6b1255b0743cfc6
2078 F20101106_AABBHR jung_y_Page_37.txt
262d9ee7c691cd732d3ca291350ae930
7cda568a203815da467d3f3aa551d8e20dba3f45
30737 F20101106_AABBMQ jung_y_Page_72.QC.jpg
a5205cb0766c14a172029629e5691042
824b2d77b007ab367aa5401f7ae97e6ab3d71456
F20101106_AABBCV jung_y_Page_02.tif
7247895bddfd32610eb4b0512558f5e5
8c8ddcd5c1b256bfd096469eaa6e64ed4ba4c6a3
6655 F20101106_AABAVY jung_y_Page_64thm.jpg
b165ba88b9350ac0773f80c88ebc6674
c2586c96f8efdb912f231c0517ba7c3e9851c58a
1711 F20101106_AABBHS jung_y_Page_38.txt
c70a9391f4a78299be3955caf1406ce7
3001fffbd897919962f437825c99e0b3c5b8a1a3
7024 F20101106_AABBMR jung_y_Page_73thm.jpg
39e05340dc510c60750f2556e730ca27
f0e3e413de2bf65ca54d9189664bb1b3b21142b7
F20101106_AABBCW jung_y_Page_03.tif
c8c039723b9bc64014cef0a922bc3497
b37415709471edf219f7650a78e8b5de03d53277
6219 F20101106_AABATA jung_y_Page_43thm.jpg
56e6f25ff6338e1975d05914e4318896
b7e7872e7d857bec0d57419707ef95c37ba759cc
22557 F20101106_AABAVZ jung_y_Page_61.pro
f0ef2419244586b67b75df1a690fdaec
85cfcbab6751ea53435905f74dfae37e2f74e39f
1818 F20101106_AABBHT jung_y_Page_39.txt
3f4554e2432098a3b9945b0aef6da8c3
e47f4e5231dd11f9f70db2debed758f96fc4f9f5
18638 F20101106_AABBMS jung_y_Page_74.QC.jpg
4eb171a4a7f90ec472b2c782a78f8d00
9e5a0513c8ff436f96b2ee50a89d1f134412cb36
F20101106_AABBCX jung_y_Page_06.tif
5cddddec81e819168de31778836b9306
55842b343247e0cb1f53eb4fc76f8492e308c8c7
F20101106_AABATB jung_y_Page_23.tif
ae62b839cc4e1a776353cf5afacfa178
ef9bfd60c34b9a366bff3a052e14b2ae90a462db
1995 F20101106_AABBHU jung_y_Page_40.txt
bff9695dbbf908b4bc8d8775b5f43ed7
7c466332bc2261e43e7bd606f1a9f367f2bac670
4456 F20101106_AABBMT jung_y_Page_74thm.jpg
f96e8fa5653a5f79433b4ba68320cc9a
921dcfefab9f8faca9984de051259f5e26223b39
F20101106_AABBCY jung_y_Page_07.tif
cab121ddfde6f81666154ccb61a370cc
3d92d340bb5fae943c7f8b31b167ec4a32c15ecb
79528 F20101106_AABATC jung_y_Page_36.jpg
efe8ec70a87c11fbda6756663cb39f48
be082adcbe6ea2d3ea9b6c643e42f27dff554723
1720 F20101106_AABBHV jung_y_Page_41.txt
639bcea1fe153daf48e8b83c6392af58
a7711cc85eaa5a23ea823dadd6faf091b72ab445
1243 F20101106_AABAYA jung_y_Page_06thm.jpg
22535ab5860e3d29e099011787b4f886
f15251a44ff9a7133dc7ff9667283bf061ec3c4f
50336 F20101106_AABBAA jung_y_Page_57.jpg
1cb90ad52e2b38ea5570e0b1cce4ca63
b80e70b637c3306c7cf2ca3dbdcdd5323454ee1f
6302 F20101106_AABBMU jung_y_Page_75.QC.jpg
c3331c32373bcd56ce731e47c074aab6
009120cd667c93b33623b7f844ff14a716099c46
F20101106_AABBCZ jung_y_Page_08.tif
4a7ec80acb6137bc9476474ae29d2f1f
647f05da581d744fc7e19c41fc3741f895c41a4b
3815 F20101106_AABATD jung_y_Page_06.QC.jpg
5c5bdffba726ef69712f2e3fd544e2c5
248c9533b0934db6b62a3e9e3af10d2eb3d1c6de
1523 F20101106_AABBHW jung_y_Page_42.txt
5c1741289b3b0aec3640b38a1c6622fb
cc1834ecda54da5dd0327a040a605e151f53a202
7010 F20101106_AABAYB jung_y_Page_56thm.jpg
bb1d67580fd24929052b2be97f6a2c81
cf2ab9e5796cac07d08732a2df7e7c907b674d70
70416 F20101106_AABBAB jung_y_Page_58.jpg
984d8dbab9067dfe4ecb4c08be366754
b80f4be2035b22b997e98a7d883bd2b5093f5e5a
87598 F20101106_AABBMV UFE0022204_00001.mets FULL
1f0fcf05e39434fe97276fc8c35f0a28
f9a848cf205bf7176fcda1c67840de31d316c82b
995250 F20101106_AABATE jung_y_Page_10.jp2
f92b4fc3b8a45154b639c63fd948e729
03f1af1ad93bc8b516c8a97362ebee4e80aede99
2006 F20101106_AABBHX jung_y_Page_43.txt
4d3f5bcb32a4b0d7375c568a82f7c7d0
a746140b0130c8e87f13c25a0720ace2857a5adb
2077 F20101106_AABAYC jung_y_Page_13.txt
ccf4f33f8469d6b9a7f5b2eef7ee0666
490a26d915e20a17dda2250945b393fef21010cd
80538 F20101106_AABBAC jung_y_Page_60.jpg
892dc91253495418ef9f3cd7a93feae3
620ab5ce2c98eb35c715eda9d984b8f2998aaa64
781 F20101106_AABBHY jung_y_Page_45.txt
9ff91a619d34755112fca5af4279e5ce
e95600b36d438bef7055464ded1cbf2fb7714445
71184 F20101106_AABATF jung_y_Page_08.pro
63660500e9afd27a20d85d71dd1090e0
1f2b9e9a9bb2c66819e6ea84ba092bef687b29dc
780 F20101106_AABBFA jung_y_Page_02.pro
69a85d097bc60a429cbda639afb863ba
f8c659dd69b09a5ae783ec8da30d5ea42f043a4d
57551 F20101106_AABBAD jung_y_Page_61.jpg
64da70e0dc4430d8118a43d7885f49ae
24b1f7cf8471df4fc6cc6d6b40789e97c38a330b
1618 F20101106_AABBHZ jung_y_Page_47.txt
3445ed4dac6bb22ca81bb8d54b663c87
005f5dd497e310930fec14e68c9429f3c704ee3a
5453 F20101106_AABATG jung_y_Page_10thm.jpg
bbb4449ac22d256711ba1a69eb3ee4ab
4557949344a70e00a075629c6c1fa4d1b9481c48
1428 F20101106_AABBFB jung_y_Page_03.pro
3e58209933c7c90f2a3fe445988f2f41
ddc4830a676893e28f3959723019f69acbfba605
53962 F20101106_AABAYD jung_y_Page_67.pro
78b30c15fa6230a2e5f8063be426a3b3
7929ffd8e23fa549879f2b96548be507047f6e1c
86384 F20101106_AABBAE jung_y_Page_63.jpg
49b990b57841b29fde7eabb96a1d45e3
c1bddb90876511f98e910859a4fb27bf2069ab74
58199 F20101106_AABATH jung_y_Page_71.pro
1270e900d2561953587d1eda40b5c07c
4f4ac5f1a337a63fd086dd670c77f0e3108accc2
10708 F20101106_AABBFC jung_y_Page_06.pro
af42e26dcc833a7c1afaeb11c4cc6d01
54539b9e97a7b13be4c5f497c2418e7032c6d80d
5504 F20101106_AABAYE jung_y_Page_38thm.jpg
0710035ecdf9c62a1198f4e61da48069
420c56ec8b77376e8d629e3b02c14da3e3f33010
90687 F20101106_AABBAF jung_y_Page_64.jpg
a14cbb293dfb69d529eb136c9ad7ac1c
99a829d76d5b7fdea3047d9f49f295f9d9065340
26984 F20101106_AABBKA jung_y_Page_24.QC.jpg
d20fe8ce4583e290c43bbdfddd97c374
c15e945feac39d2535cf3466602f734a2763d81a
1040092 F20101106_AABATI jung_y_Page_51.jp2
4b5867812d36f517ec465f43f247e0a6
ac3b30e0ce66075aecee819b2f1cd58a6f542528
21934 F20101106_AABBFD jung_y_Page_09.pro
d8fb22c3debebe6045a70e182478a325
9954394060955210b3f45a94169e92885f8ab8c5
F20101106_AABAYF jung_y_Page_10.tif
9ae0e5d7d1f4f402453a003f37f0e2c3
9991537a6bc4cc0fb99b65b8a5eccdfdad00ca5b
88172 F20101106_AABBAG jung_y_Page_66.jpg
a20c69b869c34a766cb85db1055e0e19
818ac510d00d1a7a6fe5a437a85b21d41188b5fa
6497 F20101106_AABBKB jung_y_Page_24thm.jpg
4d9dc2aeb43b987bdf591da8bce66bff
bf254b14d342f5b8b5b4067187b5f6232a0c22e2
F20101106_AABATJ jung_y_Page_33.jp2
7b052f423612a748385987b5521e906d
6c8dc3ec765eb34979bddbbeb084fc47b93b714b
42520 F20101106_AABBFE jung_y_Page_10.pro
5ff751ab9af399762081c08fdd54ae12
6ea01c0d9219aed2cd9b7483f54afdffd9ea1c13
28439 F20101106_AABAYG jung_y_Page_69.QC.jpg
e5c7bad403f0bb9ee2b44ef0b806f1dc
2b16fc0a607bed132c98b9c323f38534286ae70e
91277 F20101106_AABBAH jung_y_Page_69.jpg
923792c3c76d971bb5f655d800669d07
b761c8c1662b54179fefef29fecff88c9f973c9a
5001 F20101106_AABBKC jung_y_Page_25thm.jpg
1d601205dc9c320fd1986f731d42fe8b
c91dec584a54a09e339a2ea9811a9b091da38dfc
F20101106_AABATK jung_y_Page_73.tif
6bb2f5f42e12e930a6c388805ddc815f
8ba7d30829ddcb9bf4c4d5cada37f33c789439de
42441 F20101106_AABBFF jung_y_Page_11.pro
853a6440351be6643e3347dced9efa69
e466c010f44568b3c2adc84325f7111eebb87eec
24565 F20101106_AABAYH jung_y_Page_53.QC.jpg
ab8843f01ffedbbd468ea6ea89d9a1ae
be32eeaebff14813979d8a068908470e2e217f69
53737 F20101106_AABBAI jung_y_Page_70.jpg
1e1abcfd190f977287e6a3076d3efb42
2ad99b48bae8bdbcb3439c559de04392c4271325
27625 F20101106_AABBKD jung_y_Page_26.QC.jpg
4b071e18b174fa8456032dc5abcf9505
bd2ae0d3068ae6200d72abf9563a2c0ba31dd6ca
F20101106_AABATL jung_y_Page_04.tif
3be2fe5fcc5e1e392286dd4d123400da
062ffb95d4ba8ab3f513b79a185dd444f0823468
52017 F20101106_AABBFG jung_y_Page_12.pro
a4822d0c173eb0e6164e52cb727377a5
28037deb4a7bb71a2dc5a776c62b2a9fd9cebc08
F20101106_AABAYI jung_y_Page_34.tif
7e6dabc60316d9d2de97f03c28529e7c
e2a3952fcf862fc0bd1a7bd2ab69f6f266eb8ff3
100444 F20101106_AABBAJ jung_y_Page_73.jpg
0f7d2792d2e22c35e4cacf3e380b5479
935401e3d730fdc79ece1ca0de58cca933207563
6527 F20101106_AABBKE jung_y_Page_26thm.jpg
b7a467798c8b85057ee90974859fcbf1
8cfac522ae02a04ccd33053c059b55a2c6654ca1
21440 F20101106_AABATM jung_y_Page_41.QC.jpg
33ee88cbe8a983dd01bc5ca9022e7e2d
3f9eb2fee8ef1a6343257cde3a5285a13d7a8de2
53466 F20101106_AABBFH jung_y_Page_14.pro
f8c3c86a96b3091dee1054f7cdc9f860
61f40ddf904c9ca8406c209fd83bdbf56f7836f3
112949 F20101106_AABAYJ UFE0022204_00001.xml
6a5260350e0c6e1e959810c7e693b921
30bc6095b2431542d2e2196f585463ff1464e5e0
22152 F20101106_AABBKF jung_y_Page_27.QC.jpg
dc128106143b6d8de673e9262b2d4354
333b31036e87e864cc20f02dda234efd693cec60
1051985 F20101106_AABATN jung_y_Page_42.jp2
f4651f5c64466e7b93f376eb85477cb0
bda5a5fc0da1743b61735581c0758d897eb9f357
46954 F20101106_AABBFI jung_y_Page_17.pro
d9c10b3a54dbfa2a1dfd4e63728a2f25
4e265a4e872884bfe6ee534d2f71c5dac57c3466
66383 F20101106_AABBAK jung_y_Page_74.jpg
1b0ca830bb614ac6478d93b38f5b11a9
f99eb292146a6088e0b559fc695900ef0530d2e0
F20101106_AABBKG jung_y_Page_27thm.jpg
5344a0fdb9f197c04762e2ebcf1044f0
398c26377aa4ae6749e9d1284c7304c9fb0b6862
28192 F20101106_AABATO jung_y_Page_18.QC.jpg
229104e7c92520f95e1d1172b14aea4c
ef6980da99e882ee576830835a682a4321a00ab9
54294 F20101106_AABBFJ jung_y_Page_21.pro
abce4e4cf794291ecfbf7c47dad51e50
7f51a21cd5437f282b130f68ed6d7ae2edacbcba
20853 F20101106_AABBAL jung_y_Page_75.jpg
a02dd4c38d2e728976227b6ec0f14f4a
b79649f36dc3d9b3ddd888b3259c9d1e96b78491
24717 F20101106_AABBKH jung_y_Page_28.QC.jpg
c6fad51923cae5db1923d554be19f828
3c409f4798a21e6a8fb89f305371e4d5eb8788eb
22530 F20101106_AABBFK jung_y_Page_22.pro
f07b2a4f1d6e852113fe7764f776d72d
8e8ece4457e307b5ca79cb14c7c89195a72201d4
22250 F20101106_AABAYM jung_y_Page_01.jpg
8b1677141cc25739339c04907ca0f60b
e8d6510bb469fcd0237a77b6cd5b6d0f02bd1d26
254446 F20101106_AABBAM jung_y_Page_01.jp2
6114248ffbd7658c6fb168ac19044ac4
ebc544f0ec8871cbe0643c5d1b18b39a50171d19
2094 F20101106_AABATP jung_y_Page_49.txt
8e820908ef8eea7395711030c489a6ac
6d077f7b5f932a691c77e4e30ad27346d76ec05b
6282 F20101106_AABBKI jung_y_Page_28thm.jpg
c542f1c6b9b7ec6c78f4fadd7f80f0cd
f2d28fe5162d7236025c3834e7f1c7f997705551
50895 F20101106_AABBFL jung_y_Page_23.pro
788625c6706efd48996c371d50837aba
1507cb7815dec42756bff703b258383bffb28f61
3308 F20101106_AABAYN jung_y_Page_02.jpg
91bafb11b693c996aea753aa3ee39bc6
d039759e769e42c47d4281a561e463a6d232b945
24028 F20101106_AABBAN jung_y_Page_02.jp2
296d6a94677f1ac3fde7b16f16bcb7c2
c82459471e81325f862aecdef0fdc7a80edb90cd
5266 F20101106_AABATQ jung_y_Page_47thm.jpg
ca15e32ae3cbba50707d25f9409d9756
7d7c35d5980e4a60ba545b763548d143b2b85305
21616 F20101106_AABBKJ jung_y_Page_29.QC.jpg
5d3387e32b9d05a650807c05217997e6
f358acee6ed3f64555f1d5250cffac493617625c
51607 F20101106_AABBFM jung_y_Page_24.pro
e26a77747ccdcf395dcbe56632a27b87
21d0b9b27ba1e28d412ed53c26d6f9497bb02ada
4670 F20101106_AABAYO jung_y_Page_03.jpg
234b855ca3ebd2566482c6ec8934b833
aa03723f2b61d148699cb9b54e17b0dec4804dd5
37745 F20101106_AABBAO jung_y_Page_03.jp2
7bfea4c86be84ba44acb3b124a264cb9
6ee4fee8a5a8bfc1e6949bb624afd9dae69775cc
72375 F20101106_AABATR jung_y_Page_10.jpg
d680be3772de906fee3ea5ad89c180fb
a0d56a0fbbdbdfec26dfc2bc4c0cb961554e8cee
26073 F20101106_AABBKK jung_y_Page_30.QC.jpg
4b99b137f5fab7c44aeb4abee279e4cb
4e1aa3aa941b614418a65aa6817309e4b057f4bb
21667 F20101106_AABBFN jung_y_Page_25.pro
a78b15f3eca79e78a1b5674375f7d6a8
d00b106b18dc448e2f2244e9ce994a8bee7c014c
93287 F20101106_AABAYP jung_y_Page_05.jpg
9ed985584e629d267f70ca1ce7631fd8
4331ca4b76785c61d0c8780fe4201d681f0dd27e
343762 F20101106_AABBAP jung_y_Page_07.jp2
691277e87fbc52115f04a5faed3dc0a5
a316f40e6f74aa9e822f6925a86fd85419d083e5
73858 F20101106_AABATS jung_y_Page_34.jpg
e1c603a2c5b76d644f8e1f3942ef1142
53a9515fb0bd45192776c3c2e48c543236699cc0
21928 F20101106_AABBKL jung_y_Page_31.QC.jpg
66d2fae7ceff9809ffd5de9a810ae43b
a61902bd612e8f45ba175ca649225419774340f8
52824 F20101106_AABBFO jung_y_Page_26.pro
9ccc7718fb78c76267fe87d2f3d33e91
7b7978c7d73d6ac688de53a49b4d3dda1c4574f3
13890 F20101106_AABAYQ jung_y_Page_06.jpg
8de2a6af3d58f1b135512cf15209451d
e20ca2e698dde1366b19d89624cbeba6773de8da
1051983 F20101106_AABBAQ jung_y_Page_08.jp2
4f5eabeef478d3e1195ef21514fca8b5
78ff5ca07537c5330ba7e83a0c7e5c63dd0fd3a5
18240 F20101106_AABATT jung_y_Page_04.pro
7464da27e1186bdb4dbdd07929595481
5b660238a5a2d92b4360350cb24bd12b7e515a4f
5853 F20101106_AABBKM jung_y_Page_31thm.jpg
00d3577352f0ac16920d0fc777a7385c
aedf2e6765c085e811b94507b147d74fe0e0aab9
17790 F20101106_AABAYR jung_y_Page_07.jpg
6acc0b38827ce41633aad24874a10452
b22945a59aae480ac261a0c95699a889bd449416
571785 F20101106_AABBAR jung_y_Page_09.jp2
eeca792f70760973c7a7cfed5386e78a
8f77fda0e975de64691e4a1b97605ea5b9485b0c
5839 F20101106_AABATU jung_y_Page_29thm.jpg
398e52fd8ff0ac4d7629bf31f300a391
85cc6df3817084f0b721f4d4f195a639ca8538a0
24038 F20101106_AABBKN jung_y_Page_32.QC.jpg
27caf32388559bbaf7c2fa6df8a7b154
70ebc88e235cf6d8573ab139d9669fe4b5b94cce
35041 F20101106_AABBFP jung_y_Page_27.pro
9e4ccf1119d2807a0107693158cde466
ccecd85512c1b769f3ca317d4dc71fe79e19ac13
26057 F20101106_AABAYS jung_y_Page_09.jpg
8c555005e00161634f04e7c0917c8bc1
8072d1484ef1a666d36b1bcd3cfe2b2037e01ca7
F20101106_AABBAS jung_y_Page_12.jp2
3a99d7ab9acd18d1f1152dc88eb2ee7c
0fec61d9ac2eee8483682a89bd0ae94d89c65eba
7070 F20101106_AABATV jung_y_Page_72thm.jpg
4a5ae209ccbf5bc2ab273e3d712acbb0
3a54f76ef6a102fcaddbada100bcccab5cc8a5bc
6108 F20101106_AABBKO jung_y_Page_32thm.jpg
d43ba9689aece10aade049281581701b
9f3295dd98f0de7a3362b6c3ed5dca65861c4bdf
58320 F20101106_AABBFQ jung_y_Page_28.pro
38abe929a0c9b50147a825d0a9a0e20a
d567b3c91b0d07ee36f8ee16f1f18583422a56bb
73779 F20101106_AABAYT jung_y_Page_11.jpg
5489bd8d1be3f08d39014190b9b933ff
8b129902ccc0a8e479ae66dba88e87bbfeea020c
F20101106_AABBAT jung_y_Page_13.jp2
07cff6c05c938b020f0d4395aaffb565
46bf4ca928e2f1a68109e1681189c5fb20cb57d9
6881 F20101106_AABATW jung_y_Page_66thm.jpg
5d85e7a5b9b2dcea7aafb6a087768dda
ebac77c531a27cd9d158c182d091fb5d5d5a75a1
6252 F20101106_AABBKP jung_y_Page_33thm.jpg
bbad69868f544ad4b6e468ba0d3b31ea
3d7d0bef748465e9f6516b5ceb43958f90a6a285
57499 F20101106_AABBFR jung_y_Page_29.pro
549aa14b5ca6df2090799634eab669ba
ee9fb4eb4b6529a886f95e76c8733710000a6db6
87398 F20101106_AABAYU jung_y_Page_12.jpg
6298f750b482b963fe530994f0ed744b
bc6fca581aea30aed391ece6d81fa1ac567b923c
1051925 F20101106_AABBAU jung_y_Page_14.jp2
d015c8079e74b929753e80d1e7d90b5a
b96b78f6f2e7361ae311b7bb428a61011f8af802
1135 F20101106_AABATX jung_y_Page_02.QC.jpg
9be198643891d45b034440b6c0e4e748
9fcf58f1bc870a8bdb0a8f6ccd49043a3824515c
5430 F20101106_AABBKQ jung_y_Page_34thm.jpg
5b995217c9c8b58f5b47caf57a2924a6
610b21e2cc7c365f241538abb64b15f2c71e4b11
55534 F20101106_AABBFS jung_y_Page_30.pro
4091a9f1338ef5b23e8c9ce72862e1bd
c03db470189ad237ab63f0f53348706e0b0d79b1
88644 F20101106_AABAYV jung_y_Page_13.jpg
91e27eca83d7cc27a74eb107f20e4e49
f925d6cb572bcd9ece0e04ad414475f3e733fe43
F20101106_AABBAV jung_y_Page_15.jp2
79d297052cec3adb76630ed927068212
723aee8bc0f83dc48904086771d3e0658b63768f
1051946 F20101106_AABATY jung_y_Page_62.jp2
c27bc9327ac249e2e704fd50272ae656
844a4e6b2924ad9ecc5f603820c96fa57f67d7b7







ONTOLOGY-BASED APPROACH TO SIMULATION WITH APPLICATION TO CITRUS
WATER AND NUTRIENT MANAGEMENT

















By

YUNCHUL JUNG


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

UNIVERSITY OF FLORIDA

2008
































2008 Yunchul Jung


































To my advisor, my parents and my girl friend









ACKNOWLEDGMENTS

After beginning a study at graduate school level, it was not easy to get an academic result.

I always liked pursuing new topics and learning new technologies for the future. I would like to

thank my advisor, Dr. Howard Beck. It was lucky for me to meet and collaborate with him. His

ideas and advice stimulated me to achieve my research objectives. His passion for research and

intellectual insights were motivating factors in my work. I would also like to extend my

appreciation to my cochair, Dr. Kelly Morgan, and committee member, Dr. James Jones for their

encouragement and guidance. Finally, I take this opportunity to thank my parents and my girl

friend, Kyungmi.









TABLE OF CONTENTS

page

A CK N O W LED G M EN T S ................................................................. ........... ............. .....

L IS T O F T A B L E S ................................................................................. 7

LIST O F FIG U RE S ................................................................. 8

ABSTRACT ........................................... .. ......... ........... 10

CHAPTER

1 INTRODUCTION ............... ............................ .............................. 12

2 L IT E R A TU R E R E V IE W ........................................................................ .. ....................... 17

O ntology B asked Sim ulation .......................................................................... ......... ........... 17
M odel-B ased A approach to O ntology ......................................................................... ...... 19

3 ONTOLOGY-BASED APPROACHES AND TOOLS FOR SIMULATION ......................23

B ackgrou n d T echn ologies ........................................................................... .....................2 3
O ntology ................................................................................................23
Ontology M anagem ent System (OM S)................................... ..................................... 24
M odel and Sim ulation O ntology ................................................. ............................... 24
E qu ationE editor .................................................................................................26
E quation O bject M odel (E O M ) .......................................................................... ... ... 26
Com ponents of the EquationEditor ....................................................... .... ........... 27
Sim ulationE editor .................. ................... ....................................................3 1
A additional T ools and F acilities...................................................................... ...................36

4 APPLICATION TO CITRUS WATER AND NUTRIENT MANAGEMENT .....................40

T h e C W M S M o d el ....................................... ................................................... .. 4 0
D description of M odel ..................................... ......................... ........... 40
M odel B ase of the CW M S M odel ....................................................................... ...44
Examples of Model Representing Process .............. ............................................47
Examples of CWMS Model Implementation ............ .............................................49
Soil geom etric dim en sion .............................................................. .....................49
T im e step ............................................................................ 52
R o ot d en sity ............................................................................................. .... 53
W ater dy n am ics............................. .................................................. ............... 55
N nitrogen balance ......................................................................58
A application Im plem entation ............................................................. ................... 60
M o d el E x ten sion ....................................................... ................ 6 3
Model Performance ................................. ... .. .... ...... .................64
M odel Sen sitivity A naly sis........................................................................... .................... 64









5 SU M M ARY AN D FU TURE W ORK ......................................................... .....................68

L IST O F R E F E R E N C E S .............................................................................. ...........................7 1

B IO G R A PH IC A L SK E T C H .............................................................................. .....................75



















































6









LIST OF TABLES


Table page

3-1 List of operators in EquationEditor........................................................ ............... 30

4-1 Input factors related with water input and hydraulic conductivity ..................................65

4-2 Sensitivity analysis result including main effects and two-factor interactions ................67









LIST OF FIGURES


Figure page

3-1 Concepts and relations in model ontology for simulation ...........................................25

3-2 Class diagram of Lyra equation object model retrieving instances of model ontology.....27

3-3 Features of the symbol editor in the EquationEditor: symbol ID, symbol, unit and
description of Cell Crop Evapotranspiration ........................................ .....................28

3-4 Database constraints and array dimension description of a symbol ................................29

3-5 Equation ID and equation of Cell Evapotranspiration in the mathematical statement
editor ................................ .............................. ............... ................. 2 9

3-6 Unit ID, unit and definition of centimeter of water in unit editor...................................31

3-7 Structure editor showing a compartmental diagram of a soil-water model.....................32

3-8 Relationships between ontology-based simulation system, generated code and
ap p lic atio n ............................. .................................................................... ............... 3 5

3-9 Symbol reference diagram focusing on Layer Daily Evapotranspiration........................36

3-10 Result of generating markup language for the equation A=B+C ................................39

4-1 Conceptualization of soil geometry of CWMS model................ ...............42

4-2 Exam ples of soil profile areas........................ .. ...................... ................. ............... 42

4-3 Relationship among equation symbols for water and nitrogen balance ..........................45

4-4 Taxonomy diagram of the CWMS model................................... .....................46

4-5 Morgan's infiltration rate model in the CWMS model .............. ...................................50

4-6 E quations of profile num ber ......................................................................................... 50

4-7 L ayer thickness m atrix ............................................................................... ............. 52

4-8 Example of dimension description of a symbol in soil layer......................................52

4-9 Total number of time steps symbol, t...................................................... .............. 53

4-10 Root density m atrix .................. .................. ................. ........... .. ............ 54

4-11 L ayer root density equation ....................................................................... ..................55









4-12 W after balance equations .......................................................................... ....................57

4-13 Enhanced hydraulic conductivity related equations in CWMS model ...........................58

4-14 L ayer nitrification equation ..................................................................... .....................59

4-15 Accumulated profile volatilization equation........................... .......................... 59

4 -16 S etu p P h a se ............. ................... ........... .... ...................... ................ 6 1

4-17 Irrigation scheduling result ........................................................................ .................. 6 1

4-18 Sim ulation results........... .... ....................... ......... 62











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 Engineering

ONTOLOGY-BASED APPROACH TO SIMULATION WITH APPLICATION TO CITRUS
WATER AND NUTRIENT MANAGEMENT

By

Yunchul Jung

August 2008

Chair: Howard Beck
Cochair: Kelly Morgan
Major: Agricultural and Biological Engineering

Simulation in agriculture and natural resource management is a popular methodology for

studying environmental and agricultural system problems. Traditionally, building a simulation is

treated as a software engineering problem, and simulations are implemented through manual

coding in a particular programming language. Problems of implementing a model and

developing a simulation system include difficulties in managing and reusing existing models and

simulation system because it is hard to understand the detailed specification of the system model

when it is written in a specific program language. Also, model specification may be lost during

the programming process, and it is difficult to maintain documentation describing the system

because documents are external to the programming process. Visual simulation environments

reduce the burden of programming, but there are still problems related to sharing knowledge

about the system.

An ontology is an explicit specification of a conceptualization, which can be used to create

a formal representation describing and categorizing concepts and relationships among the

concepts in a particular domain. Ontologies enable sharing through a common understanding of









the structure of information in a domain, enhance reuse of domain knowledge, and make domain

assumptions explicit by separating domain knowledge from operational knowledge. While

ontologies have been used in many domains as a way to represent generic domain knowledge, an

ontology-based approach to modeling and simulation in the domain of agriculture and natural

resources has not been well explored.

In this thesis, ontology-based modeling and simulation methodologies and tools are

developed which can be used by modeler and researcher to build mathematical models and

simulations, and in the process provide a better way of representing knowledge about models,

improve sharing and reusability, and provide a new basis for analysis of models and model

elements. These tools are applied to develop CWMS (Citrus Water Management System) model

as a way of evaluating the effectiveness of the proposed approach. An ontology for CWMS was

developed using the Lyra ontology management system. Tools that were developed for building

ontology-based models and simulations include the SimulationEditor, which is a high level

modeling environment for designing a system structure based on a graphic interface, and the

EquationEditor, which is a tool for designing a model in equation form and representing

knowledge of each equation and symbol by using the underlying ontology.

The main contribution of this thesis is the application of ontology-based techniques to

modeling and simulation in agriculture and natural resource domains through the development of

these tools and their application to a particular problem.









CHAPTER 1
INTRODUCTION

Simulation in agriculture and natural resource management is a popular methodology for

studying environmental and agricultural system problems. There has been much works on

modeling crop, soil, water and nutrients in specific research domains (Peart and Curry, 1998),

and recently interests in modeling and simulation methodology have moved to a reuse of existing

models and simulation systems for building a large system (Leon et al., 2002). This will require

better communication of model structure and components to the community of model builders

who collaborate on an international level.

Traditionally, modeling and simulation are tasks based on programming to implement the

processes necessary for operating or solving a model to mimic real system behavior within a

particular domain. General processes include developing computer logic and flow diagrams,

writing computer code, and implementing code on a computer to produce desired outputs (Peart

and Curry, 1998). While visual simulation, an approach to modeling and simulating based on

building diagrams of system components, is an intuitive and simple way to do this, programming

languages are still used for developing more complicated simulating system because there are

limitations on representing ability.

Classical problems of implementing a model and developing a simulation system include

difficulties in managing and reusing existing models and system that are written in a particular

programming language. Understanding a program written in a specific program language is

difficult because it is too hard to get information about the detailed specification of the system.

Usually this information is lost during the transformation to the program code (Furmento et al.,

2001) and documents describing the model are physically separated from the implementation.

Although documentation such as a paper or a manual and descriptions in program code partially









cover the gap in understanding, often documentation contain inaccurate information, the

document description does not adequately explain the entire system in detail, and it is difficult to

maintain both the system implementation and supporting documents.

Some simulation programs, such as Stella (Steed, 1992) and Simile (Muetzelfeldt and

Massheder, 2003), solve many of these problems by providing a visual modeling environment

and supporting embedded simulation and reporting tools. Visual environments eliminate or

greatly simplify the process of programming and make models much easier to design and

develop compared to hand coding of models in a traditional programming language, but sharing

of modeling products is still restricted because these tools use proprietary model representation

formats. Program source code is assumed to be an easily reusable, executable, flexible and

expandable way of sharing models, so that even visual programs provide the functionality for

generating program source code in a specific programming language from the models. However,

the problem is that different symbols and mathematical expressions are used for the same

concepts at the different viewpoints of modeller, so that there are enormous overlaps of concept

and interaction in models. These issues motivate research on ways to explicitly represent the

knowledge in a model (Lacy and Gerber, 2004; Cuske et al., 2005).

An ontology is an explicit specification of a conceptualization (Gruber, 1995), which has

been applied to create a formal representation describing and categorizing concepts and the

relationships among concepts in a particular domain. Classes are main elements of an ontology,

which describes concepts in the domain, and properties represent various features and attributes

of the concept. There is no single correct ontology of a particular domain, and several different

ontologies might exist depending on the task or role of ontology in that domain (Guarino, 1997).









Ontologies are based on object-oriented design, and thus appear to be similar to object-

oriented programming (Rumbaugh et al., 1991) and Unified Modeling Language (UML) (Booch

et al., 1997), but they are different in several important aspects (Noy and McGuinness, 2001). In

object-oriented programming, classes are regarded as types for instance, and each instance has

one class as its type, whereas, ontologies declare that classes are regarded as sets of individuals,

and each individual can belong to multiple classes. Also, in object-oriented programming classes

have behavior defined through functions and methods. Ontologies are not programming

languages, and classes in ontologies make their meaning explicit without any methods

(Knublauch et al., 2006). This is an important distinction, because methods are coded using a

programming language and thus behavior is not explicitly represented and is largely unknown

except through manual analysis or processing of the program code.

Ontologies enable sharing of a common understanding of the structure of domain

knowledge, reuse of domain knowledge, making domain assumptions explicit, separating

domain knowledge from operational knowledge, and analysis of domain knowledge (Noy and

McGuinness, 2001). These capabilities can be applied to the modeling and simulation domain. In

recent studies, it has been determined that ontologies increase the potential for interoperability,

integration, and reusability of simulation models (Miller et al., 2004). Also, ontologies can be a

useful for the description, development, and composition of simulation models, and for mapping

of input/output data.

In the domain of agriculture and natural resources, an ontology-based approach to

simulation, which represents a model with ontology concepts, can address several problems with

current methodology used to develop simulations. Whereas ontologies have been used as a way

to build generic domain knowledge, only recently have attempts been made to develop an









ontology-based approach to modeling and simulation. There have been many well-studied

physical processes in agriculture and natural resources, and different perspectives on the

problems led to development of many similar but varied models. Also, as the problems in the

agricultural domain has been diversified and widened to the environmental and natural resource

domain, requirements increase for modeling and simulation to solve multi-scale problems and to

integrate existing models rather than to develop new models for specific problems (Ewert et al.,

2006). Therefore, a comprehensive management system to manage these diverse models is

needed.

The objectives of the research presented in this thesis are twofold; 1) to develop ontology-

based methodologies and tools to be used by modeler and researcher for building mathematical

models and simulation in the agricultural and natural resource domain and 2) to apply the

methodology and tools to develop a sophisticated soil water and nutrient model for evaluating

the efficiency of the proposed ontology-based simulation approach.

Several software components were developed as a part of this research. The

SimulationEditor is a high-level modeling environment for specifying a system structure based

on a graphical interface. The EquationEditor is a tool for describing a model in equation form

and for representing knowledge of the equations and symbols used in the model. The Citrus

Water Management System (CWMS) is an application program applying the ontology-based

simulation methodologies. CWMS provides growers with site-specific optimal nutrient and

irrigation recommendations by simulating models based on soil characteristics, nutrient uptake

patterns and weather conditions. The motivations behind this research are 1) to create

methodologies based on ontology techniques to explicitly represent models and related

mathematical expressions of parameters, 2) to create an environment for building reusable and









sharable model knowledge, 3) to provide the core representational facilities of structure

diagrams, symbols/equations and descriptions, 4) to use the ontology as a database for

systematically storing models and model elements, and 5) to assess the value of ontology-based

simulation approach by applying it to modeling and simulation of CWMS.

The main contribution of this research includes the development of a methodology of

ontology-based simulation, which provide two main software tools; the SimulationEditor and

EquationEditor. These tools were used successfully to design and build a model and simulation

for citrus water and nutrient management.









CHAPTER 2
LITERATURE REVIEW

Ontology Based Simulation

Recently, ontologies have received much attention for implementing mathematical models

and building simulation systems. The aim of adapting ontologies for simulation systems is

similar across various projects, but the design and implementation of an ontology is different

depending on the problem domain (Benjamin et al., 2006).

Miller et al. (2004) noted that for modeling and simulation an ontology provides standard

terminology which increases the potential for application interoperability and reuse of simulation

artifacts. Furthermore, semantics represented in an ontology can be used for discovery of

simulation components, composition of simulation components, implementation assistance,

verification, and automated testing. He proposed a web-accessible ontology for discrete-event

modeling (DEMO), which defines a taxonomy of models by describing structural

characterization (State-oriented, Event-oriented, Activity-oriented, and Process-oriented models)

and a model mechanism explaining how to run the model.

Although Miller focused on the creation of an ontology for general stochastic models such

as Markov Processes or Petri Nets, Fishwick and Miller (2004) placed emphasis on capturing

mostly object or instance-based knowledge. He presented a software framework, RUBE, which

provides an integration method for the phenomenon of model and model object, and multiple

visual modes of display to provide interfaces for developing dynamic model. 3D visualization

(Park and Fishwick, 2005) is used to animate the responses of models. An ontology is used to

define a schema of simulation model types and models, and a sample air reconnaissance scene is

represented with the Web Ontology Language, OWL.









Some studies (Raubel and Kuhn, 2004; Cuske et al., 2005) addressed the use of a static

ontology (Jurisica et al., 2004), which describes static aspect of the world focusing on entities,

and in a simulation focusing on the data and the rules governing the simulation. They understood

that data used by a model is a key characteristic of semantics, which an ontology of an

information system should define, rather than building an ontology which is independent from

simulation form or contents. For example, ontology-based task simulation (Raubel and Kuhn,

2004) uses an ontology for evaluating the usability and utility of a task or data for the decision-

making process. JOntoRisk (Cuske et al., 2005), which is an ontology-based simulation platform

in risk management domain, proposed a three level ontology hierarchy, consisting of a meta risk

ontology, a domain risk model, an a risk knowledge base. Especially, a meta risk ontology

defines the common characteristics of risk management simulation with world elements which

are affected by risk, functional dependencies between world elements, random elements which

are input parameters, and stochastic dependencies between random elements. Models refined

from a meta risk ontology at a domain risk model have a strength on validating or reviewing the

meta structure of simulation system.

SEAMLESS (Ittersum et al., 2007) is a component-based framework for agricultural

systems which is used to assess agricultural and environmental policies and technologies from

the field-farm level to the regional level in the European Union. For SEAMLESS, an ontology is

designed to relate different concepts from models, indicators, and source data at different level,

and to structure domain knowledge and semantic meta-information about components for

retrieving and linking knowledge in components. It also is used to check the linkage between

components through input and output variables in the system. An ontology, the Model Interface

Ontology, encapsulates knowledge of biophysical agricultural models. Static and dynamic









models are included, and the system dynamics approach which describes a system with stocks

and flows are applied to conceptualize models. This approach to model ontologies provides

advantages which include the simplicity of model representation by using states, inputs, and

outputs, but it has limits on representing mathematical expressions of models and manipulating

models to build complex system. SEAMLESS does not attempt to represent models based on

their mathematical equation form in the ontology.

A web-based simulation using an ontology in the hydrodynamic domain (Islam and

Piasecki, 2004) is used to solve the governing equations for a two dimensional hydrodynamic

model. A model ontology is created to describe a numerical model by defining a specific

metadata set that describes hydrodynamic model data, which is used to search and retrieve

metadata information. This approach gives an advantage in prescribing geospatial data and

model data at model level. However, there is a limitation on building and describing model

equation, and model should be provided in a specific form required by the system.

The Modelling Support Tool, MoST (Scholten et al., 2007), a software framework for

supporting the full modeling process, used an ontological knowledge base (KB). The KB is a

collection of knowledge on modeling for various domains of water management, which is

developed by domain experts. They adopt ontological approaches to develop a knowledge

structure, store the knowledge to the KB following an ontological structure, and build software

applications to use the KB.

Model-Based Approach to Ontology

A model base is a massive collection of models and model components. As the number

and scale of models grow, the conceptualization and role of models within a problem domain

becomes wider and more complex. Some models may be considered as an integration of related

unit process models, while previously a single-process model itself was enough to make a









simulation. As various concepts are applied to develop an ontology to build a model, it becomes

a challenge to develop an ontology which contains different categorical views and which can be

used to manage models (Ewert et al., 2006).

As there are diverse aspects to understanding and describing models in a specific domain,

it is not easy to reuse existing model with other models or to replace a model with other models

which satisfies the same requirements of input data and parameters. In large-scale problem

domains, the need increases for comparing and evaluating models in order to locate an adequate

model for a given environment. Lu et al. (2004) compared different models for estimating leaf

area, and Eitzinger et al. (2004) performed a evaluation and comparison of water balance

components in different models. To provide a model base, there is an effort to develop a set of

crop models for a various crops and integrating models with farm decision support system

(Reddy and Anbumozhi, 2004).

A modular approach to model development (Jones et al., 2001) introduced by categorizing

and organizing crop model with biological, environmental and management module as a form of

software component, which is an executable unit of independent production, in the agro-

ecological domain (Donatelli et al., 2006a,b). Although they offer useful ideas on categorizing

and reusing the existing components, they cannot fully address the difficulties of model

management because they are developed for a specific program environment such as a

FORTRAN and C++.

These difficulties make it important to organize a model base that can compare similar yet

different models and components. It will be useful to categorize and organize models into a well

designed framework for the purpose of locating and reusing models. There have been many









efforts to construct model bases, and recently ontologies are being applied to this purpose

because of their strength in categorizing and organizing knowledge.

Watershed modeling is considered as aggregating a complex system of unit hydrology and

chemical processes, which includes precipitation, infiltration, evapotraspiration and erosion.

Haan et al. (1982) presented a collection of generic processes and practical models which have

been used to study the hydrologic cycle in watersheds. MoST (Scholten et al., 2007) developed a

model ontology following the structure of components in the simulation system to manage

models, and it made it possible to switch one model with other models in the same process level

for seeking appropriate model composition resulting in an adaptable conclusion. But, the

complexity of the representation is not enough to describe detailed processes, and the large scale

of the system makes it difficult to manage models. Although it enables model switching, it is

limited to simple models.

Some research to support a decision making process over a farm or water management area

provides a library of models that allows a user to build up a simulation system easily with unit

process models (Athanasiadis et al., 2006; Scholten et al., 2007). The libraries contain ontologies

for storing the farm management model knowledge which is gathered from references or experts.

Usually, in those cases, models can be repeatedly used for building up a system, but there are

limitations in modifying or creating another model from known models, even models which the

system provides. A simple case is that an ontology is not designed originally to allow any

manipulation, and this problem is usually found at the multi-scale simulation model.

To solve the difficulties of managing models in ontologies, the SEAMLESS built a model

ontology which contains multi-scaled categories over an agricultural domain, and provided an

interface for managing model knowledge, which is an authoring tool supporting to create and









categorize models and to modify model knowledge (Rizzoli et al., 2004; Athanasiadis et al.,

2006). Model knowledge appearing in the interface includes a model description, creator, a

components list using selected model, and model elements. Model elements describe model

input, output, and state variables which can be used to select models. Although input, output, and

state variables can be dictated in the interface, it does not represent the detailed and complicated

mathematical relations between them. A model ontology just contains knowledge of concepts

related with a mode as input/output or state variables, and their mathematical relationship is

coded or internally described in the system. To resolve these limitations, it is required to focus on

designing a model ontology based on their mathematical representation and meaning explicitly.









CHAPTER 3
ONTOLOGY-BASED APPROACHES AND TOOLS FOR SIMULATION

This approach to ontology-based simulation focuses on model authoring facilities and

simulation execution tools. In the following sections, supporting technologies which enable

modelers to develop ontology-based simulations are described. The SimulationEditor and the

EquationEditor are the two main tools for building a simulation system. Additionally, system

validation tools, a symbol referencing flow diagram and a sensitivity analysis tool, which

provide facilities for model analysis, are also described.

Background Technologies

Ontology

An ontology is a formal explicit representation of concepts in a specific domain (Gruber,

1995). Specifically, in computer science and information science ontology is considered to be a

data model which represents a collection of concepts in a domain and relationships between

those concepts. To form the representation of a data model, several elements are considered

including class, individual, attribute, and relationship.

An instance (also known as an object or individual) is a concrete (e.g. people) or abstract

(e.g. number) object in an ontology, and a class (concept) is an abstract group of similar objects

in the domain, which may contain individuals, other more specific classes or combination of

both. An attribute characterizes and describes a property of a class, and has at least a name and a

value. Since an important use of attributes is to describe the relationships between objects, a

relationship is an attribute whose value is another object in the ontology. The power of ontology

comes from the flexibility in describing relationships. A common relationship is the subsumption

relation such as 'is-superclass-of, 'is-subclass-of, which defines classes that are more general or

specific than other classes of instances, and the relation part-of which represents how instances









combine together to form composite instances (Noy and McGuinness, 2001). The procedure of

developing an ontology consists of defining classes and individuals, arranging the classes in a

taxonomic (subclass-superclass) hierarchy, defining attributes and describing allowed values for

these attributes, and filling in the values of attributes for instances (Guarino, 1997).

Ontology Management System (OMS)

The Lyra ontology management system (Beck, 2007) is used to build an ontology for

modeling, to develop tools for entering symbols and equations into the ontology, and to

implement the tools that execute simulation and show their results. Lyra is an object database

management system for ontology data, which provides a data model of the linguistic and

semantic concepts in an ontology based on a formally defined ontology language. It supports

management of large collections of ontology objects, reasoning facilities that help in organizing

and searching for concepts, visual ontology design tools, and application development tools. It is

designed as a server/client system implemented with Java. Clients communicate remotely with a

database located on a remote server through Java Remote Method Invocation (RMI) technology.

Model and Simulation Ontology

The model and simulation ontology is developed with Lyra for building conceptual

diagrams graphically, representing mathematical models with symbols and equations, and

describing information related to each symbol and equation. It consists of two parts, system

design and model implementation (Figure 3-1).

In order to build a simulation system diagram, three classes (project, diagram, graphic

elements) are created in the model ontology. System design is a process of building a conceptual

or structural diagram of concepts and relations between them with graphic elements, and each

diagram belongs to a specific project. A graphic element may represent a set of equations









describing a detail (or independent) process in a complete set of models or explain a structural

relationship which means a physical part of a system.


Figure 3-1. Concepts and relations in model ontology for simulation



Classes for representing a mathematical expression focus on symbols and equations. An

equation includes knowledge about mathematic operators' hierarchical relations, symbols, and

meaning. One challenge in describing a symbol is that a term for the symbol composing an

equation may be used in other equations with the same or different meaning. A symbol (concept)

is unique but having multiple names (terms) which are mapped to symbols, and enables

reusability of the symbol in different models. A symbol may have one of three different sources

for its value; constant, equation, and database.









EquationEditor

The EquationEditor is a tool for creating equations associated with a model, and properly

defining symbols appearing in these equations. It provides a facility for creating, browsing, and

inspecting all equations, symbols, and units appearing in the model. It uses an interface that

resembles other equation editors such as Microsoft Office Equation Editor (Microsoft, 2003) and

MatyType (MathType, 1996), but differs significantly because all the symbols in equations and

equations are represented internally by using ontology objects. This provides a way to represent

the meaning of equations and symbols that is not possible with other equation editors.

Equation Object Model (EOM)

The Equation Object Model (EOM) is an intermediate collection of basic objects that

represents information describing a mathematical expression and communicates with the Lyra

physical storage manage to retrieve and store equations and symbols (Figure 3-2). The main

purpose of EOM is to represent the elements of mathematical expressions. Operators and other

symbols of an equation are objects of the two main classes, MathTemplate and MathPrimitive.

MathTemplate defines a type of operator and a collection of arguments. Character symbols and

numerical symbols are subclasses of MathPrimitive. MathSymbol objects representing a symbol

contain two properties; a linguistic-level property and a programmatic-level property. Symbol (in

multiple terms), symbollD, and definition are linguistic-level properties. Programmatic-level

contains three properties; source, matrixType, and matrixSymbolUsage. A property source

represents the origin of the numerical value of the symbol (equationType, databaseType, or

constant). If the symbol is a matrix, property matrixType gives the matrix dimensions. A flag,

"constant" or "variable", is a value representing matrixSymbolUsage property. The value

"constant" means that the value never changes, whereas "variable" means the value can change.













I I UML Designer Lyra Eti-


Figure 3-2. Class diagram of Lyra equation object model retrieving instances of model ontology





Components of the EquationEditor


The EquationEditor has three sub-editors, Symbol Editor, Mathematical Statement Editor,


and Unit Editor, to create and maintain symbols, equations and symbol units.


Symbol Editor (Figure 3-3) is an editor for individual symbols appearing in equations and


includes a symbolic expression of a symbol, a quantity of measurement, and a description of the


linguistic and programmatic properties of the symbol. A symbol is implemented as a class in the


ontology, which has a unique meaning within a specific domain. Often, the same symbolic


character (term for the symbol) is used over different domains, but is used in different ways and


has different meaning. Since a symbol has a distinguishing identifier representing a specific


Projects Search Windows Help
S T B C cR R mr I



r,,11r I: I ."'- -1
i: I ,.





I- ri II A, .
_l-r,.- .,, ,-











I. I r.rj .
r 1-



S r Iri.-:,,, i., -
S, .r.- I :l: 1,I
1r 1 r i ,. iI 7 ii-
.. r .,r r: :111. .. .





,,, I T j-' I : 1 -







-1-- r 111:,


r I 1- 7 1 r.~I .I
r~ I 1-7 1 1










concept in the ontology, a use of the same term for different symbols is permitted, and the

domain ontologies can be used to resolve their ambiguous meaning.











Figure 3-3. Features of the symbol editor in the EquationEditor: symbol n it SdID, symbol, unit and












database, or from a constant which is directly assigned to the symbol. In the case where the
symbol value is determined by an equation, there must be an equation in the database in which












this symbol appears alone on the left side. To obtain the value from the database, some
current time and a soil layer number for querying a soil temperature at a sp eific date), and these
a a-I V J-- a .


















constraints can be specified as a part of the symbol's properties (Figure 3-4).














Symbols can also be arrays, when a symbol can be used in different discrete intervals in
space and time. For example, soil water content can be expressed in different soil layers which
occur in different soil profiles, characterized by the depth from the soil surface, the soil profile
number and timed 5 Ia C 0a1|
Sla" a* L it 1r Tp; Vo .- -r- -






Figure 3-3. Features of the symbol editor in the EquationEditor: symbol ID, symbol, unit and
description of Cell Crop Evapotranspiration




The value of a symbol is determined by one of three methods: from an equation, from a

database, or from a constant which is directly assigned to the symbol. In the case where the

symbol value is determined by an equation, there must be an equation in the database in which

this symbol appears alone on the left side. To obtain the value from the database, some

constraints may be required in order to locate and query a database to obtain the value (e.g. a

current time and a soil layer number for querying a soil temperature at a specific date), and these

constraints can be specified as a part of the symbol's properties (Figure 3-4).

Symbols can also be arrays, when a symbol can be used in different discrete intervals in

space and time. For example, soil water content can be expressed in different soil layers which

occur in different soil profiles, characterized by the depth from the soil surface, the soil profile

number and time.

















E juat.. s -Tt 5 Uln.c s


Symbols
Fraction of Passive Carbon


storical UnShaded Soi Layer Tc
vdraulic Conductivity
ydrauli Conductivity Active Fla
Hydraulic Conductivity Parametel
ydraulk Conductivity Paramete
hydraulic Conductivity Wetting F
1aed lif..pcr..Ic'r..2r.f.r.r. Ict.
d.al.? ri 1rmrAL-Tr.r. r Y. 4c(
'1 r 31 r j ir .tc h er. :.r f :.r l



d :1 P i.T.r;. l ', l. .-. rrc.m : r,.
Tdael' I.T.r.-..-13 _r,-..- rCr...T A.: r.



i ]1P
16si F lvrn.,bIcsri. nrr 'ror, !.lpl3t


New




Show al Equations/Symbols for:
SCWMS sal lyer
OCWMS


Symbol is Being Correction

Symbol Der.r r.or.. u,-, .,,.* :i I1.

X b a 8 U L -S V I- i- E +13 -] Tcc

Historical Soil Layer Tempeaure

Historical Soil Layer Temperature


(


Enable Index Symbol: el-.:r -.'ri.bDllO t rbanretale I:
Sourre Type Equation type wy-. -. v 3. k- 0

( Database :ondrition 2

0 Constant 10
Value Type : 0 Scear ( Matrix Dimension as r anc
Iu ] [ *s.:,s.t.o.n 1 rjo ] [ Destr, r.h.i 1
L c- fert*e LaoYr 1-*e t v 1 Tatl jI.-rt.e, Cr rcr.ie T,De. v
,Z rcr s INurrrber -i Ti.e [epns sD ill VI rjur.;.bei c r rcd.Ie;
&I 15...nh1.- rF I r, v.
a., T: Ijl. l tIc cF Tme cC s -


DB Constraints


SymbolArray


I save _


Figure 3-4. Database constraints and array dimension description of a symbol



P=O =yfNORM


air S


~.quti, fi~uanlo.n


1. L, h -f



SeltDaLy sdl Coeffarien -lqtiS -.
Cell oav9 Ek i.eJNrhlt Er
.r' ,- ., '" -. ,-..
<.e L .* -1. < 1. p
elU'yFIqEqjtaon
elDry Flag EqJaon BK MayO9
l EfLfew Rain Equa n


CIrLe U C~ ty wnrdr DeRth
ellkrrlation Waer A-munI to
ll Leaching Anrnim Amrun
elachin AronumAmun






khow olEquatio sfTv5utds for;
Scwmssolical
0 CW4S


c :-. r ; I



A -. .I -.. Euan Is* L'"

lA A 5 .' r I. V L,,. .



-' Ah lk ,- ,
.. : l r












Delete


CwrN5 sol call
O .
I"' "a


Figure 3-5. Equation ID and equation of Cell Evapotranspiration in the mathematical statement

editor


ec-uaions k sy I urIs









The mathematical statement editor is designed to graphically create an equation from

existing symbols and mathematical operator templates (Figure 3-5). An equation is an expression

that has a hierarchical tree data structure composed of symbols and operators. The equal operator

is the root node of the tree structure, containing a single symbol on the left branch of the tree.

The value of the left side symbol is defined by the calculation of the right side terms. Thus the

equation is assumed to be a function which has symbols as arguments. The editor provides an

operator template which can describe specific argument sets. There are eight operator groups

used to compose an equation (Table 3-1).

Table 3-1. List of operators in EquationEditor
Operator group Operators
Exponential Subscript, double subscript, superscript, exponent, sub and super
script, function, square root, root, log
Fence Parenthesis, bracket, brace, absolute, ceiling, floor
Trigonometry Sine, cosine, tangent, arcsine, arccosine, arctangent
Calculus Limit, differential, indefinite, definite, summation, product,
maximum, minimum
Logic And, or, not
Arithmetic Add, subtract, multiply, divide, negation
Relation Less than, greater than, less and equal, greater and equal, equal,
equivalent, not equal, not equivalent, less than and less than equal
to, less than equal to and less than
Case n-case, matrix


The Unit editor is an interface to create and maintain the unit for a symbol and its

compositions for representing the quantity of measurement of symbols (Figure 3-6). Unit

includes not only the generic collection of global standard unit of metric system (e.g.

international system of unit (SI) and the English unit system), but also domain specific units such

as "cm3 of soil" in soil engineering. It is very important to carefully track the units associated

with symbols, since different models may use the same symbol but having different units. A unit

is not represented by a simple string, but by a composition of symbols (like an equation). The












unit can be expressed using a composition of limited operators (multiply, divide, and power


operator) and other units. Thus, basic units such as length and weight can be reused for creating a


composite unit, and this makes it possible to automatically calculate conversion of units from


one form to another (e.g. the English unit to the metric unit).

.udr.id .. i .i .l. L' .-.






'r aI. ,ni U r



Ce tIr.ete'
c -.. .dl 4 .-ct- r : I:
.........,*,-...... || u u V -=- = o- ia n
r..r IL.-.il. ;[r.r -. Irr +p

o.T.r rn.'.a cm of water
qrdaT. r nlrrO ,r.

E'J .. Sy- r bols Unr,
meter



CJi'. r-..- "1. ,r t Symbol
*r,, Ferr,,T...r ,.
Iaticcentkneter of water [ X Kf B z U W F = M3 43 ,.*A

grrr cl anvun an
rf rntrate centi meter or water
r f nitrogen
onter
square centineter


Figure 3-6. Unit ID, unit and definition of centimeter of water in unit editor





SimulationEditor


The SimulationEditor is used to describe the structure of dynamic systems using graphic


elements such as source, sink, storage, and flow. It adopts concepts from the compartmental


modeling technique (Peart and Curry, 1998) and Forrester notation (Forrester, 1971) which is


widely used in agriculture and natural resource models. However, like the EquationEditor, these


concepts are represented internally using the ontology and stored in the Lyra database. The


SimulationEditor is used for specifying the overall model structure in the form of elements, and










incorporates the EquationEditor described in the previous section in order to build equations

associated with each element. The SimulationEditor provides a graphic user interfaces to create

and maintain a simulation system which includes a structure design interface, a simulation

control interface, and a simulation result reporting interface. The SimulationEditor also contains

facilities for automatically generating and running simulations and generating reports.


Projects Windows Help









soil profile


soil layer









Figure 3-7. Structure editor showing a compartmental diagram of a soil-water model

The structure editor is the main interface of the SimulationEditor and provides

functionalities which enables modelers to create and maintain a simulation project by designing

the structure of a system, and to interact with the EquationEditor and the simulation controller.

Structural design of a system is a procedure by which a modeler creates physical or

environmental elements and relationships in the system by using graphic elements. For example

(Figure 3-7), a 3-dimensional soil profile system may be designed as a composition of soil cell

(production unit), soil profile (horizontal division), and soil layer (vertical division) concept.

These three elements may be defined as an instance of storage element, and relationships









between these elements are represented by 'Part of (e.g. block, soil cell and soil profile

composed of soil cell, soil profile and soil layer respectively ). Irrigation may be realized with

the flow element representing the flow of water into the cell.

The simulation controller is a collection of simulating engines used to generate a

simulation program from the mathematical model, to run the simulation, and to generate reports.

For simulating a model, a simulation engine automatically converts ontology objects to program

source code. It then compiles and runs the generated program to execute the simulation.

Currently, Java is the target language, although in theory a simulation program may be generated

using other programming language. It is not necessary for the modeler to examine, work directly

with or otherwise be concerned about the generated program source code. This process is

completely internal to the operation of the software, and transparent to the modeler. However,

the compiled source code can be used as a component that can be accessed by other software

environments after the model is developed.

The data object conversion and generation of program source code follows these steps:

A class representing a module in the SimulationEditor forms a single class in Java. The

class contains member variables and methods for all the symbols and equations in the

module.

Each symbol in the module is declared as a member variable named after the name of the

symbol. If the symbol is a matrix, the member variable is declared as an array with the

same dimensions as the symbol. A method is created that contains code for obtaining

the value of the symbol. The name of this method is based on the name of the symbol.

o If a symbol is a constant, the return value of the method is a constant for the

symbol's value.









o If a symbol obtains its value from a database, the method returns a value

obtained by querying the database for the value of the symbol, subject to

constraints specified in the symbol's properties.

o If a symbol obtains its value from an equation, the method contains code for

solving the equation to obtain the value. Since the equation contains other

symbols, these values of these symbols (on the right hand side of the equation),

are obtained by recursively calling methods for determining their values.

Generated source code set is designed to be independent from the SimulationEditor, so that

it can be used as part of a component library and inserted into other application independently.

There are two levels to the system: the core level and the application level (Figure 3-8). The core

level is the ontology-based simulation environment including the SimulationEditor and, the

EquationEditor integrated within the Lyra OMS which also provides the database management

facilities for storing the ontology objects created by the SimulationEditor and the

EquationEditor. At the application level, the generated code library is used by the other

applications which can be implemented independently from the core level. For example, the

resulting simulation application can be integrated into a desktop application used by growers, it

can be part of a larger decision support system such as DISC (Beck et al., 2004) which is a citrus

planning and scheduling program or a Web-based simulation environment (users can run the

simulation through a web page, or the simulation can be part of a web service that is part of a

distributed simulation environment).









Core Engine Level


Figure 3-8. Relationships between ontology-based simulation system, generated code and
application

Running a simulation involves compiling and executing the automatically generated source

code. The simulation is controlled by recursively evaluating the value of a target symbol. Within

the SimulationEditor, there is an interface to communicate with the model code library, which

contains a method for calling the target symbol's method which results in execution of the

simulation. The generated source code contains variables for storing all values of variables,

which are retrieved by a report generator to display model results when the simulation has

finished executing.

The report generator displays simulation results by showing the values of specific symbols

in the form of a table or a graph as a function of time and proper dimension described in a

symbol. A list of symbol ID which is stored in the ontology is provided to create reports, and a

report is designed by selecting and adding to the target variables list and the dependent variable

after simulation. A designed report form can be categorized and maintained in the ontology.









Additional Tools and Facilities


A system may be composed of many small models, and these models reference other

models or equations. There is a need to verify interactions between such complex structural

relationships, and to assess the behavior of models statistically. Tools are also available for

exporting the model into XML in two different formats, MathML (Ausbrooks et al., 2003) and

OpenMath (Buswell et al., 2004).

Simuatio Conrol ente


kl Exp420 V Rad12 V ZomI


Figure 3-9. Symbol reference diagram focusing on Layer Daily Evapotranspiration

To verify the complex flow of referencing equations in a model, it is useful to visualize

those flows and call sequences as a diagram. The calling sequence is generated by interpreting

attributes recursively in a process that is similar to the process of generating the program code.

Figure 3-9 shows an example of a calling sequence of equations which are parts of CWMS









models presented in Chapter 4. To calculate a value of a symbol such as the layer daily

evapotranspiration (ETLcs), some values are required including the total root length (TR), the

cell crop evapotranspiration (ETCc), the soil coefficient (ks), the today number of time step

(TodayTS) and the layer root length (rl). Arrows starting from the target symbol point to other

symbols which are required to calculate it. Numbers over each node indicate of nodes which are

related to that node, but not shown. The calling sequence interface provides a convenient way for

displaying every connected symbol as a visual network diagram, and it can be used to browse

symbols and navigate the model.

A sensitivity analysis process is implemented using computer experiments. The aim of

sensitivity analysis is to determine how sensitive the output of a system is with respect to the

elements of the model which are subject to uncertainty or variability (Wallach et al., 2007). This

is useful as a guide when the model is under development as well as to understand model

behavior, to seek the main affecting factors in the system, and to figure out the significant

interaction between input factors selected from the system variables. The procedure consists of

two step; factor screening using Morris randomized OAT (One-factor-At-a-Time) design

(Morris, 1991; Alam and McNaught, 2004) and global sensitivity analysis with screened factors

from previous step. The Morris' randomized OAT design as a factor screening method is used to

determine which factors are really significant when there are potentially a large number of

factors involved. A computer experiment is a set of simulation runs designed to explore the

model responses when the input varies within given ranges. The number of executions required

to do this is dependent on a number of selected factors and levels. After choosing input factors in

the system, discrete levels are automatically decided by the maximum value, minimum value and









number of levels. An experiment engine feeds factor values at a specific level, and control the

iterative simulating process.

The XML generator is a tool to generate a markup language form for a model built in the

ontology. XML enables the model to be shared outside of the Lyra OMS environment. Two

forms of markup language, MathML and OpenMath, can be generated. MathML is an

application of XML for describing mathematical notation and capturing both its structure and

content. It aims at integrating mathematical formulas into Web documents. It is a

recommendation of the W3C (World Wide Web Consortium) math working group. Whereas

MathML has strength on presentation of formulae, OpenMath is a document markup language

for mathematical formulae, which provides a mechanism for describing the semantics of

mathematical symbols. To generate these XML formats from equations in the ontology, each

operator template class which is declared in the EquationEditor has a method transforming

operator and arguments to a string containing a XML tag expression. An operator template can

have other operator templates as arguments. An equation may be considered as a tree data

structure composed of operator and symbol. The XML generator traverses this tree from the root

operator template (which is always the "equal" operator) to each leaf operator template, similar

to the way in which the code for solving the equation is generated. An example of generating

markup language is shown in Figure 3-10 which a simple equation "A=B+C" is generated in

MathML and OpenMath format (generated Java code for solving this equation is shown as well).












[ Source Equation]
A=B+C

[ generated documents]




A
=

B
+
C


< Jaa--->

public double AP

"returnBO +CO";



Figure 3-10. Result of generating markup language for the equation A=B+C


In Chapter 3, methodologies were covered to represent mathematical models in the


ontology using the Lyra ontology management system. To utilize the constructed model


ontology for simulating models, two main tools, the EquationEditor and the SimulationEditor,


are developed. The EquationEditor provides interfaces for describing symbols and equations in a


model and for retrieving ontology objects, and the SimulationEditor helps to conceptualize the


circumstance in which models are applied. Simulation handling process is facilitated by


functions including automatic generation of program codes and reports, sensitive analysis, and


calculation sequence diagram. Model representation adopting ontology-based methodologies


simplifies to create a deliverable model expression such as a XML form.









CHAPTER 4
APPLICATION TO CITRUS WATER AND NUTRIENT MANAGEMENT

Ontology-based simulation methodologies covered in Chapter 3 were applied to building a

model describing water and nutrient balance processes for the Citrus Water and Nutrient

Management System (CWMS) (Morgan et al., 2006a). To aid growers in water management

decision making, a computer-based decision support system was developed to facilitate more

efficient use of water and nutrients by basing recommended application rates on site specific

characteristics and local weather data. The purpose of this work was to test the feasibility of

utilizing ontology-based simulation to build a moderately complex model, resulting not only in a

simulation that can execute rapidly, but that also can be incorporated into a user interface for

delivery to and use by growers or in other applications.

The CWMS Model

Description of Model

The CWMS model has been designed for the sandy soils of central and southern Florida

which have low water and nutrient retention capacities. At citrus production sites, nutrients may

be leached from the sandy soils by excessive irrigation events. The CWMS model was developed

to anticipate the potential contribution to the groundwater contamination and to provide

appropriate irrigation scheduling strategies. The CWMS model uses two main water inputs,

rainfall and irrigation events. Rainfall amount is assumed to be affected by the canopy volume

covering the soil surface. An irrigation event contains information including nutrient

concentration (in the case of fertigation or injection of liquid fertilizers into the irrigation water),

amount and event date, and it consists of several distinct irrigation processes. The soil water

budget models in the CWMS model are based on crop water use, soil-water storage capacity, and









vertical soil water movement. Horizontal water movement is excluded due to lack of lateral

movement in the sandy soil.

The model is based on a restricted area, a soil cell in a block representing a single citrus

tree and the drainage field surrounding it, which is the basic unit of the geometry (Figure 4-1). A

commercial block of citrus consists of many soil cells since it has many trees, but in this model

to simplify the simulation process the model is based on a single soil cell, and the single tree

represented by that soil cell is characteristic of all the other trees in the block. A soil cell is

defined as a cubical soil area containing one citrus tree, having a depth of 200cm from the top of

the soil. A soil cell is further divided into soil profiles within a cell and soil layers within a soil

profile. As shown in Figure 3-7 in Chapter 3, seven concepts are defined for the model structure;

block, soil cell, soil profile, soil layer, root, irrigation, and weather.

A soil cell consists of a one-tree planting row area with the tree in the center. The width

and length of the cell are in-row and between-row distances to adjacent trees. It includes four-

types of zones (i.e. a non-irrigated & dry-fertilized area, an irrigated & dry-fertilized area, a non-

irrigated & non-dry-fertilized area, and an irrigated and non-dry-fertilized area as shown in

Figure 4-2) according to the irrigation status and the dry-fertilized status, and each zone may

have from 1 to 5 soil profile(s) which consist of n soil layers. The total soil layer number (n) is

determined based on the particular soil type or a depth of each soil layer.














Grove


Rain
Irrigation / Transpiration






Leaching
Leaching


[ Profile Area Definition ]
!~i


Soil Cell (i: irrhon-irr)
Soil Profile (j)
/ -Soil Layer ()


< Case of Partially Irrigated >


< Case of Fully Irrigated >


INDF Area


IDF Area














NIDF Area

NINDF Area


[ i index ]
i = 1 NIDF : Non Irrigated-Dry Fertilized i = 2 IDF : Irrigated-Dry Fertilized
i = 3 NINDF : Non Irrigated- Non Dry Fertilized. i = 4 INDF : Irrigated-Non Dry Fertilized


Figure 4-2. Examples of soil profile areas


rea Boundary by Canopy ----
lation Area (additional profile boundary)
Irrigated Profiles Basic Profile Layer 1 -
\ / .^r Boundary -----
\------------ --.i------

Non-Irrigated Profiles 200c -- .



SLayern ..................



Conceptualization of soil geometry of CWMS model


Figure 4-1.


I NINDF 1


NINDF 1


NIDF v '


NIDF 4



ID F IDF 12 1 ID F 1 ID F 3 ID F .


DF
.I-- --, ,.


__









Water balance for each profile is determined using rainfall and irrigation events as water

inputs and evapotranspiration (ET) and leaching water as water losses from the balance.

Basically, the water budget calculation is based on the tipping bucket model, enhanced by

considering the effect of the delayed soil water drainage caused by the soil hydraulic

conductivity during one daily time step. It is assumed that water inputs are applied at the first soil

layer. Water infiltration depth is calculated as a function of the infiltration characteristics of the

soil.

A pure tipping bucket model assumes that water moves the entire layer depth in one time

step. Tipping bucket does not always reflect soil water content on a daily time step due to soil

hydraulic characteristics. To account for the hydraulic characteristics of the soil, CWMS assumes

that all irrigation, water with N application and rainfall occurs at noon and has a maximum of 12

hours to move through the soil on the first day. The model calculates the wetting front speed and

the time for which the wetting front travels a layer thickness, thus to obtain the layer index of

wetting front at the end of the day and let unfinished drained water continues to drain at next

day.

As stated above, water loss from the water balance is water drainage below the 200 cm

maximum depth and crop ET. Daily crop ET is calculated in CWMS using reference ET as user

input or from weather data. A crop coefficient based on Morgan et al. (2006b) is applied to the

reference ET based on seasonal variability. The crop ET deducted from each soil profile layer is

proportional to the root length density of the profile layer. A layer root length density distribution

is a function of tree size (Morgan et al., 2006a). An irrigation is scheduled when the CWMS

determines that water content in the irrigated zone is below the allowed depletion. The irrigation









duration is determined by the water amount needed to bring water content to field capacity and is

determined by the emitter flow rate, irrigation efficiency, and irrigation depth.

For nutrient management, especially application of nitrogen by irrigation (fertigation), the

CWMS model provides processes for calculating nitrogen balance followed by transformation of

nitrate and ammonium. Transformations are composed of complicated sub-processes and are

affected by input and flow (drainage from above layer) amount of nutrient and water between

layers. The model assumed that there are four nutrient flow processes by four drain events: drain

by rain, pre-irrigation, during-irrigation, and post-irrigation. After four-drain steps,

transformation processes is applied to ammonium and nitrate following in order of volatilization,

uptake of ammonium, uptake of nitrate, and nitrification.

Model Base of the CWMS Model

The CWMS models is implemented by using a taxonomy representing physical

relationship of natural resources (soil profile, crop, and environment), which consists of 4 soil

related concepts (block, soil cell, soil profile, and soil layer), root density, weather, and

irrigation. The system structure taxonomy is built graphically with the SimulationEditor (Figure

3-7).

For the CWMS model approximately 700 symbols and 500 equations were created.

Symbols and equations developed for the CWMS model are interrelated, and their relationship

can be visualized as a graph diagram (Figure 4-3), which displays connections of symbols and

equations used to model the water and nitrogen balance process. Symbols and equations are

represented as rectangular boxes, and the number above a box shows the count of connected

boxes but not displayed. An arrow means that the target symbol is required to calculate the

source equation.















































Figure 4-3. Relationship among equation symbols for water and nitrogen balance

In the diagram, nitrogen balance process group is connected with the water balance process

group by referencing several equations for water balance. They can be switched with similar

process groups independently. Particular processes, uptake, nitrification and volatilization, are

clustered clearly from other equations in the equation group of nitrogen balance, and for water

balance a similar pattern was found for processes of evapotranspiration, infiltration, and

irrigation. This suggests possibilities for organizing and categorizing models and subprocesses.












The graphic in Figure 4-3 is generated automatically from the ontology objects, and an animated

interface allows the model to navigate through the space of symbols.

A model base is a database of many models, model elements, equations, and symbols. It

can be utilized for reusing models by applying a taxonomic organization to an ontology. In

Figure 4-4, a model base is shown, which consists of equations used in the CWMS model. The

taxonomy contains 6 classes including weather, water, nutrient, soil, crop and site. The water

class has 5 subclasses (infiltration, evapotranspiration, precipitation, irrigation and runoff), and

each subclass contains related equations and symbols. For example, the CWMS model includes

two different infiltration models (Tipping Bucket Water Content Equation and Enhanced

Wetting Front Water Content Equation).

k FdT WMsolayer 1 Zoi -1 | I


To ee he nodedeescr pion, move mouse on the node

Figure 4-4. Taxonomy diagram of the CWMS model









Examples of Model Representing Process

Representing a model formed as theoretical or logical expressions (equations) in the

ontology-based simulation system is a process of redefining and adjust them into a real world

simulation for describing a phenomenon. Theoretical models provide the conceptual knowledge,

and there are usually many reformulated models depending on assumptions and the given

situation.

Morgan et al. (2006a) described how they derived the infiltration model used in CWMS

from theoretical models. In summary, their infiltration model is based on Green-Ampt infiltration

(Green and Ampt, 1911) and unsaturated flow based on Richard's equation (Richards, 1931)

derived from Darcy's Law for irrigation and rainfall moving into soil from the surface and

moving between soil layers. The Green-Ampt model in the case of no ponding at surface can be

expressed as:

fJ, = -I- KC AM S> /F

Where,

fp : infiltration capacity of soil (the rate that water will infiltrate as limited by soil factors)

F : cumulative infiltration

Ks : the hydraulic conductivity of the transmission zone

M : the difference between initial and final volumetric water contents

Sf: the effective suction at the wetting front



Furthermore, the model adopted Mein and Larson's equation (Mein and Larson, 1973)

applying the Green-Ampt model for rainfall conditions by determining cumulative infiltration at

the time of surface ponding. Some assumptions addressing the situation in which the rainfall









intensity is less than the infiltration capacity of the soil are focused. The general Mein and

Larson equation can be described as:

f = R = K + KS AS_ / F

Where,

Say : the average suction at wetting front

Fp : the cumulative infiltration at the time of surface ponding

R : the rainfall intensity

By considering the relation between rainfall intensity, infiltration capacity, and saturated

hydraulic conductivity, the Mein and Larson equation can be written like as:


mrainlntensity (t ind~ i ll be 0) if rainIntensit'
infiltrationRate= f < t i if Ks tinfiltrationCapacity .:.tei-' ie
Ii Iti atr ,io I-',-Capacity ifKs \kWhere
Ks: Saturated Hydraulic Conductivity
t: time in hours
tpoding : Ponding time when rainIntensitv> i i Ir6 tin' r!:' 3p 3i: ity
i]i, iIri tI:,aiC apac Ty-i: decrease as t increase


Morgan et al. assumed that no ponding occurred in the sandy soil (e.g. for Candler soil, Ks

is 25 cm/hour) because rainfall is less than 5 cm/hour (except during hurricanes) and irrigation

rates are typically 0.315 cm/hour or less. It was also assumed that runoff does not exist since the

rainfall and irrigation intensity is usually smaller than Ks. Thus, the infiltration rate equals the

rainfall rate and the amount of infiltration water equals amount of rainfall rate multiplied by time

t. If rainfall rates are larger than the irrigation rate, the infiltration rate equals the rainfall

intensity. Otherwise, it equals irrigation intensity. Thus, the Mein and Larson equation is

simplified for such rainfall and irrigation conditions. The condition term for comparing the rain









intensity with the saturated hydraulic conductivity was simplified to comparing the rainfall

amount (represented by PWID4) with the irrigation amount at soil cell (represented by CIAM)

for creating the average infiltration rate of soil profile. To apply the world system to the model,

additional condition term s describing whether current soil profile exists in the irrigated-area or

non-irrigated-area were added to the model. With the above result, an equation for the infiltration

rate of soil profile was formed in Figure 4-5. CIAM is the cell irrigation amount, and i is the

profile type, and numbers (1, 2, 3 and 4) represent the profile type, a non-irrigated & dry-

fertilized area, an irrigated & dry-fertilized area, an irrigated and non-dry-fertilized area, and an

non-irrigated and non-dry-fertilized area. SAKtop is the saturated hydraulic conductivity at

surface, and PWID4 is the soil profile water input amount from rainfall effect. IAM is the

equation calculating irrigation amount from irrigation duration, and IDur is the irrigation

duration time.

Examples of CWMS Model Implementation

Concepts developed above were entered into the ontology system. The geometry relating

soil cell and profile, soil water redistribution, and root density are given below as examples.

Soil geometric dimension

Basically, a profile is determined by the distance from the trunk of a tree to three root zone

radii (75, 125, and 175cm). Other profile boundaries are the irrigation diameter and the dry-

fertilized area. Depending on the irrigation type (360 degree or less than 360 degree), soil

profiles can be divided into irrigated-areas and non-irrigated areas. Irrigation and dry-fertilizing

events are assumed to be conducted in a soil cell area except two equipment drive paths between

tree rows. Therefore, the two drive paths are always considered as a non irrigated & non dry-

fertilized area (NINDF). An irrigated & non-dry fertilized area (INDF) is an irrigated area in the

drive paths.













Equations Ei..t :1: Units


I M A 123 &^U (0u)[0i fi AV -X 0>
61 A r- u I


EETI


Active Inactive All
Infiltration Maximum Depth Eq A
Infiltration Maximum Depth Eq -

Irrigated Profile Radius Matrix
ayer Depth at Specific Layer
ayer Number of Wetting Fror
ayer Number of Wetting Fror v
< Ne


[ New ]


Equation Equation ID


Rlntensityt


nfat RIntensityt
InfRate =
Irrlntensity

0


(i =1 v i = 3) PWID4 >0


(i = 2 v i = 4) PWID4 s CIAM


(i = 2 v I= 4) A PWID4
otherwise


Where,


1.75xR
Rlntensity =min ( O. xSAKtop, )
3


PWID4 = R x(1 shadeAreaRatiox(1 shadedRainRatiot))



CIAM = IAM(IDurt)


Figure 4-5. Morgan's infiltration rate model in the CWMS model


2 2XRZR1
3 2XRZR2 NPIDF =
4 2xRZR3
1 otherwise


NPNIDF =


4 WD- 2XRZR1 v WD-2xRZR2 WD-2xRZR3

5 otherwise

4 NPIDF WD = 2 RZR v WD =2xRZR2 v D = 2 xRZR3

5 NPIDF otherwise


NPINF = 1 WD >2xDistToHedgingBoundary
NPINDF otherwise
S0 otherwise


Figure 4-6. Equations of profile number


SpP <360



otherwise









Equations for calculating each profile number and area are defined at the cell, and Figure

4-6 shows equations of three different profile number (profile number of a non irrigated & non

dry-fertilized area is always 1 and it is defined as a constant). NPIDF, NPNIDF, and NPINDF

are respectively symbols of a profile number of an irrigated & dry-fertilized area, a profile

number of a non-irrigated & dry-fertilized area, and a profile number of an irrigated and non-dry-

fertilized area. RZR is a symbol of the root zone radius matrix, and WD is a symbol of the

wetting diameter, and SpP is a symbol of the spray pattern.

A soil layer is one vertical element of a soil profile, and the number of soil layers in a soil

profile is determined by layer thickness and total depth of the soil profile, whose maximum depth

is 200cm. The thickness of soil layers can be grouped, and it is represented as a matrix as in

Figure 4-7. Each row is a layer group, the first column is a thickness of layers in a group, and the

second column is the number of layers in a group, and third column is a cumulative layer number

to that group.

Symbols belonging in concepts, block, soil cell, soil profile, and soil layer, are required

dimensions. The time dimension is a common dimension required by most symbols as one

dimension of the matrix. Symbols in block and soil cell need only the time dimension, whereas

those in soil profile need three dimensions, one for time, one for soil profile type, and one for

soil profile number. Symbols in the soil layer need four dimensions, and they include three

dimensions from soil profile and one for soil layer number. For example, a symbol, historical soil

layer temperature, defined in soil layer has four dimensions, and it is described through the

EquationEditor (Figure 4-8).













5 10 10


LThickRangeArr= 10 6 16


15 6 22


Figure 4-7. Layer thickness matrix


ei,] t -l h r `o '.. 'rS ri', ..



.... J;
L. Ir r- I 5
h.r: ri': I: -rj; .r. P.r t-r l"




0oe iii Ir rt L;. Ir Ir -r c r3rD
!oa F ij r :'.- l 1 c r >





"-c -r;,0 a.id r


_-.-b n aa .rru "_: )- riot,

Historical BSol Layer Teperaure


Historical Soil Layer Temperature


fle TT.pe *. 13-'Ce

ar rPI
"e'ak,- Type 'C lia
(ru i
aI S:' :.. ':e &eLa e' 1-3
- f ? i f
i. 51Zrq Ti-ill


0C -1 W a


* b- LE. -r e.-.0ie l ~ b


C.

[ ~ I| ( tw 2.: ..' .-." i
M.I LC1 f. ..I' l .



*4 T I' .- L r T er-

SynmhI %Arnay


Figure 4-8. Example of dimension description of a symbol in soil layer






Time step


The main time step for the model is daily. Rain event, irrigation event, and fertilizer



applications are assumed to occur at 12:00 pm, but adjustments to the time period of 0-12:00 pm


and 12:00 pm 24:00 pm are controlled by using some time flag symbols. The minimum period


of simulation which can be executable is 14 days. The first 13 days (for 14 days calculation) or


first 5 years (for 20 years calculation) are used to initialize and stabilize values of symbols whose


initial condition are not known at the simulation starting time. For example, the initial values of



soil water content is usually unknown, but assumed to be a default value and computed over the


13 days (or 5 years) to initialize the values.


--I


Eq~~m;r*a..tw. Edt- CWM .H My













Equations Symbols Units

Symbols Symbol
umber of Non-irrigated Profiles ^
Dverlap Between Row Arc Area Symbol DeFinition |.-. Symbol ID
overlap Between Row Arc Half A
verlapInRow Arc Area I i f B U V ;- =- 1 4-
Dverlap In Row Arc Half Angle
representativee Number of Trees English Spanish German
:esentative Amount of N Applica Total number of time steps for simulation It equals the number of A
saturated Hydraulic Conductivity days user selected (current day is ending day) plus the number of
et Point Depth
shade Top Soil Temperature days predicted for future irrigation schedule
Today Number of Time Step
op Layer Soil Temperature Daily Enable Index Symbol : select SymbolID translentable: NO
total Area of Irrigated Source Type : OEquation type general / nit value: I
Total Area of Non Irrigated
Total Number of Profile Types ( Database : Condition #
Total Number of Profiles
O Constant Fo 1
l >I ; ] Value Type : ( Scalar O Matrix : Dimension 1 v as constant v
New [ No ] [Description]
Delete

Show all Equations/Symbols For:
CWMS soil cell
O CWMS

Save




Figure 4-9. Total number of time steps symbol, t


There are two concepts related with time. One is Total Number of Time Steps for


assigning system time dimension size, a value that is stored in and obtained from the database


(Figure 4-9). The other concept related with time is Index variable of Time Step, t, which is used


as the general variable for the current time.


Root density


As shown in Figure 4-10, root density is represented as a matrix from the model for each


soil depth and root sections based on root density distribution as a function of tree size (Morgan


et al., 2006a). A column is a root section and a row is a soil layers group to a specific depth.


Following Morgan's model, the root horizontal area is divided into four sections (0-75 cm, 75-


125 cm, 125-175cm and 175cm-boundary), and 10 soil layer groups are used. The first 6 rows


have thicknesses 15 cm and last 4 rows have 30 cm thicknesses, which are calculated by


proportional equations based on the root density value of 6th soil depth group. RD is a symbol of













the root density, and cv is a symbol of the canopy volume, and RSL is a matrix symbol of the


root density regression parameters for layers below 6th soil depth group.


The root density equation is designed for specific soil layer depths and root ranges. The


CWMS model utilizes it for an equation of LRD which is the layer root density (Figure 4-11).


LD and LT are symbols of the layer depth and thickness of a layer at specific depths.


1 2+0 023xcv 0
1 1 +0043xcv- 0000880cv2 15
1 45+00022xcv 35
1 3+0006xcv cv>50

03+002xcv 0
03+003xcv- 00007xcv2 15< cv 35

00- 00025xcv 35
022 0 005xcv cv>50

012+0012xcv 0
00019xcv- 00004xcv2 15
035- 0002 xc 35
O0005xc cv>50

007+0013xcv 0
0 06+0016xc- O0002xcv2 15
013+0007xcv cv>35


S007+0008xcv 0 S003+001xc 20
023+0006xcv cV>50


001 xc- 0007 07i cy< 20

0013xc- 006 20 cv< 50

029+0006xcv cv> 50

RSL1 FI


02 +0027xcv 0
026+0024xcv 20
11+0007xcv cv>50


01 +002xcv 0
007+0035xcv- 00000xcv2 15
019 35
002xcv- 081 c >50

001+0012xcv 0
001xc- 0004- 00004 xcv2 15
016+00005xcv 35< cv 50

002xcv- 08 cv>50


S001+0006xcv
001xc 005

02+0 005 xcv


002+0004xcv

S0015xc- 015

045+0003xcv


S0015xcv- 0002

001xcv- 013

012+0005xcv

RSL. T


RSL2 Fi

RSL3 FI

RSL4 Fi


0
15r
c >50


0
15
c >50


14
15
cv>50


RSL2 T-,

RSL3 .

RSL4 .


0005xcv- 001 2 0054xcv- 08
002- 00075xcv+00007xcv2 15 1 0002xcv
0 26+ 001xcv 35 c'v 50
055+0007xcv
05+0005xcv cv>50


00053xcv- 0008 15
0 00xcv- 004 cv>15





0 002xc 0002 1
00043xcv- 0037 cv>15


S00013xcv
0 012 cv- 016

09 +0005Xcv


1
15
c >35


0001xc- 00008 08
0013xcv- 018 15
001xc'v- 0075 35
022+0004xcv cv>50

S0008xcv- 00009 12
0009xcv- 012 15
0006xcv- 0005 cv>35

RSL FF
,I


RSL2 F.I

RSL3 F I

RSL4 FF
4 i'


148
32
cv 50


S0005xc- 0125 cv>25
S0 otherwise





00002xcv- 005 cv>25

S0 otherwise


0 0018 cv- 0045
0


S0001 x
0


cv>25

otherwise


0025 cv>25

otherwise


0 0008xcv- 002 cv>25

0 otherwise


RSL FI

RSL2 F-

RSL3 Fi

RSL4 Fi


Figure 4-10. Root density matrix





Equations | Symbols || Units |
a *S A I:;1 | I,,,--',,1'"="= .-| i. | vi|- |l >i =: :|

Sive... InaItve All Eqa Equation ID
ayer Dained Water by Wate Ir
er Drained Wate by Water I
aer Drained Water by Water Ir
ayer Drained Water by Water Ir
aer Drained Water by Water Ir 5(RD,
ayer Dained Wae by Water Ir RD211 R 1i
ayer gained Water by Water I,
ayer Dry Flag Equation RD RD
ayer Ideal Needed Water Amoul 1 II-21
ayer Ideal Neded Water Amoul 2
ayer Initial Soil Temperature Eqi
ar Maximum Damping Depth E-
Nitrate Initial Value EQUAT 211 1 21
-ayer Roo. Density Equation BK
ayer Soll Temperature Equation 211
yaer Th lkness Equation
RDII RD311


.LRD(Ij) = D311

: : .: ,R D 411
RD5II

S New ]RD611
Delete
RD711
Show all Equationsymbols for
* CWMS soil layer s811
S)CWMS

RD1 il
________________________- ^ ________ u ________ I __________


Figure 4-11. Layer root density equation


Water dynamics


The water balance model consists of rainfall, irrigation, evapotranspiration, and infiltration.



As stated previously, surface runoff and subsurface lateral flow are not considered due to the



high saturated hydraulic conductivity of sandy soils. Basically, the water budget calculation is



based on the tipping bucket model, enhanced by considering the effect of the delayed soil water



drainage which is caused by the soil hydraulic conductivity during one daily time step.



Rainfall and irrigation are water inputs into the system, and it is assumed that they are



applied at the first layer. The symbol Profile Water Input (PWI) in the soil profile module



represents input water amount from water resources into the soil profile. According to the



CWMS model, it considers the complicated nutrient management processes by calculating



different sources of water and nutrient separately. In the equation for PWI, sources of water



inputs are distinguished as 4 different types depending on the considering event: irrigation



(water), N application, post N application, and rainfall. Since the system time step is daily and


I


[ Equ--n r:.i-- C-. -ii. .-- Q CM) cx)


LD +0 5 LT <9 375

9 375 LD 0 5 LT< 13125

13125 a LD+0 5 XLT< 17.075

17875 5 LD + 0 5 xLT< 20.625

20 625 a LD+0 5 xLT< 26.25

26 25 LD +0 5 XLT< 33 75

33 75 L LD +05 LT< 45

45 LD + 05 XLT< 60

60 LD + 0.5 xLT< 75

75 LD + 0.5 x LT< 90

90 LD+ 0.5 LT< 120

120 LD+0 5LT< 150

150- LD+05xLT< 100

180s LD +0 5XLT









there is no sub-time step, symbols are created for every stage and distinguished by subscript

number. These are four-drain process named. Figure 4-12a shows related PWI equations (PWI1,

PWI2, PWI3, and PWI4), where it is assumed that irrigation and N application are applied to just

irrigated-area (i=2).

The symbol WI (Figure 4-12b) is the layer water input, and the amount is determined by

the difference of amount of input water into a layer from environment (PWI) or from the upper

layer (DW) and the amount of layer evapotranspiration (nowET). For the four-drain process,

calculation of layer water input is conducted for every stage separately. The symbol WIflag is for

indicating the status of the remained water drain by the hydraulic conductivity.

A layer water content is represented by the symbol WC whose amount is calculated by the

difference of increasing amount (WCpos) by drained water and decreasing amount (WCneg) by

evapotranspiration during the previous time step. At the start of simulation all layer water

content has same amount calculated with constant depletion level. For the four-drain process, a

layer water content at time t would be a WCD4 of previous time step which is the layer water

content applying irrigation, N application, post N application, and rainfall (Figure 4-12c).

The CWMS model enhanced the pure tipping bucket model by computing the wetting front

moving speed during the simulation, whereas the pure tipping bucket model assumed that water

moves the entire layer depth in one time step. The model assumed that all irrigation, water with

N application, water with post N application and rainfall occurs at noon and has a maximum of

12 hours to move through the soil on the first day after irrigation, water with N application, water

with post N application or rainfall. And, the model calculates the wetting front speed (Figure 4-

13) and the time for which the wetting front travels a layer thickness, thus to obtain the layer

index of wetting front at the end of the day and let unfinished drained water continues to drain at

















next day. WFSP is the layer wetting front speed, and it can be calculated by the symbol of the



infiltration WFSP (Figure 4-13b) or the symbol of the hydraulic conductivity WFSP (Figure 4-



13c). Figure 4-13d is the equation for hydraulic conductivity.







P PWIWIDI + PWD2 + PVD3 + PWID4


PWID1 CI i2





PWID3= = A(posttplicaon) i 2
0 otherwise


PlD2 IApreApplicaion +withApplication) i ) 2
PWIDD2 -
0 otherwise



PWID4-Rx(l shadeAreaRaio x(1 shadediRainRaio,)


WID1 +WID2 +WID3 +WID4

PWVI nowETxLT PWVI nowETxLT

PWI otherwise

DWk- 1- nowETxLT DWk-

DWk 1 otherwise


DWk- lt- 1- nowETxLT DWk_


DWk 1 t 1 othemi

SDWkl nowETXLT DWk 1


DWk- 1 otherwise


DWk- t- 1 +DW- 1- nowETxL-


DWk- It- 1 +DWk- 1


DWk lt- 1+DWk 1 nETxL


DWk- It- 1 +DWk-

0


FourDrainPerDayFlaa


nowETxLT A PWI >0





lt 1 nowETxLT

se


Wlflag =DF_12hrTdyAir v (k=1 A HCFlag =0)




k >1 A (Wiflag =F_12hrTdyAbvLyr HCFlag =)




k>1 A t>1 WIflag=DF 24hrYstrdyAbvLyr


nowETXL


k >1 A Wflag = F_24hrTdyAbvLyr


S DWk- lt

otherwise


T DWk- t


otherwise


+ DWV




1+ DWk


>nowETxLT




SnowFTxlT


otherwise


k >1 A t>1 A Wlflag =DFYstrdyl 2hrTdvAbvLyr




k > A t>1 Wflag= DF_Ystrdy24hrTdyAbvLyr


otherwise


12hrTdAirv (k=1 HCF ag=0))

ag 1 DF_12hrTdyAbvLyr v HCFag = 0)

g =DF 24hrYstrdVAbvLyr

g DF_24hrTdyAbvLyr

1 xLTA Wflagt 1 =DF_Ystrdyl 2hrTdy

SxLT A WVIfagt 1 DF_Ystrdy24hrTdy


Sma-(wpo 1 WCnegt 1 PWPWC) otherwise C




Figure 4-12. Water balance equations: A) profile water input amount equations, B) layer water

input amount equation, and C) layer water content equation


WCD4 1

WWC(d)

WCpost_

WCpost


WC 1= { WCpost_

WCpost

fCPtr
WCpost

WCp-os


t >1 i FourDrainPerDayFlag 1


-1

1 A PWI1_ 1>nowET

>1 A k>1 A DWVk It

>2A k>1 DWk_ It




> 2 A k >1 A DWk 1 t

>.2 k>1 DWk 1 t


S=DF

S(Wifla

SA WIfla


-A Wfla

> nowETt

> nowET,


1xLT (VWflagt


2 >nowETt 1 x

ro1 L
2>nowET- 1xL

I >nwETt 1XL


2 +Wk- 11-1

2 DWk 1 t 1


AbvLyr

AbvLyr


otheMise











r LT
InfWVFSP LD +- < InfMaxDep
WFSP = 2
HCWFSP otherwise A


InfRate
In1VVFSP =
SWC wc B

S (min( ))- KTheta(WCbeloWF)
.1 1
2 5 k< NLi rin(' I .. i. DWV >0
mi( I' .I ) WCbelow~ F I 1 k-
I 1

I (min( ,i 1.1 ))- Kheta(WCbelowWF)
1 HCWFSPt 1------>1 A 25 k< NLA min(' ) .I.. II A WFLNfag 1(k)=- WFLNfag (k)i WFLNflag (k 0
HCWFSPF- rmn( i VC ,L eow F I 1

Khea(SWC) KTheta(WCbeloNVF)
k=1A PWI>0
SWC WCbelo F

0 otherwise C


HCPm 2
WC PWPWC0.5 (1 WC PWPWC HCPm) SAK
KTheta(WC) =min( SAKx( )W x -- ( -) 1
SWC PWPWC SWC PWPWC 2 D


Figure 4-13. Enhanced hydraulic conductivity related equations in CWMS model: A) Wetting
front speed equation, B) infiltration wetting front speed equation, C) hydraulic
conductivity wetting front speed equation, and D) hydraulic conductivity equation

Nitrogen balance

For nutrient management, especially application of nitrogen by irrigation (fertigation), the


CWMS model provides processes for calculating nitrogen balance followed by transformation of


nitrate and ammonium. Transformations are composed of complicated sub-processes and are


affected by input and flow (drain from above layer) amount of nutrient and water between layers.


The model assumed that there are four nutrient flow processes by four drain events: drain by


rain, pre-irrigation, during-irrigation, and post-irrigation. After four-drain steps, transformation


processes is applied to ammonium and nitrate following in order of volatilization, uptake of


ammonium, uptake of nitrate, and nitrification.


The amount of nitrogen input into a soil profile is an equation formed with the dissolved


ammonium (NH4) and nitrate (NO3) nitrogen amount of dry fertilizer by rain or irrigation and











fertilizer amount during the fertigation. Nitrogen drain amounts (ammonium and nitrate) of a


layer from the above layer are represented by symbols, Layer Drained Nitrate and Layer Drained


Ammonium, and they are calculated by equations of soluableRate. For uptake amounts of


nitrogen, passive uptake of ammonium and active/passive uptake of nitrate are formed as an


equation. Also, the model assumed that the uptake process of ammonium occurred after


volatilization and nitrification follows a nitrate uptake process, and they are utilized by using


separate symbols for each stage.


The amount of nitrification is represented by symbol NIT, and its equation includes the


rate of maximum nitrification as a function of ammonium content (NH4NC3), maximum


nitrification amount at a time (NITVmax), half-saturation constant of nitrification (NITkm), and


soil moisture factor (NITwf). Nitrification is allowed till 80% of current ammonium content


(Figure 4-14).

nE ito s
Equations symbols Units
S S A 12 4 'll I I>
Equation Equation ID
,trifica tio n V m a x o f r .- -. i: : ,'- I i
itrificationVmaxof r -. i NH4NC3xLV
trificatlon Water Factor Equation x NITVrma
WA
i ftrif'ato for Layer Equation BK Jul16207 NIT =mn( 0 8xNH4NC3, NH4N 3 xNITwl)
NH4NC3xLV Krn
N.. I W A



Figure 4-14. Layer nitrification equation


Equa mbolrs Unis

_-- Equa1ion EquaI;on TDI |
l ocaton Equation Eq
s s oIn ratln Dept h Equaton a VolPNH4NIn qmt NApplyElapse l
_ln tWSFtY qmNAppRateFactorl NApplyElapset 0
depletion to Irrigation Depth I umulativeVOLA-l 1 T -n 1 1 -V S t qrN pR 9F co ,
Rained Water Content at Merged of a Specific Layer EQL
S 0 otherwise

SNem I I-f


Figure 4-15. Accumulated profile volatilization equation









Volatilization (cumulativeVOLA) is created at the soil profile module, since volatilization

occurs till 5cm soil depth from the surface. It represents a cumulative volatilization loses over a

day, which is determined by the relationships between ammonium content (VolPNH4NIn), days

counting from the volatilization starting time (NApplyElapse), percentage of maximum

cumulative volatilization (qm), and site-specific temperature/wind parameters (Figure 4-15).

Application Implementation

The CWMS model is used to implement a CWMS application program for use by grower

that utilizes crop, soil and weather data. A CWMS application consists of the automatically

generated simulation code and a graphic user interface. The generated simulation code is plugged

into the graphic user interface without further coding required.

A graphic user interface allows growers to interact with the system through three phases:

the setup, the irrigation scheduling, and reporting. The setup phase is for configuring the cell and

block information for a particular grove (Figure 4-16a) and for describing simulation site,

start/end data, and resource location (Figure 4-16b). The simulation period is separated into an

initializing period and a simulation period. For a long term simulation, an initializing period can

be longer than 14 days which is the default initializing period for farm irrigation scheduling.

The irrigation scheduling phase, shown in Figure 4-17 provides irrigation scheduling

information to growers. By default, it is based on a 14-day simulation followed by a 3-day

prediction period (the simulation period can be extended) to provide immediate term

recommendations on irrigation rates. In order to plan a strategy on irrigation scheduling,

simulation system tested for full seasons (over 250 days per year and over 30 years).
























Prn114t-'p



-lfl~


.1 n- p.



I -.* -' --~l
nb,.tY-


L. _- I nI ...* u ...a 1 c .J F


















Figure 4-16. Setup Phase: A) grove/block and B) simulation period










iqnkpac. Menu : ..s..ac'



"- -. W ..











.. 1q lrsonn.,iuan












SI. nH.v.... E I It.r I WIa l | 1 Ia:n
Descrea


Figure 4-17. Irrigation scheduling result


a~. .0 i t~ I *atlC A I~V~



t1 I -' h.- -~~ r"~(11


Uln lulr *rrlllurcl~Y~nn ~YI

n~l I~~~






Il~n~n~


~m~n~n~n~n~n~ n~


R-IICII
Ynlu CII~CL
r~~L


LLUU


~- rc-r
r U-~-ll Y1
Il~r~- i~
L~ll~l-l yl




I~U IYrUiI

ICU*YI*II
UUllic~I*C

Illr I~-11III-

III LUI ~-
~ I~ I-;


P~cl I~ He
E1~r*-~ UlIIl

*ICIL9wl*rr

O-olr


rOsav,


Vul KiL. *--








*TCRacma :rwmn mi | 1<*

SmmlaMn fIfNI
u* Curet m :lIut) ort~it < --]



















Projea Maode HTop



WrKspaceT MeI: WorKspac:


C work
D Irngabon Schmdwdu
C3 Scenaro
Setup
























dloil minmkuin al

ofspedi dl


rl"latOO schedultna i Scenaro Setup


BlInklD. iCt'enTlil&mdiol01 DAr. I II 231, 6


Ara a jllq lJedJAIM


I ,


I.I





I -~ar I Ip ~lbRp


I lTdi l.nLengtlh.


RfiAfI Plulle Ipenww Idalm


SUo92J51l71 ExcelRsurn


,,. A;
















v1 ..I -


.6 I ,
,', ,,. I -
,.,, ".

,',,' : .I


... .1 ;
,6 ,,
6 ,,










iJI ii I


MUoinrlIyRIiRI O YealIlRepol 1 TioalLawers


Saw veRsi


6 "1.














I I ,


I I. l








-I I


, ModIeVs Eqtxpinnn I


i ...............aD. 7


Prect Mode Help
Experrnewd WIadl

WoUIKspw Menu.

























o rl ", O

Dihr.kan
ofl --tlB Iil
o@ fgrallMoflkn
lplnieir.lal n
*<.>.=**= .....


mWolkspace

tilliacr.Schedulino | SCenuil Selup


Blotl Co arlwf lairmulo I- Picfrl t Iliaie 4vne 45
II I -O h l Pi.ola r. i 1 IIp



Water Content



r. ./, /. -













II
coo

Oala
n n









sIrimnv [ inpullnri Darnpuf Morthfrnpren YSarlReof Ia .o LayerL Madelv.Erpelnt


Figure 4-18. Simulation results: A) table and B) graph














62


i I


. i.-r r i l .i II .i P .]r. P. l









Detailed simulation results are provided in the form of daily, monthly, and yearly reports

(Figure 4-18). The daily report contains each layer's root length, water content, nitrate and

ammonium content, and soil coefficient. Data can be browsed by selecting a specific date and

profile type. The monthly report shows data for a particular month including irrigation interval

days and duration, evapotranspiration, crop coefficient, soil coefficient, water and nutrient

leaching amount at 2 m depth and irrigation depth, rain, and irrigation. The yearly report

provides the monthly total value of irrigation, rain, water and nutrient leaching amount at 2 m

depth and irrigation depth, and fertilizing amount. Water, nitrate, and ammonium content

contained in the daily report can be displayed as a chart.

Model Extension

The CWMS Java code was generated automatically by the SimulationEditor and the

EquationEditor, and can be used as a software component that can be used independently of the

model building environment. The compiled Java code can be easily connected to other user

interfaces or simulation system. For example, the Watershed Assessment Model (WAM) which

is used as a part of a larger FDACS BMP simulation, adopted the CWMS model as a sub-

component of land use in citrus production. The CWMS model provides information about

leached water and nutrient amounts to the host simulation system.

A key issue was how easily the CWMS model could be modified for integration with

WAM. A water and nitrogen balance model needed by the WAM was required to use several

new soil types, and modifications to the model were made to support these new parameters.

Using the EquationEditor, new symbols for soil types and parameters were created and existing

equations were modified by replacing old symbol and by adding new formula, and some new

equations calculating required values by the WAM were added into the model. This was all









accomplished with less than 10 hours of work including creation of required file input/output

protocol by WAM (that took about 80% of total work hours).

Model Performance

Simulation of the CWMS model containing approximately 700 symbols and 500 equations

uses different numbers of equations depending on the combination of processes (e.g. use process

of tipping bucket or effect of hydraulic conductivity). To test the performance of the CWMS

model for the maximum number of processes, the combination including hydraulic conductivity,

four-drain process and moving wetting front is chosen, which consists of 330 equations.

Applying dimension size to these equations, total number of calculation at double precision level

is 330,000 for 14 days period. It takes approximately 5 seconds by a computer with Intel Pentium

1.7GHz CPU speed and 1G RAM. Calculation time for 1, 10 and 30 years are 24 seconds, 5

minutes and 15 minutes, respectively. For an extension for WAM which contains process of

tipping bucket and moving wet front, it takes 5 minutes for 30 years simulation. In order to

validate the accuracy of the model, Morgan (Morgan et al., 2006a) compared the observed water

contents at soil depth 10, 20, 30 and 50cm from foil surface with 2 years simulation result for

two different sites. According to the validation result, R2 values were varied in the range

between 0.46 and 0.75, and for soil depth 10 and 20 cm it gave 0.7 R2 average value which is

higher than another two points. The validity of the model depends on the accuracy of the

equations and parameters, and is impacted by the quality of the model implementation platform.

Model Sensitivity Analysis

Sensitivity analysis is useful method to guide model development as well as to understand

model behavior when the model is under construction. This method is applied to the CWMS

model for identifying most significant factors to Cell Water Amount (CWA) and determining

their interaction, which has been implemented at the SimulationEditor as an additional feature.









The procedure consists of two steps: a factor screening with Morris randomized OAT design and

a global sensitivity analysis with screened factors.

At the factor screening step, a variable (e.g. CWA, a sum of water amount in the soil cell)

is selected to analyze the response of the system to the CWMS models' water balance processes.

Variables related with the tree characteristics, water inputs (rain and irrigation) and water

movement are selected as an input factor (Table 4-1).

Table 4-1. Input factors related with water input and hydraulic conductivity
no Symbol ID Symbol min max unit
1 Canopy Volume cv 3.2 12.6 m
2 Emitter Flow Rate EFR 20 70 L/hr
3 Hydraulic Conductivity Parameter n HCPn 2.6 4.6
4 Initial Depletion Dini 0.09 0.19
5 Irrigation efficiency IE 79 89 %
6 Readily Available Coefficient KRA 0.03 0.13
7 Wetted Diameter WD 100 500 cm


The selected variable used for tree characteristic was canopy volume (CV, volume of a

citrus tree as function of tree age). Selected water input variables were emitter flow rate (EFR,

the volume of water discharged from the emitter within a period of time), initial depletion (Dini,

the initial condition of depletion), and irrigation efficiency (IE, the percentage of water pumped

into the irrigation system that actually gets distributed by the emitter) and wetted diameter (WD,

the diameter of the irrigation emitter). The selected variables for water movement were soil

hydraulic conductivity (HCPn), was assumed to same for all soil layers, and readily available

coefficient (KRA, coefficient used to calculate the readily available water content for uptake for

uptake).

The minimum and maximum values of factors six different variables in Table 4-1 were

applied. Six orientation matrices are generated according to the Morris factor-screening design,

and the respective elementary effects for 7 different factors per orientation matrix are estimated









from the simulation response for CWA. Following Morris OAT design 8 simulation

configurations are generated for each of six orientation matrices. In each orientation matrix, the

first row represents the base case (configuration) and the remaining 7 are used to determine the

elementary effects for all 7 factors involved.

After comparing the elementary effects of 7 factors for 6 different trajectories and the

corresponding mean and variance of the distribution, factor 7, the WD, appears significantly

separated from the other factors and it means that the wetted diameter dominate the simulation

result. The WD increases value of CWA since it determine direct water input amount from

irrigation event, but its effect compared with other factors was explained before this analysis.

Whereas, factor 5 and 6, IE and KRA, has lower mean-variance relation value than other factors,

so that model results are less sensitive those two factors were screened before sensitivity analysis.

At the sensitivity analysis step, 35 complete factorial design makes 243 scenarios with the

5 factors selected by the Morris OAT screening-factor method and 3 different factor levels. The

analysis of variance on the simulation result of CWA was performed, which included

interactions between two different factors. The results presented in Table 4-2 shows the sum of

squares and the sensitivity index which is calculated by dividing the sum of squares with the total

variability.

From the result of sensitivity analysis, the CWA was apparently governed by the WD, and

Dini with a significant impact on the system than other factors. The HCPn was more sensitive

than the CV and the EFR. The EFR appeared as the least sensitive factor among the main factors.

The WD related the water input to the amount of irrigation. Thus, water balance could be

affected significantly by the irrigation amount when there was less rainfall. For the interaction

between two factors, the interaction of the WD and the CV was more significant than others, and









interactions with the HCPn had relatively larger value than other interactions because it

contributed to increase water amount in deep soil layer. The CV affected rainfall into the system

by blocking direct rain, and limited amount of rain could reach to soil.

Table 4-2. Sensitivity analysis result including main effects and two-factor interactions
Effects SS Sensitivity Effects SS Sensitivity
(x\ O) Index (\1' i') Index
cv 7,915 0.00102 cv*WD 7,847 0.00101
EFR 805 0.00010 EFR*HCPn 493 0.00006
HCPn 2,084 0.00027 EFR*dini 82 0.00001
dini 120x103 0.01545 EFR*WD 171 0.00002
WD 7.,i4 \ 0.98102 HCPn*dini 63 0.00001
cv*EFR 145 0.00002 HCPn*WD 914 0.00012
cv*HCPn 993 0.00013 dini*WD 99 0.00001
cv*dini 19 0.00000 residuals 5,919



Sensitivity analysis result provides information about impact factors and related factors

impacts. It may be useful to reconfigure model parameter and to create other models using these

symbols.

In Chapter 4, the CWMS model developed by Morgan et al. (Morgan et al., 2006a; Morgan

et al., 2006b) is implemented using ontology-based simulation methodologies and tools covered

in Chapter 3. Soil geometry structure is designed with 4 concepts, soil block, soil cell, soil profile,

and soil layer and their dimensions are defined with index concepts describing array size. With

the existing mathematical models symbols and equations are defined and entered into the

ontology, and they formed a model base containing symbols and equations and structuring

relation between them. Model performance is tested under two different simulation conditions,

and using a sensitive analysis tool, critical input factors and the associated factor relations are

revealed for water amount in the soil system. Through these processes the CWMS model is

created and executed efficiently by the graphic interface program.









CHAPTER 5
SUMMARY AND FUTURE WORK

The following methodologies were developed utilizing ontology-based simulation

techniques to build mathematical models.

1) The EquationEditor includes a symbol dictionary for entering symbols appearing in

the equations along with their definitions and units. Symbols are defined by

specific concepts. Equation are rendered visually using classic mathematical

notation, but internally a hierarchical data structure (tree) is used for storing

operators and symbols. The equal operator is the root node of the equation tree.

Operators (like + and -) used in the equation become a node in the tree with child

nodes being additional operators or symbols. The Equation Object Model (EOM)

used in the Equation Editor is a collection of basic objects which represent

information describing a mathematical expression and defining data type of

attributes, and which communicate with ontology-based database system to retrieve

data.

2) The SimulationEditor incorporates the EquationEditor and is designed to represent

the structure of dynamic systems using graphic elements. The SimulationEditor

also contains facilities for automatically generating and running simulations and

providing reports. The SimulationEditor provides a graphic user interface to create

and maintain a simulation system; a structure design interface for the simulation

system, a simulation control interface, a simulation result reporting interface, and

some additional interfaces including a math markup language generator and a

statistical model analyzer.









These methodologies were applied to develop a model of the Citrus Water Management

System. Approximately 700 symbols and 500 equations are conceptualized and stored in

ontology database using the SimulationEditor and EquationEditor. A Java program for running

the simulation was generated automatically from the modeling environment and incorporated

into both a stand-alone application for grower and WAM. The modeling environment provides

adequate tools to create and modify models without any programming knowledge. From the

results, it can be concluded that ontology-based simulation offers a significant improvement in

the methodology for building, publishing, and managing model.

For the next step, it will be useful to study model reusability within the existing model base

of the CWMS model. Issues on model reusability are related with the scale of models in the

problem domain, which requires the consistency of the temporal and spatial scale to guaranty

compatibility in models (Leon et al., 2002). For a similar scale domain problem with the CWMS

model, such as models using different soil profiles, limitation placed on reusing these sub-

models are that they have rigid and sophisticate connections with other processes such as a

hydraulic conductivity which has strict requirement for spatial scale. On the other hand, from an

extension study of the CWMS model, WAM, in which two different models cooperate

independently keeping the internal scale of the model may cause significant inefficiency of

simulation time. There are needs for an efficient way to organize and classify models and

processes within a model base and for a flexible way to constructing a set of simulation model

for different project and modeler by switching with a different process. Diverse level taxonomic

categories may help the model base to be handled more intuitively.

The data reusability could follow the model reusability. Especially, relational databases are

important sources of data for simulation model, but it is expensive to identify existing data, to









determine the exact format of data and to use data in a model. There has been research to convert

directly relational databases directly to ontology using a mapping language (Barrasa et al., 2004)

and to transform/service various heterogeneous database sources including database as

ontologies (TopBraid, 2003). It is difficult to provide a formal way to use databases as ontologies

since security and restriction level of data varies and case depends on the domain. Therefore, it

will be useful to study sharing databases as a form of ontology for a simulation model in the

agriculture domain to build an efficient simulation model network. This study can include

methods to publish/search databases, negotiate/communicate automatically to get wanted data,

and provide a protocol for networking within the domain. Finally, a visual environment will be

required to enhance the reusability of models and data and ontology since they have a

complicated relationship and structure. There are some studies on solving complexity problem

in displaying ontology visually (Bosca et al., 2005; TouchGraph, 2005), but it needs to develop

methods focusing on the relations in model and data.









LIST OF REFERENCES


Alam, F. M. and K. R. McNaught, 2004, Using Morris's Randomized OAT Design as Factor
Screening Method for Developing Simulation Metamodels, Proc. Of the 2004 Winter Sim. Conf.
Vol. 1:930-938

Athanasiadis, I. N., A. E. Rizzoli, M. Donatelli and L. Carlini, 2006, Enriching software model
interfaces using onotlogy-based tools, iEMSs, Burlington, Vermont, July 2006

Ausbrooks, R., S. Buswell and D. Carlisle, 2003, Mathematical Markup Language (MathML)
Version 2.0, http://www.w3.org/TR/MathML

Barrasa, J., O. Corcho and A. G'omez-P'erez, 2004, R20, an Extensible and Semantically Based
Database-to-ontology Mapping Language, Second Workshop on Semantic Web and Databases
(SWDB2004). Toronto, Canada. August 2004

Beck, H. W., 2007, Lyra ontology management system,
http://orb.at.ufl.edu/ObjectEditor/index.html

Beck, H. W., L. G. Albrigo and S. Kim, 2004, DISC citrus planning and scheduling program,
Proceeding of the Seventh International Symposium on Modelling in Fruit Research and Orchard
Management: 25-32

Benjamin, P. C., M. Patki and R. J. Mayer, 2006, Using ontologies for simulation modeling,
Winter Simulation Conference 2006: 1151-1159

Booch, G., J. Rumbaugh and I. Jacobson, 1997, The Unified Modeling Language user guide,
Addson-Wesley

Bosca, A., D. Bonino and P. Pellegrino, 2005, OntoSphere: more than 3D ontology visualization
tool, SWAP 2005, the 2nd Italian Semantic Web Workshop, Trento, Italy, December 14-16,
2005, CEUR Workshop Proceedings

Buswell, S., O. Caprotti, D. P. Carlisle, M. C. Dewar, M. Gaetano and M. Kohlhase, 2004, The
OpenMath Standard 2.0, http://www.openmath.org/standard/om20-2004-06-30/

Cuske, C., T. Dickopp and S. Seedorf, 2005, JOntoRisk: An Ontology-based Platform for
Knowledge-based Simulation Modeling in Financial Risk Management, European Simulation
and Modeling Conference 2005

Donatelli, M., G. Bellocchi and L. Carlini, 2006a, Sharing knowledge via software components:
models on reference evapotranspiration, Europ. J. Agronomy Vol. 24(2): 186-192

Donatelli, M., G. Bellocchi and L. Carlini, 2006b, A software component for estimating solar
radiation, Environmental Modelling and Software Vol. 21(3): 411-416









Eitzinger, J., M. Trnka, J. Hosch, Z. Zalud and M. Dubrovsky, 2004, Comparison of CERES,
WOFOST and SWAP models in simulating soil water content during growing season under
different soil conditions, Ecological Modelling Vol. 171(3): 223-246

Ewert, F., H. Van Keulen, M. K. Van Ittersum, K. E. Giller, P. A. Leffelaar and R. P. Roetter,
2006, Multi-scale analysis and modelling of natural resource management, Proceedings of the
iEMSs, Burlington, Vermont, July 2006

Fishwick, P. A. and J. A. Miller, 2004, Ontologies for Modeling and Simulation: Issues and
Approaches, Proceeding of 2004 Winter Simulation Conference, Vol. 1:251-256

Forrester, J. W., 1971, World Dynamics, Cambridge, MA: Productivity Press: 144

Furmento, N., A. Mayer, S. McGough, S. Newhouse, T. Field and J. Darlington, 2001,
Optimisation of component-based applications within a grid environment, Proceedings of the
2001 ACM/IEEE conference on Supercomputing, Denver, CO, November 2001

Green, W. H. and G. Ampt, 1911, Studies of soil physics, part 1.-the flow of air and water
through soils, J. Agricultural Science Vol. 4: 1-24

Gruber, T. R., 1995, Toward Principles for the Design of Ontologies Used in Knowledge
Sharing, International Journal of Human Computer Studies Vol. 45:907-928

Guarino, N., 1997, Understanding, building and using ontologies, Int. J. Human-Computer
Studies Vol.46, 293-310

Haan, C. T., H. P. Johnson and D. L. Brakensiek, 1982, Hydrologic modeling of small
watersheds, ASAE Monograph No. 5:533

Islam, A. S. and M. Piasecki, 2004, A Stategy for Web-Based Modeling of Hydrodynamic
Processes, EM2004 June 13-16

Ittersum, M. K. v., F. Ewert, T. Heckelei, J. Wery, J. Alkan Olsson, E. Andersen, I. Bezlepkina,
F. Brouwer, M. Donatelli, G. Flichman, L. Olsson, A. E. Rizzoli, T. van der Wal, J. E. Wien and
J. Wolf, 2008, Integrated assessment of agricultural systems A component-based framework for
the European Union (SEAMLESS), Agricultural Systems, Vol. 96(1-3):150-165

Jones, J. W., B. A. Keating and C. H. Porter, 2001, Approaches to modular model development,
Agricultural Systems 70: 421-443

Jurisica, I., J. Mylopoulos and E. Yu, 2004, Ontologies for Knowledge Management: An
Information Systems Perspective, Knowledge and Information Systems Vol. 6: 380-401

Knublauch, H., D. Oberle, P. Tetlow and E. Wallace, 2006, A Semantic Web Primer for Object-
Oriented Software Developers, W3C Working Group Note 9 March 2006









Lacy, L. and W. Gerber, 2004, Potential modeling and simulation applications of the web
ontology language OWL. WSC '04: Proceedings of the 36th conference on Winter simulation,
Winter.

Leon, L. F., D. Lam, S. Hamilton, N. Crookshank, D. Bonin and D. Swayne, 2002, Multi-model
integration in decision support system: a technical user interface approach for watershed and lake
management scenarios, Proceeding of 2002 iEMSs Vol. 3: 306

Lu, H.-Y., C.-T. Lu, M.-L. Wei and L.-F. Chan, 2004, Comparison of Different Models for
Nondestructive Leaf Area Estimation in Taro, Agron J Vol. 96(2): 448-453

MathType, 1996, Design Science, http://www.dessci.com/en/products/mathtype/

Mein, R. G. and C. L. Larson, 1973, Modeling infiltration during a steady rain, Water Resour.
Res. Vol. 9(2): 384-394

Microsoft, 2003, Microsoft Office Equation Editor, http://office.microsoft.com/en-
us/word/HP051902471033. aspx

Miller, J. A., G. T. Baramidze, A. P. Sheth and P. A. Fishwick, 2004, Investigating ontologies
for simulation modeling. Simulation Symposium, 2004. Proceedings. 37th Annual.

Morgan, K. T., T. A. Obreza and J. M. S. Scholberg, 2006a, Characterizing citrus tree root
distribution in space and time, J. Am. Soc. Hort. Sci. Vol. 131: 149-156

Morgan, K. T., T. A. Obreza, J. M. S. Scholberg, L. R. Parsons and T. A. Wheaton, 2006b,
Citrus water uptake dynamics on a sandy Florida Entisol, Soil Sci. Soc. Am. J. Vol. 70(1): 90-97

Morris, M. D., 1991, Factorial Sampling Plans for Preliminary Computational Experiments,
Technometrics Vol. 32(2)

Muetzelfeldt, R. and J. Massheder, 2003, The Simile visual modelling environment, Europ. J.
Agronomy Vol. 18: 345

Noy, N. F. and D. L. McGuinness, 2001, Technical Report KSL_01_05, Ontology Development
101: A Guide to Creating Your First Ontology, Stanford Knowledge Systems Laboratory

Park, M. and P. A. Fishwick, 2005, Integrating Dynamic and Geometry Model Components
through Ontology-Based Interface, Simulation Vol. 81(12): 795-813

Peart, R. M. and R. B. Curry, 1998. Agricultural systems modeling and simulation. New York,
Marcel Dekker.

Raubel, M. and W. Kuhn, 2004, Ontology-based task simulation, Spatial Cognition and
Computation Vol. 4: 15-37









Reddy, V. and V. Anbumozhi, 2004, DEVELOPMENT AND APPLICATION OF CROP
SIMULATION MODELS FOR SUSTAINABLE NATURAL RESOURCE MANAGEMENT,
International Agricultural Engineering Conference (IAEC)

Richards, L. A., 1931, Capillary conduction through porous mediums, Physics Vol. 1: 313-318

Rizzoli, A. E., M. Donatelli, R. Muetzelfeldt, T. Otjens, M. G. E. Sevensson, F. v. Evert, F. Villa
and J. Bolte, 2004, SEAMFRAME, A Proposal for an Integrated Modelling Framework for
Agricultural Systems, Proc. of the 8th ESA Congress: 331-332

Rumbaugh, J., M. Blaha, W. Premerlani, F. Eddy and W. Lorensen, 1991, Object-oriented
modeling and design, Englewood Cliffs, New Jersy: Prentice Hall

Scholten, H., A. Kassahun, J. C. Refsgaard, T. Kargas, C. Gavardinas and A. J. M. Beulens,
2007, A methodology to support multidisciplinary model-based water management,
Environmental Modelling & Software Vol. 22(5): 743-759

Steed, M., 1992, Stella, a simulation construction kit: cognitive process and educational
implications, The Journal of Computers in Mathematics and Science Teaching Vol. 11(1): 39

TopBraid, 2003, TopQuadrant White Paper, http://www.topquadrant.com

TouchGraph, 2005, TouchGraph White Paper, http://www.touchgraph.com/

Wallach, D., D. Makowski and J. W. Jones, 2007, Working with Dynamic Crop Models,
Elsevier









BIOGRAPHICAL SKETCH

Yunchul Jung, hailing from Ulsan, Republic of Korea, finished his schooling from

Haksung High School. He studied agricultural engineering and acquired a bachelor's degree

from Seoul National University, Seoul. He is currently working toward completion of his

master's degree program in agricultural and biological engineering at the University of Florida.





PAGE 1

ONTOLOGY-BASED APPROACH TO SIMULA TION WITH APPLICATION TO CITRUS WATER AND NUTRIENT MANAGEMENT By YUNCHUL JUNG 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 ENGINEERING UNIVERSITY OF FLORIDA 2008 1

PAGE 2

2008 Yunchul Jung 2

PAGE 3

To my advisor, my parents and my girl friend 3

PAGE 4

ACKNOWLEDGMENTS After beginning a study at gradua te school level, it was not easy to get an academic result. I always liked pursuing new topics and learning new technologies for the future. I would like to thank my advisor, Dr. Howard Beck. It was luck y for me to meet and collaborate with him. His ideas and advice stimulated me to achieve my research objectives. His passion for research and intellectual insights were motivating factors in my work. I would also like to extend my appreciation to my cochair, Dr. Kelly Morgan, and committee member, Dr. James Jones for their encouragement and guidance. Finally, I take this opportunity to thank my parents and my girl friend, Kyungmi. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES.........................................................................................................................8 ABSTRACT...................................................................................................................................10 CHAPTER 1 INTRODUCTION................................................................................................................. .12 2 LITERATURE REVIEW.......................................................................................................17 Ontology Based Simulation....................................................................................................17 Model-Based Approach to Ontology......................................................................................19 3 ONTOLOGY-BASED APPROACHES AND TOOLS FOR SIMULATION......................23 Background Technologies......................................................................................................23 Ontology..........................................................................................................................23 Ontology Management System (OMS)...........................................................................24 Model and Simulation Ontology............................................................................................24 EquationEditor........................................................................................................................26 Equation Object Model (EOM).......................................................................................26 Components of the EquationEditor.................................................................................27 SimulationEditor............................................................................................................... ......31 Additional Tools and Facilities...............................................................................................3 6 4 APPLICATION TO CITRUS WA TER AND NUTRIENT MANAGEMENT.....................40 The CWMS Model................................................................................................................. 40 Description of Model.......................................................................................................40 Model Base of the CWMS Model...................................................................................44 Examples of Model Representing Process......................................................................47 Examples of CWMS Mo del Implementation..................................................................49 Soil geometric dimension.........................................................................................49 Time step..................................................................................................................52 Root density..............................................................................................................53 Water dynamics........................................................................................................55 Nitrogen balance......................................................................................................58 Application Im plementation...................................................................................................60 Model Extension................................................................................................................ .....63 Model Performance................................................................................................................64 Model Sensitivity Analysis..................................................................................................... 64 5

PAGE 6

5 SUMMARY AND FUTURE WORK....................................................................................68 LIST OF REFERENCES...............................................................................................................71 BIOGRAPHICAL SKETCH.........................................................................................................75 6

PAGE 7

LIST OF TABLES Table page 3-1 List of operators in EquationEditor....................................................................................30 4-1 Input factors related with wa ter input and hydraulic conductivity....................................65 4-2 Sensitivity analysis result including ma in effects and two-factor interactions..................67 7

PAGE 8

LIST OF FIGURES Figure page 3-1 Concepts and relations in model ontology for simulation.................................................25 3-2 Class diagram of Lyra equation object model retrieving instances of model ontology.....27 3-3 Features of the symbol editor in the EquationEditor: symbol ID, symbol, unit and description of Cell Crop Evapotranspiration.....................................................................28 3-4 Database constraints and array dimension description of a symbol..................................29 3-5 Equation ID and equation of Cell Evapot ranspiration in the mathematical statement editor..................................................................................................................................29 3-6 Unit ID, unit and definition of cen timeter of water in unit editor......................................31 3-7 Structure editor showing a compartmental diagram of a soil-water model.......................32 3-8 Relationships between ontology-based simulation system, generated code and application.................................................................................................................... ......35 3-9 Symbol reference diagram focusi ng on Layer Daily Evapotranspiration..........................36 3-10 Result of generating markup language for the equation A=B+C......................................39 4-1 Conceptualization of soil geometry of CWMS model.......................................................42 4-2 Examples of soil profile areas............................................................................................4 2 4-3 Relationship among equation symbols for water and nitrogen balance............................45 4-4 Taxonomy diagram of the CWMS model..........................................................................46 4-5 Morgans infiltration rate model in the CWMS model......................................................50 4-6 Equations of profile number..............................................................................................50 4-7 Layer thickness matrix..................................................................................................... ..52 4-8 Example of dimension descrip tion of a symbol in soil layer.............................................52 4-9 Total number of time steps symbol, t.................................................................................53 4-10 Root density matrix....................................................................................................... .....54 4-11 Layer root density equation............................................................................................... 55 8

PAGE 9

4-12 Water balance equations................................................................................................... .57 4-13 Enhanced hydraulic conductivity related equations in CWMS model..............................58 4-14 Layer nitrification equation.............................................................................................. ..59 4-15 Accumulated profile volatilization equation......................................................................59 4-16 Setup Phase............................................................................................................... .........61 4-17 Irrigation scheduling result.............................................................................................. ..61 4-18 Simulation results........................................................................................................ .......62 9

PAGE 10

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 Engineering ONTOLOGY-BASED APPROACH TO SIMULA TION WITH APPLICATION TO CITRUS WATER AND NUTRIENT MANAGEMENT By Yunchul Jung August 2008 Chair: Howard Beck Cochair: Kelly Morgan Major: Agricultural and Biological Engineering Simulation in agriculture a nd natural resource management is a popular methodology for studying environmental and agricultural system pr oblems. Traditionally, building a simulation is treated as a software engin eering problem, and simulations are implemented through manual coding in a particular programming language. Problems of implementing a model and developing a simulation system include difficultie s in managing and reusing existing models and simulation system because it is hard to understand the detailed specificati on of the system model when it is written in a specifi c program language. Also, model sp ecification may be lost during the programming process, and it is difficult to maintain documentation describing the system because documents are external to the progra mming process. Visual simulation environments reduce the burden of programming, but there are still problems related to sharing knowledge about the system. An ontology is an explicit specif ication of a conceptualization, which can be used to create a formal representation describing and categorizing concepts and relationships among the concepts in a particular doma in. Ontologies enable shari ng through a common understanding of 10

PAGE 11

the structure of information in a domain, enha nce reuse of domain knowledge, and make domain assumptions explicit by separating domain know ledge from operational knowledge. While ontologies have been used in many domains as a way to represent generic domain knowledge, an ontology-based approach to modeling and simulati on in the domain of agriculture and natural resources has not been well explored. In this thesis, ontology-based modeling a nd simulation methodologi es and tools are developed which can be used by modeler and re searcher to build mathematical models and simulations, and in the process provide a bett er way of representing knowledge about models, improve sharing and reusability, and provide a new basis for analysis of models and model elements. These tools are applied to develop CWMS (Citrus Water Management System) model as a way of evaluating the effectiveness of the proposed approach. An ontology for CWMS was developed using the Lyra ontology management syst em. Tools that were developed for building ontology-based models and simulations include the SimulationEditor, which is a high level modeling environment for designing a system st ructure based on a graphic interface, and the EquationEditor, which is a tool for designi ng a model in equation form and representing knowledge of each equation and symbol by using the underlying ontology. The main contribution of this thesis is th e application of ontol ogy-based techniques to modeling and simulation in agriculture and natu ral resource domains through the development of these tools and their applicati on to a particular problem. 11

PAGE 12

CHAPTER 1 INTRODUCTION Simulation in agriculture a nd natural resource management is a popular methodology for studying environmental and agricultural system problems. There has been much works on modeling crop, soil, water and nutrients in spec ific research domains (Peart and Curry, 1998), and recently interests in modeling and simulation methodology have moved to a reuse of existing models and simulation systems for building a larg e system (Leon et al., 2002). This will require better communication of model st ructure and components to the community of model builders who collaborate on an in ternational level. Traditionally, modeling and simulation are ta sks based on programming to implement the processes necessary for operating or solving a m odel to mimic real system behavior within a particular domain. General processes include developing computer logi c and flow diagrams, writing computer code, and implem enting code on a computer to produce desired outputs (Peart and Curry, 1998). While visual simulation, an approach to modeling and simulating based on building diagrams of system components, is an intuitive and simple way to do this, programming languages are still used for developing more co mplicated simulating system because there are limitations on representing ability. Classical problems of implementing a model and developing a simulation system include difficulties in managing and reusing existing models and system that are written in a particular programming language. Understanding a program written in a specific program language is difficult because it is too hard to get informati on about the detailed specification of the system. Usually this information is lost during the tran sformation to the program code (Furmento et al., 2001) and documents describing the model are p hysically separated from the implementation. Although documentations such as a paper or a manua l and descriptions in program code partially 12

PAGE 13

cover the gap in understanding, often documen tation contain inaccurate information, the document description does not adequately explain the entire system in detail, and it is difficult to maintain both the system implemen tation and supporting documents. Some simulation programs, such as Stella (Steed, 1992) and Simile (Muetzelfeldt and Massheder, 2003), solve many of these problems by providing a visual modeling environment and supporting embedded simulation and reporting t ools. Visual environments eliminate or greatly simplify the process of programming an d make models much easier to design and develop compared to hand coding of models in a traditional programming language, but sharing of modeling products is s till restricted because these tools use proprietary model representation formats. Program source code is assumed to be an easily reusable, executable, flexible and expandable way of sharing models, so that even visual programs provide the functionality for generating program source code in a specific programming language from the models. However, the problem is that different symbols and math ematical expressions are used for the same concepts at the different viewpoi nts of modeller, so th at there are enormous overlaps of concept and interaction in models. These issues motivate research on ways to explicitly represent the knowledge in a model (Lacy and Gerber, 2004; Cuske et al., 2005). An ontology is an explicit spec ification of a conceptualiza tion (Gruber, 1995), which has been applied to create a formal representation describing and categoriz ing concepts and the relationships among concepts in a particular dom ain. Classes are main elements of an ontology, which describes concepts in the domain, and proper ties represent various f eatures and attributes of the concept. There is no singl e correct ontology of a particul ar domain, and several different ontologies might exist depending on the task or role of ontology in that domain (Guarino, 1997). 13

PAGE 14

Ontologies are based on object-o riented design, and thus appear to be similar to objectoriented programming (Rumbaugh et al., 1991) and Unified Modeling Language (UML) (Booch et al., 1997), but they are diffe rent in several important aspe cts (Noy and McGuinness, 2001). In object-oriented programming, classes are regarded as types for instance, and each instance has one class as its type, whereas, ontol ogies declare that classes are re garded as sets of individuals, and each individual can belong to multiple classe s. Also, in object-oriented programming classes have behavior defined through functions and methods. Ontologies are not programming languages, and classes in ont ologies make their meaning explicit without any methods (Knublauch et al., 2006). This is an important distinction, beca use methods are coded using a programming language and thus behavior is no t explicitly represented and is largely unknown except through manual analysis or processing of the program code. Ontologies enable sharing of a common understanding of the structure of domain knowledge, reuse of domain knowledge, making domain assumptions explicit, separating domain knowledge from operational knowledge, a nd analysis of domain knowledge (Noy and McGuinness, 2001). These capabilities can be applied to the modeling and simulation domain. In recent studies, it has been determined that ontolo gies increase the potential for interoperability, integration, and reusability of simulation models (Miller et al., 2004). Also, ontologies can be a useful for the description, development, and composition of simulation models, and for mapping of input/output data. In the domain of agriculture and natural resources, an ontology-based approach to simulation, which represents a model with ontol ogy concepts, can address several problems with current methodology used to develo p simulations. Whereas ontologies have been used as a way to build generic domain knowledge, only recently have attempts been made to develop an 14

PAGE 15

ontology-based approach to modeling and simu lation. There have been many well-studied physical processes in agriculture and natural resources, and different perspectives on the problems led to development of many similar but varied models. Also, as the problems in the agricultural domain has been dive rsified and widened to the environmental and natural resource domain, requirements increase for modeling and simulation to solve multi-scale problems and to integrate existing models rather than to devel op new models for specific problems (Ewert et al., 2006). Therefore, a comprehensive management sy stem to manage these diverse models is needed. The objectives of the research pr esented in this thesis are twofold; 1) to develop ontologybased methodologies and tools to be used by modeler and researcher for building mathematical models and simulation in the agricultural and natural resource domain and 2) to apply the methodology and tools to develop a sophisticated soil water and nutrient model for evaluating the efficiency of the proposed ontology-based simulation approach. Several software components were devel oped as a part of this research. The SimulationEditor is a high-level modeling environment for specifying a system structure based on a graphical interface. The Equa tionEditor is a tool for descri bing a model in equation form and for representing knowledge of the equations and symbols used in the model. The Citrus Water Management System (CWMS) is an ap plication program applying the ontology-based simulation methodologies. CWMS provides grower s with site-specific optimal nutrient and irrigation recommendations by simulating models based on soil characteristics, nutrient uptake patterns and weather conditions. The motivations behind this research are 1) to create methodologies based on ontology techniques to explicitly represent models and related mathematical expressions of parameters, 2) to create an environment for building reusable and 15

PAGE 16

sharable model knowledge, 3) to provide the co re representational facilities of structure diagrams, symbols/equations and descriptions, 4) to use the ontology as a database for systematically storing models and model elements and 5) to assess the value of ontology-based simulation approach by applying it to modeling and simulation of CWMS. The main contribution of this research includes the development of a methodology of ontology-based simulation, which provide two main software tools; the SimulationEditor and EquationEditor. These tools were used successful ly to design and build a model and simulation for citrus water and nutrient management. 16

PAGE 17

CHAPTER 2 LITERATURE REVIEW Ontology Based Simulation Recently, ontologies have received much attention for implementing mathematical models and building simulation systems. The aim of adapting ontologies for simulation systems is similar across various projects, but the design an d implementation of an ontology is different depending on the problem domain (Benjamin et al., 2006). Miller et al. (2004) noted that for modeli ng and simulation an ont ology provides standard terminology which increases the pote ntial for application interoperability and reuse of simulation artifacts. Furthermore, semantics represented in an ontology can be used for discovery of simulation components, composition of simulation components, implementation assistance, verification, and automated testing. He proposed a web-accessible ontology for discrete-event modeling (DEMO), which defines a taxonomy of models by describing structural characterization (State-oriented, Event-oriented, Activity-oriented, and Process-oriented models) and a model mechanism explaining how to run the model. Although Miller focused on the creation of an ontology for general stochastic models such as Markov Processes or Petri Nets, Fishwick and Miller (2004) placed emphasis on capturing mostly object or instance-based knowledge. He presented a software framework, RUBE, which provides an integration method for the phenom enon of model and model object, and multiple visual modes of display to provide interfaces for developing dynamic model. 3D visualization (Park and Fishwick, 2005) is used to animate th e responses of models. An ontology is used to define a schema of simulation model types and m odels, and a sample air reconnaissance scene is represented with the Web Ontology Language, OWL. 17

PAGE 18

Some studies (Raubel and Kuhn, 2004; Cuske et al., 2005) addressed th e use of a static ontology (Jurisica et al., 2004), whic h describes static aspect of the world focusing on entities, and in a simulation focusing on the data and th e rules governing the simulation. They understood that data used by a model is a key characteri stic of semantics, which an ontology of an information system should define, rather than building an ontology which is independent from simulation form or contents. For example, ont ology-based task simulation (Raubel and Kuhn, 2004) uses an ontology for evaluating the usability a nd utility of a task or data for the decisionmaking process. JOntoRisk (Cuske et al., 2005), which is an ontology-based simulation platform in risk management domain, proposed a three le vel ontology hierarchy, cons isting of a meta risk ontology, a domain risk model, an a risk knowledge base. Esp ecially, a meta risk ontology defines the common characteristics of risk management simulation with world elements which are affected by risk, functional dependencies between world elements, random elements which are input parameters, and stochastic dependencies between random elements. Models refined from a meta risk ontology at a domain risk model have a strength on validating or reviewing the meta structure of simulation system. SEAMLESS (Ittersum et al., 2007) is a com ponent-based framework for agricultural systems which is used to assess agricultural a nd environmental policies and technologies from the field-farm level to the regional level in the European Union. For SEAMLESS, an ontology is designed to relate different concep ts from models, indicators, and source data at different level, and to structure domain knowledge and semantic meta-information about components for retrieving and linking knowledge in components. It also is used to check the linkage between components through input and output variables in the system. An ontology, the Model Interface Ontology, encapsulates knowledge of biophysical agricultural models. Static and dynamic 18

PAGE 19

models are included, and the system dynamics appr oach which describes a system with stocks and flows are applied to conceptualize models. This approach to model ontologies provides advantages which include the simplicity of mo del representation by usin g states, inputs, and outputs, but it has limits on repres enting mathematical expressions of models and manipulating models to build complex system. SEAMLESS doe s not attempt to represent models based on their mathematical equation form in the ontology. A web-based simulation using an ontol ogy in the hydrodynamic domain (Islam and Piasecki, 2004) is used to so lve the governing equations for a two dimensional hydrodynamic model. A model ontology is created to descri be a numerical model by defining a specific metadata set that describes hydrodynamic model da ta, which is used to search and retrieve metadata information. This approach gives an advantage in prescribi ng geospatial data and model data at model level. However, there is a limitation on building and describing model equation, and model should be provided in a specific form required by the system. The Modelling Support Tool, MoST (Scholte n et al., 2007), a software framework for supporting the full modeling process, used an ont ological knowledge base (KB). The KB is a collection of knowledge on modeling for various domains of water management, which is developed by domain experts. They adopt ont ological approaches to develop a knowledge structure, store the knowledge to the KB followi ng an ontological structur e, and build software applications to use the KB. Model-Based Approach to Ontology A model base is a massive collection of m odels and model components. As the number and scale of models grow, the conceptualization and role of models w ithin a problem domain becomes wider and more complex. Some models may be considered as an integration of related unit process models, while previously a singl e-process model itself was enough to make a 19

PAGE 20

simulation. As various concepts are applied to de velop an ontology to build a model, it becomes a challenge to develop an ontology which contains different categorical views and which can be used to manage models (Ewert et al., 2006). As there are diverse aspects to understanding and describing models in a specific domain, it is not easy to reuse existing model with other models or to replace a model with other models which satisfies the same requirements of input data and parameters. In large-scale problem domains, the need increases for comparing and eval uating models in order to locate an adequate model for a given environment. Lu et al. (2004) compared different models for estimating leaf area, and Eitzinger et al. ( 2004) performed a evaluation and comparison of water balance components in different models. To provide a model base, there is an effort to develop a set of crop models for a various crops and integrati ng models with farm decision support system (Reddy and Anbumozhi, 2004). A modular approach to model development (J ones et al., 2001) introduced by categorizing and organizing crop model with biological, enviro nmental and management module as a form of software component, which is an executable unit of independent production, in the agroecological domain (Donatelli et al., 2006a,b). Although they offer useful ideas on categorizing and reusing the existing components, they cannot fully address the difficulties of model management because they are developed for a specific program environment such as a FORTRAN and C++. These difficulties make it important to organize a model base that can compare similar yet different models and components. It will be useful to categorize and organize models into a well designed framework for the purpose of locating and reusing models. There have been many 20

PAGE 21

efforts to construct model bases, and recently ontologies are being applied to this purpose because of their strength in categorizing and organizing knowledge. Watershed modeling is considered as aggreg ating a complex system of unit hydrology and chemical processes, which incl udes precipitation, in filtration, evapotraspiration and erosion. Haan et al. (1982) presented a co llection of generic processes a nd practical models which have been used to study the hydrologi c cycle in watersheds. MoST (Scholten et al., 2007) developed a model ontology following the structure of compone nts in the simulation system to manage models, and it made it possible to switch one model with other models in the same process level for seeking appropriate model composition resu lting in an adaptabl e conclusion. But, the complexity of the representation is not enough to describe detailed processes, and the large scale of the system makes it difficult to manage mo dels. Although it enables model switching, it is limited to simple models. Some research to support a decision making pro cess over a farm or water management area provides a library of models that allows a user to build up a s imulation system easily with unit process models (Athanasiadis et al., 2006; Scholten et al., 2007). The libraries contain ontologies for storing the farm management model knowledge wh ich is gathered from references or experts. Usually, in those cases, models can be repeated ly used for building up a system, but there are limitations in modifying or crea ting another model from known models, even models which the system provides. A simple case is that an ont ology is not designed originally to allow any manipulation, and this problem is usually found at the multi-scale simulation model. To solve the difficulties of managing models in ontologies, the SEAMLESS built a model ontology which contains multi-scaled categories ov er an agricultural domain, and provided an interface for managing model knowledge, which is an authoring tool supporting to create and 21

PAGE 22

categorize models and to modify model knowledge (Ri zzoli et al., 2004; Athanasiadis et al., 2006). Model knowledge appearing in the interface includes a model description, creator, a components list using selected model, and mode l elements. Model elements describe model input, output, and state variables which can be used to select models Although input, output, and state variables can be dictated in the interface, it does not represent the detailed and complicated mathematical relations between them. A model ontology just contains kno wledge of concepts related with a mode as input/output or state va riables, and their mathematical relationship is coded or internally described in the system. To resolve these limita tions, it is required to focus on designing a model ontology based on their mathema tical representation and meaning explicitly. 22

PAGE 23

CHAPTER 3 ONTOLOGY-BASED APPROACHES AND TOOLS FOR SIMULATION This approach to ontology-based simulati on focuses on model authoring facilities and simulation execution tools. In the following sections, supporting technologies which enable modelers to develop ontology-based simulations are described. The SimulationEditor and the EquationEditor are the two main tools for building a simulation system. Additionally, system validation tools, a symbol referencing flow diagram and a sensitivity analysis tool, which provide facilities for model an alysis, are also described. Background Technologies Ontology An ontology is a formal explicit representation of concepts in a specific domain (Gruber, 1995). Specifically, in computer sc ience and information science ontology is considered to be a data model which represents a collection of co ncepts in a domain and relationships between those concepts. To form the representation of a data model, several elements are considered including class, individual, attribute, and relationship. An instance (also known as an obj ect or individual) is a concrete (e.g. people) or abstract (e.g. number) object in an ontology, and a class (concep t) is an abstract group of similar objects in the domain, which may contain individuals, ot her more specific classes or combination of both. An attribute characterizes a nd describes a property of a class, and has at least a name and a value. Since an important use of attributes is to describe the relations hips between objects, a relationship is an attribute w hose value is another object in the ontology. The power of ontology comes from the flexibility in describing relations hips. A common relationship is the subsumption relation such as is-superclass-of is-subclass-of, which defines cl asses that are more general or specific than other classes of instances, and th e relation part-of which represents how instances 23

PAGE 24

combine together to form composite instances (Noy and McGuinness, 2001). The procedure of developing an ontology consists of defining classes and individuals, arranging the classes in a taxonomic (subclass-superclass) hierarchy, defining attributes and describing allowed values for these attributes, and filling in the values of attributes for instances (Guarino, 1997). Ontology Management System (OMS) The Lyra ontology management system (Beck, 2007) is used to build an ontology for modeling, to develop tools for entering symbols and equations into the ontology, and to implement the tools that execute simulation and s how their results. Lyra is an object database management system for ontology data, which pr ovides a data model of the linguistic and semantic concepts in an ontology based on a formally defined ontology language. It supports management of large collections of ontology objects reasoning facilities that help in organizing and searching for concepts, visual ontology design t ools, and application development tools. It is designed as a server/client system implemented with Java. Clients communicate remotely with a database located on a remote server through Ja va Remote Method Invoc ation (RMI) technology. Model and Simulation Ontology The model and simulation ontology is devel oped with Lyra for building conceptual diagrams graphically, representing mathemati cal models with symbols and equations, and describing information related to each symbol an d equation. It consists of two parts, system design and model implementation (Figure 3-1). In order to build a simulation system diagram, three classes (project, diagram, graphic elements) are created in the model ontology. System design is a process of building a conceptual or structural diagram of concep ts and relations between them with graphic elements, and each diagram belongs to a specific project. A gra phic element may represent a set of equations 24

PAGE 25

describing a detail (or independent ) process in a complete set of models or explain a structural relationship which means a phys ical part of a system. Figure 3-1. Concepts and relations in model ontology for simulation Classes for representing a mathematical ex pression focus on symbols and equations. An equation includes knowledge about mathematic opera tors hierarchical relations, symbols, and meaning. One challenge in describing a symbol is that a term for the symbol composing an equation may be used in other equations with th e same or different meaning. A symbol (concept) is unique but having multiple names (terms) which are mapped to symbols, and enables reusability of the symbol in different models. A symbol may have one of three different sources for its value; constant, equation, and database. 25

PAGE 26

EquationEditor The EquationEditor is a tool fo r creating equations associated with a model, and properly defining symbols appearing in th ese equations. It provides a facility for creating, browsing, and inspecting all equations, symbols, and units appear ing in the model. It us es an interface that resembles other equation editors such as Micros oft Office Equation Editor (Microsoft, 2003) and MatyType (MathType, 1996), but differs significantly because all the symbols in equations and equations are represented intern ally by using ontology objects. This provides a way to represent the meaning of equations and symbols that is not possible with ot her equation editors. Equation Object Model (EOM) The Equation Object Model (EOM) is an inte rmediate collection of basic objects that represents information describing a mathemati cal expression and communi cates with the Lyra physical storage manage to retrieve and store equations and symbols (F igure 3-2). The main purpose of EOM is to represent the elements of mathematical expressi ons. Operators and other symbols of an equation are objects of the two main classes, MathTemplate and MathPrimitive. MathTemplate defines a type of operator and a collection of arguments. Character symbols and numerical symbols are subclasses of MathPrimitive. MathSymbol objects representing a symbol contain two properties; a linguistic-level propert y and a programmatic-level property. Symbol (in multiple terms), symbolID, and definition are linguistic-level properties. Programmatic-level contains three properties; source, matrixType and matrixSymbolUsage. A property source represents the origin of the numerical value of the symbol (equationType, databaseType, or constant). If the symbol is a matrix, property matrixType gives the matrix dimensions. A flag, "constant" or "variable", is a value repres enting matrixSymbolUsage property. The value "constant" means that the value never changes, wh ereas "variable" means the value can change. 26

PAGE 27

Figure 3-2. Class diagram of Ly ra equation object model retrieving instances of model ontology Components of the EquationEditor The EquationEditor has three sub-editors, Sym bol Editor, Mathematical Statement Editor, and Unit Editor, to create and maintain symbols, equations and symbol units. Symbol Editor (Figure 3-3) is an editor for individual symbols appearing in equations and includes a symbolic expression of a symbol, a quan tity of measurement, a nd a description of the linguistic and programmatic properties of the symbol. A symbol is implemented as a class in the ontology, which has a unique meaning within a specific domain. Often, the same symbolic character (term for the symbol) is used over differe nt domains, but is used in different ways and has different meaning. Since a symbol has a di stinguishing identifier representing a specific 27

PAGE 28

concept in the ontology, a use of the same te rm for different symbols is permitted, and the domain ontologies can be used to resolve their ambiguous meaning. Figure 3-3. Features of the symbol editor in the EquationEditor: symbol ID, symbol, unit and description of Cell Crop Evapotranspiration The value of a symbol is determined by one of three methods: from an equation, from a database, or from a constant which is directly assigned to the symbol. In the case where the symbol value is determined by an equation, there must be an equation in the database in which this symbol appears alone on the left side. To obtain the value from the database, some constraints may be required in order to locate and query a database to obtain the value (e.g. a current time and a soil layer number for querying a soil temperature at a specific date), and these constraints can be specified as a part of the symbols properties (Figure 3-4). Symbols can also be arrays, when a symbol can be used in different discrete intervals in space and time. For example, soil water content can be expressed in different soil layers which occur in different soil profiles, characterized by the depth from the soil surface, the soil profile number and time. 28

PAGE 29

Figure 3-4. Database constr aints and array dimension de scription of a symbol Figure 3-5. Equation ID and equation of Cell Evapotranspiration in the mathematical statement editor 29

PAGE 30

The mathematical statement ed itor is designed to graphica lly create an equation from existing symbols and mathematical operator templa tes (Figure 3-5). An eq uation is an expression that has a hierarchical tree data structure composed of symbols and operators. The equal operator is the root node of the tree stru cture, containing a single symbol on the left branch of the tree. The value of the left side symbol is defined by th e calculation of the right side terms. Thus the equation is assumed to be a function which has symbols as arguments. The editor provides an operator template which can describe specific argument sets. There ar e eight operator groups used to compose an equation (Table 3-1). Table 3-1. List of operators in EquationEditor Operator group Operators Exponential Subscript, double subscript, superscript, exponent, sub and super script, function, square root, root, log Fence Parenthesis, bracket, br ace, absolute, ceiling, floor Trigonometry Sine, cosine, tangent, arcsine, arccosine, arctangent Calculus Limit, differential, inde finite, definite, summation, product, maximum, minimum Logic And, or, not Arithmetic Add, subtract, multiply, divide, negation Relation Less than, greater than, less a nd equal, greater and equal, equal, equivalent, not equal, not equivale nt, less than and less than equal to, less than equal to and less than Case n-case, matrix The Unit editor is an interface to create and maintain the unit for a symbol and its compositions for representing the quantity of measurement of symbols (Figure 3-6). Unit includes not only the generic co llection of global standard un it of metric system (e.g. international system of unit (SI) and the English unit system), but also domain specific units such as cm3 of soil in soil engineering. It is very im portant to carefully track the units associated with symbols, since different models may use th e same symbol but havi ng different units. A unit is not represented by a simple string, but by a composition of symbols (like an equation). The 30

PAGE 31

unit can be expressed using a composition of limited operators (multiply, divide, and power operator) and other units. T hus, basic units such as length and we ight can be reused for creating a composite unit, and this makes it possible to au tomatically calculate conversion of units from one form to another (e.g. the E nglish unit to the metric unit). Figure 3-6. Unit ID, unit and definition of centimeter of water in unit editor SimulationEditor The SimulationEditor is used to describe th e structure of dynamic systems using graphic elements such as source, sink, storage, and flow It adopts concepts from the compartmental modeling technique (Peart and Curry, 1998) and Forrester notation (For rester, 1971) which is widely used in agriculture and natural resource models. However, like the EquationEditor, these concepts are represented intern ally using the ontology and stor ed in the Lyra database. The SimulationEditor is used for speci fying the overall model structure in the form of elements, and 31

PAGE 32

incorporates the EquationEditor de scribed in the previous section in order to build equations associated with each element. The SimulationEd itor provides a graphic user interfaces to create and maintain a simulation system which incl udes a structure design interface, a simulation control interface, and a simulation result reporting interface. The SimulationEditor also contains facilities for automatically generating and r unning simulations and generating reports. Figure 3-7. Structure editor showing a comp artmental diagram of a soil-water model The structure editor is the main interf ace of the SimulationEditor and provides functionalities which enables modelers to create and maintain a simulation project by designing the structure of a system, and to interact with the EquationEditor and th e simulation controller. Structural design of a system is a proce dure by which a modeler creates physical or environmental elements and relationships in th e system by using graphic elements. For example (Figure 3-7), a 3-dimensional soil profile syst em may be designed as a composition of soil cell (production unit), soil profile (horiz ontal division), and soil laye r (vertical division) concept. These three elements may be defined as an in stance of storage element, and relationships 32

PAGE 33

between these elements are represented by P art of (e.g. block, soil cell and soil profile composed of soil cell, soil profile and soil layer respectively ). Irrigation may be realized with the flow element representing the flow of water into the cell. The simulation controller is a collection of simulating engines used to generate a simulation program from the mathematical model, to run the simulation, and to generate reports. For simulating a model, a simulation engine au tomatically converts onto logy objects to program source code. It then compiles and runs the generated program to execute the simulation. Currently, Java is the target la nguage, although in theory a simu lation program may be generated using other programming language. It is not necessary for the modele r to examine, work directly with or otherwise be concerned about the gene rated program source code. This process is completely internal to the operation of the soft ware, and transparent to the modeler. However, the compiled source code can be used as a component that can be accessed by other software environments after the model is developed. The data object conversion and generation of program source code follows these steps: A class representing a module in the SimulationEditor forms a single class in Java. The class contains member variables and methods for all the symbols and equations in the module. Each symbol in the module is declared as a member variable named after the name of the symbol. If the symbol is a matrix, the member variable is declared as an array with the same dimensions as the symbol. A method is created that contai ns code for obtaining the value of the symbol. The name of this method is based on the name of the symbol. o If a symbol is a constant, the return value of the method is a constant for the symbols value. 33

PAGE 34

o If a symbol obtains its value from a database, the method returns a value obtained by querying the database for th e value of the symbol, subject to constraints specified in the symbols properties. o If a symbol obtains its value from an equation, the method contains code for solving the equation to obtain the value. Since the equation contains other symbols, these values of these symbols (on the right hand side of the equation), are obtained by recursivel y calling methods for determining their values. Generated source code set is de signed to be independent from the SimulationEditor, so that it can be used as part of a component library an d inserted into other ap plication independently. There are two levels to the system: the core leve l and the application leve l (Figure 3-8). The core level is the ontology-based simulation envir onment including the SimulationEditor and, the EquationEditor integrated within the Lyra OMS which also provides the database management facilities for storing the ontology objects created by the SimulationEditor and the EquationEditor. At the applica tion level, the generated code library is used by the other applications which can be implemented independ ently from the core level. For example, the resulting simulation application can be integrated into a desktop application used by growers, it can be part of a larger decision support system su ch as DISC (Beck et al., 2004) which is a citrus planning and scheduling program or a Web-base d simulation environment (users can run the simulation through a web page, or the simulation can be part of a web service that is part of a distributed simulation environment). 34

PAGE 35

Figure 3-8. Relationships betw een ontology-based simulation system, generated code and application Running a simulation involves compiling and executing the automatically generated source code. The simulation is controlled by recursively evaluating the value of a target symbol. Within the SimulationEditor, there is an interface to communicate with the model code library, which contains a method for calling the target symbol s method which results in execution of the simulation. The generated source co de contains variables for stor ing all values of variables, which are retrieved by a report generator to di splay model results when the simulation has finished executing. The report generator displays simulation result s by showing the values of specific symbols in the form of a table or a graph as a function of time and proper dimension described in a symbol. A list of symbol ID which is stored in the ontology is provided to create reports, and a report is designed by selecting and adding to the target variables list and the dependent variable after simulation. A designed report form can be categorized and maintained in the ontology. 35

PAGE 36

Additional Tools and Facilities A system may be composed of many small models, and these models reference other models or equations. There is a need to verify interactions be tween such complex structural relationships, and to assess the behavior of models statistically. Tools are also available for exporting the model into XML in two different formats, MathML (Ausbrooks et al., 2003) and OpenMath (Buswell et al., 2004). Figure 3-9. Symbol refere nce diagram focusing on Layer Daily Evapotranspiration To verify the complex flow of referencing equations in a model, it is useful to visualize those flows and call sequences as a diagram. The calling sequence is generated by interpreting attributes recursively in a process that is sim ilar to the process of ge nerating the program code. Figure 3-9 shows an example of a calling seque nce of equations whic h are parts of CWMS 36

PAGE 37

models presented in Chapter 4. To calculate a value of a symbol such as the layer daily evapotranspiration (ETLcs), some values are requir ed including the total root length (TR), the cell crop evapotranspiration (ETCc) the soil coefficient (ks), the today number of time step (TodayTS) and the layer root length (rl). Arrows starting from the target symbol point to other symbols which are required to calculate it. Numb ers over each node indicate of nodes which are related to that node, but not shown. The calling sequence interface provides a convenient way for displaying every connected symbol as a visual ne twork diagram, and it can be used to browse symbols and navigate the model. A sensitivity analysis process is implemented using computer experiments. The aim of sensitivity analysis is to determine how sensitive the output of a system is with respect to the elements of the model which are subject to uncerta inty or variability (Wallach et al., 2007). This is useful as a guide when the model is unde r development as well as to understand model behavior, to seek the main affecting factors in the system, and to figure out the significant interaction between input factors selected from th e system variables. The procedure consists of two step; factor screening using Morris randomized OAT (One-factor-At-a-Time) design (Morris, 1991; Alam and McNaught, 2004) and global sensitivity analysis w ith screened factors from previous step. The Morris randomized OAT design as a factor screening method is used to determine which factors are real ly significant when there are potentially a large number of factors involved. A computer experiment is a set of simulation runs designed to explore the model responses when the input varies within given ranges. The number of executions required to do this is dependent on a number of selected factors and levels. After choosing input factors in the system, discrete levels are automatically decided by the maximum value, minimum value and 37

PAGE 38

number of levels. An experiment engine feeds f actor values at a specifi c level, and control the iterative simulating process. The XML generator is a tool to generate a markup language form for a model built in the ontology. XML enables the model to be shared outside of the Lyra OMS environment. Two forms of markup language, MathML and Open Math, can be generated. MathML is an application of XML for describing mathematical notation and capturing both its structure and content. It aims at integrating mathemati cal formulas into Web documents. It is a recommendation of the W3C (World Wide We b Consortium) math working group. Whereas MathML has strength on presentation of fo rmulae, OpenMath is a document markup language for mathematical formulae, which provides a mechanism for describing the semantics of mathematical symbols. To generate these XML formats from equations in the ontology, each operator template class which is declared in the EquationEditor has a method transforming operator and arguments to a string containing a XML tag expression. An operator template can have other operator templates as arguments. An equation may be considered as a tree data structure composed of operator and symbol. The XML generator trav erses this tree from the root operator template (which is always the equal operator) to each leaf operator template, similar to the way in which the code for solving the equation is generated. An example of generating markup language is shown in Figure 3-10 which a simple equation A=B+C is generated in MathML and OpenMath format (generated Java c ode for solving this equation is shown as well). 38

PAGE 39

Figure 3-10. Result of generating ma rkup language for the equation A=B+C In Chapter 3, methodologies were covered to represent mathematical models in the ontology using the Lyra ontology management system. To utilize the constructed model ontology for simulating models, two main tools, the EquationEditor and the SimulationEditor, are developed. The EquationEditor provides interfaces for describi ng symbols and equations in a model and for retrieving ontology objects, and th e SimulationEditor helps to conceptualize the circumstance in which models are applied. Simulation handling proc ess is facilitated by functions including automatic gene ration of program codes and re ports, sensitive analysis, and calculation sequence diagram. Model repres entation adopting ontol ogy-based methodologies simplifies to create a deliverable mo del expression such as a XML form. 39

PAGE 40

CHAPTER 4 APPLICATION TO CITRUS WATER AND NUTRIENT MANAGEMENT Ontology-based simulation methodol ogies covered in Chapter 3 were applied to building a model describing water and nutri ent balance processes for the Citrus Water and Nutrient Management System (CWMS) (Morgan et al., 2 006a). To aid growers in water management decision making, a computer-based decision suppor t system was developed to facilitate more efficient use of water and nutrients by basing recommended application rates on site specific characteristics and local weather data. The purpose of this work wa s to test the feasibility of utilizing ontology-based simulation to build a mode rately complex model, resulting not only in a simulation that can execute rapidly, but that also can be incorporated into a user interface for delivery to and use by growers or in other applications. The CWMS Model Description of Model The CWMS model has been desi gned for the sandy soils of central and southern Florida which have low water and nutrient retention capacities. At citrus production sites, nutrients may be leached from the sandy soils by excessive ir rigation events. The CWMS model was developed to anticipate the potential contribution to th e groundwater contamination and to provide appropriate irrigation scheduli ng strategies. The CWMS model uses two main water inputs, rainfall and irrigation events. Rainfall amount is assumed to be affected by the canopy volume covering the soil surface. An irrigation even t contains information including nutrient concentration (in the case of fertigation or injecti on of liquid fertilizers in to the irrigation water), amount and event date, and it cons ists of several distinct irri gation processes. The soil water budget models in the CWMS model are based on cr op water use, soil-water storage capacity, and 40

PAGE 41

vertical soil water movement. Horizontal water movement is excluded due to lack of lateral movement in the sandy soil. The model is based on a restricted area, a soil cell in a block repr esenting a single citrus tree and the drainage field surrounding it, which is the basic unit of the geometry (Figure 4-1). A commercial block of citrus consists of many soil cells since it has many trees, but in this model to simplify the simulation process the model is based on a single soil cell, and the single tree represented by that soil cell is characteristic of all the other trees in the block. A soil cell is defined as a cubical soil area co ntaining one citrus tree, having a depth of 200cm from the top of the soil. A soil cell is further divided into soil pr ofiles within a cell and soil layers within a soil profile. As shown in Figure 3-7 in Chapter 3, seve n concepts are defined for the model structure; block, soil cell, soil profile, soil laye r, root, irrigation, and weather. A soil cell consists of a one-t ree planting row area with the tree in the center. The width and length of the cell are in-row and between-row distances to adjacent trees. It includes fourtypes of zones (i.e. a non-irrigated & dry-fertilized area, an irri gated & dry-fertilized area, a nonirrigated & non-dry-fertilized area, and an irrigated and non-dry-fertilized area as shown in Figure 4-2) according to the irrigation status a nd the dry-fertilized status, and each zone may have from 1 to 5 soil profile(s) which consist of n soil layers. The total soil layer number (n) is determined based on the particular soil type or a depth of each soil layer. 41

PAGE 42

Figure 4-1. Conceptualization of soil geometry of CWMS model Figure 4-2. Examples of soil profile areas 42

PAGE 43

Water balance for each profile is determined us ing rainfall and irrigation events as water inputs and evapotranspiration (ET) and leaching water as wate r losses from the balance. Basically, the water budget calc ulation is based on the tipping bucket model, enhanced by considering the effect of the delayed soil water drainage caused by the soil hydraulic conductivity during one daily time ste p. It is assumed that water input s are applied at the first soil layer. Water infiltration depth is calculated as a function of the infiltration characteristics of the soil. A pure tipping bucket model assumes that wate r moves the entire layer depth in one time step. Tipping bucket does not alwa ys reflect soil water content on a daily time step due to soil hydraulic characteristics. To account for the hydrau lic characteristics of the soil, CWMS assumes that all irrigation, water with N application and rainfall occurs at noon and has a maximum of 12 hours to move through the soil on the first day. Th e model calculates the wetting front speed and the time for which the wetting front travels a la yer thickness, thus to obtain the layer index of wetting front at the end of the day and let unfinished drained water continues to drain at next day. As stated above, water loss from the water balance is water drainage below the 200 cm maximum depth and crop ET. Daily crop ET is calc ulated in CWMS using reference ET as user input or from weather data. A crop coefficient based on Morgan et al. (2006b) is applied to the reference ET based on seasonal variability. The crop ET deducted from each soil profile layer is proportional to the root length density of the profile layer. A layer root le ngth density distribution is a function of tree size (Morgan et al., 2006a). An irrigation is sche duled when the CWMS determines that water content in the irrigated zone is below th e allowed depletion. The irrigation 43

PAGE 44

duration is determined by the water amount needed to bring water content to field capacity and is determined by the emitter flow rate, ir rigation efficiency, and irrigation depth. For nutrient management, especially applicati on of nitrogen by irrigati on (fertigation), the CWMS model provides processes for calculating nitrogen balance followed by transformation of nitrate and ammonium. Transfor mations are composed of comp licated sub-processes and are affected by input and flow (drainage from a bove layer) amount of nutri ent and water between layers. The model assumed that th ere are four nutrient flow proce sses by four drain events: drain by rain, pre-irrigation, duringirrigation, and post-irrigation. After four-drain steps, transformation processes is applied to ammonium and nitrate following in order of volatilization, uptake of ammonium, uptake of nitrate, and nitrification. Model Base of the CWMS Model The CWMS models is implemented by us ing a taxonomy representing physical relationship of natural resources (soil profile, crop, and environm ent), which consists of 4 soil related concepts (block, soil cell, soil profile and soil layer), root density, weather, and irrigation. The system structure taxonomy is built graphically with the SimulationEditor (Figure 3-7). For the CWMS model approximately 700 symbols and 500 equations were created. Symbols and equations developed for the CWMS model are interrelated, and their relationship can be visualized as a graph di agram (Figure 4-3), which displays connections of symbols and equations used to model the wa ter and nitrogen balance proce ss. Symbols and equations are represented as rectangular boxes, and the number above a box shows the count of connected boxes but not displayed. An arrow means that the target symbol is required to calculate the source equation. 44

PAGE 45

Figure 4-3. Relationship among equation sy mbols for water and nitrogen balance In the diagram, nitrogen balance process group is connected with the water balance process group by referencing several equati ons for water balance. They can be switched with similar process groups independently. Pa rticular processes, uptake, nitrification and volatilization, are clustered clearly from other equations in the equation group of nitrogen balance, and for water balance a similar pattern was found for pro cesses of evapotranspi ration, infiltration, and irrigation. This suggests possibili ties for organizing and categoriz ing models and subprocesses. 45

PAGE 46

The graphic in Figure 4-3 is generated automatica lly from the ontology objects, and an animated interface allows the model to navigate through the space of symbols. A model base is a database of many models, model elements equations, and symbols. It can be utilized for reusing models by applying a taxonomic organization to an ontology. In Figure 4-4, a model base is shown, which consists of equations used in the CWMS model. The taxonomy contains 6 classes including weather, wate r, nutrient, soil, crop and site. The water class has 5 subclasses (infiltration, evapotranspi ration, precipitation, irrigation and runoff), and each subclass contains related equations and sy mbols. For example, the CWMS model includes two different infiltration mode ls (Tipping Bucket Water Content Equation and Enhanced Wetting Front Water Content Equation). Figure 4-4. Taxonomy diagram of the CWMS model 46

PAGE 47

Examples of Model Representing Process Representing a model formed as theoretical or logical expressions (equations) in the ontology-based simulation system is a process of redefining and adjust them into a real world simulation for describing a phenomenon. Theoretical models provide the conceptual knowledge, and there are usually many reformulated models depending on assumptions and the given situation. Morgan et al. (2006a) described how they de rived the infiltration model used in CWMS from theoretical models. In summary, their infiltration model is based on Green-Ampt infiltration (Green and Ampt, 1911) and unsaturated flow based on Richards equation (Richards, 1931) derived from Darcys Law for i rrigation and rainfall moving into soil from the surface and moving between soil layers. The Green-Ampt model in the case of no ponding at surface can be expressed as: Where, fp : infiltration capacity of soil (the rate that water will infiltrate as limited by soil factors) F : cumulative infiltration Ks : the hydraulic conductivity of the transmission zone M : the difference between initial and final volumetric water contents Sf : the effective suction at the wetting front Furthermore, the model adopted Mein and Larsons equation (Mein and Larson, 1973) applying the Green-Ampt model for rainfall condi tions by determining cumulative infiltration at the time of surface ponding. Some assumptions addr essing the situation in which the rainfall 47

PAGE 48

intensity is less than the infiltration capacity of the soil are focused. The general Mein and Larson equation can be described as: Where, Sav : the average sucti on at wetting front Fp : the cumulative infiltrati on at the time of surface ponding R : the rainfall intensity By considering the relation between rainfall intensity, infiltration capacity, and saturated hydraulic conductivity, the Mein and Larson equation can be written like as: Morgan et al. assumed that no ponding occurred in the sandy soil (e.g. for Candler soil, Ks is 25 cm/hour) because rainfall is less than 5 cm/hour (except during hurricanes) and irrigation rates are typically 0.315 cm/hour or less. It was also assumed that runoff does not exist since the rainfall and irrigation intensity is usually smaller than Ks. Thus, the infi ltration rate equals the rainfall rate and the amount of infiltration water equals amount of rainfall rate multiplied by time t. If rainfall rates are larger than the irrigation rate, the infiltration rate equals the rainfall intensity. Otherwise, it equals irrigation in tensity. Thus, the Mein and Larson equation is simplified for such rainfall and irrigation conditions. The condition term for comparing the rain 48

PAGE 49

intensity with the saturated hydraulic conductiv ity was simplified to comparing the rainfall amount (represented by PWID4) with the irrigati on amount at soil cell (represented by CIAM) for creating the average infiltration rate of soil pr ofile. To apply the world system to the model, additional condition term s describi ng whether current soil profile ex ists in the irrigated-area or non-irrigated-area were added to th e model. With the above result, an equation for th e infiltration rate of soil profile was formed in Figure 4-5. CIAM is the cell irrigation amount, and i is the profile type, and numbers (1, 2, 3 and 4) repr esent the profile type, a non-irrigated & dryfertilized area, an irrigated & dr y-fertilized area, an irrigated and non-dry-fertilized area, and an non-irrigated and non-dry-fertiliz ed area. SAKtop is the satu rated hydraulic conductivity at surface, and PWID4 is the soil profile water input amount from rainfall effect. IAM is the equation calculating irrigation amount from irrigation duration, and IDur is the irrigation duration time. Examples of CWMS Model Implementation Concepts developed above were entered into the ontology system. The geometry relating soil cell and profile, soil water redistribution, a nd root density are give n below as examples. Soil geometric dimension Basically, a profile is determined by the distan ce from the trunk of a tree to three root zone radii (75, 125, and 175cm). Other profile boundaries are the irri gation diameter and the dryfertilized area. Depending on the irrigation t ype (360 degree or less than 360 degree), soil profiles can be divided into irri gated-areas and non-irrigated areas. Irrigation and dry-fertilizing events are assumed to be conducted in a soil cel l area except two equipmen t drive paths between tree rows. Therefore, the two drive paths are al ways considered as a non irrigated & non dryfertilized area (NINDF). An i rrigated & non-dry fertilized area (I NDF) is an irrigated area in the drive paths. 49

PAGE 50

Where, Figure 4-5. Morgans infiltration rate model in the CWMS model Figure 4-6. Equations of profile number 50

PAGE 51

Equations for calculating each profile number and area are defined at the cell, and Figure 4-6 shows equations of three different profile number (profile number of a non irrigated & non dry-fertilized area is always 1 and it is defi ned as a constant). NPID F, NPNIDF, and NPINDF are respectively symbols of a prof ile number of an irrigated & dry-fertilized area, a profile number of a non-irrigated & dry-fe rtilized area, and a profile numb er of an irrigated and non-dryfertilized area. RZR is a symbol of the root zo ne radius matrix, and WD is a symbol of the wetting diameter, and SpP is a symbol of the spray pattern. A soil layer is one vertical element of a soil pr ofile, and the number of soil layers in a soil profile is determined by layer thickness and total depth of the soil profil e, whose maximum depth is 200cm. The thickness of soil layers can be gr ouped, and it is represented as a matrix as in Figure 4-7. Each row is a layer gr oup, the first column is a thickne ss of layers in a group, and the second column is the number of layers in a gr oup, and third column is a cumulative layer number to that group. Symbols belonging in concepts, block, soil cell, soil profile, and soil layer, are required dimensions. The time dimension is a common dimension required by most symbols as one dimension of the matrix. Symbols in block a nd soil cell need only the time dimension, whereas those in soil profile need three dimensions, one for time, one for soil profile type, and one for soil profile number. Symbols in the soil layer n eed four dimensions, and they include three dimensions from soil profile and one for soil laye r number. For example, a symbol, historical soil layer temperature, defined in soil layer has f our dimensions, and it is described through the EquationEditor (Figure 4-8). 51

PAGE 52

Figure 4-7. Layer thickness matrix Figure 4-8. Example of dimension desc ription of a symbol in soil layer Time step The main time step for the model is daily. Ra in event, irrigation event, and fertilizer applications are assumed to occur at 12:00 pm, but adjustments to the time period of 0-12:00 pm and 12:00 pm 24:00 pm are cont rolled by using some time flag symbols. The minimum period of simulation which can be executable is 14 days. Th e first 13 days (for 14 days calculation) or first 5 years (for 20 years calculati on) are used to initialize and st abilize values of symbols whose initial condition are not k nown at the simulation starting time. Fo r example, the initial values of soil water content is usually unknown, but assumed to be a default value and computed over the 13 days (or 5 years) to initialize the values. 52

PAGE 53

Figure 4-9. Total number of time steps symbol, t There are two concepts related with time. One is Total Number of Time Steps for assigning system time dimension size, a value that is stored in and obtained from the database (Figure 4-9). The other concept related with time is Index variable of Time Step, t, which is used as the general variable for the current time. Root density As shown in Figure 4-10, root density is repr esented as a matrix from the model for each soil depth and root sections base d on root density distribution as a function of tree size (Morgan et al., 2006a). A column is a root section and a row is a soil layers gr oup to a specific depth. Following Morgans model, the root horizontal ar ea is divided into four sections (0-75 cm, 75125 cm, 125-175cm and 175cm-boundary), and 10 so il layer groups are used. The first 6 rows have thicknesses 15 cm and last 4 rows have 30 cm thicknesses, which are calculated by proportional equations based on the root density value of 6th soil depth group. RD is a symbol of 53

PAGE 54

the root density, and cv is a symbol of the canopy volume, and RSL is a matrix symbol of the root density regression parameters fo r layers below 6th soil depth group. The root density equation is designed for sp ecific soil layer depths and root ranges. The CWMS model utilizes it for an equation of LRD which is the layer root density (Figure 4-11). LD and LT are symbols of the layer depth a nd thickness of a layer at specific depths. Figure 4-10. Root density matrix 54

PAGE 55

Figure 4-11. Layer ro ot density equation Water dynamics The water balance model consists of rainfall, irrigation, evapotranspira tion, and infiltration. As stated previously, surface runoff and subsurface lateral flow are not considered due to the high saturated hydraulic conductivity of sandy soils. Basically, the water budget calculation is based on the tipping bucket model, enhanced by c onsidering the effect of the delayed soil water drainage which is caused by th e soil hydraulic conductivity during one daily time step. Rainfall and irrigation are water inputs into the system, and it is assumed that they are applied at the first layer. The symbol Profile Water Input (PWI) in the soil profile module represents input water amount fr om water resources into the so il profile. According to the CWMS model, it considers the complicated nutrient management processes by calculating different sources of water and nutrient separate ly. In the equation for PWI, sources of water inputs are distinguished as 4 different types depending on the considering event: irrigation (water), N application, post N application, and ra infall. Since the system time step is daily and 55

PAGE 56

there is no sub-time step, symbols are created for every stage and distinguished by subscript number. These are four-drain process named. Figure 4-12a show s related PWI equations (PWI1, PWI2, PWI3, and PWI4), where it is assumed that irrigation and N applicati on are applied to just irrigated-area (i=2). The symbol WI (Figure 4-12b) is the layer water input, and the amount is determined by the difference of amount of input water into a layer from environment (PWI) or from the upper layer (DW) and the amount of layer evapotrans piration (nowET). For the four-drain process, calculation of layer water input is conducted for every stage separate ly. The symbol WIflag is for indicating the status of th e remained water drain by the hydraulic conductivity. A layer water content is represented by the symbol WC whose amount is calculated by the difference of increasing amount (WCpos) by dr ained water and decreasing amount (WCneg) by evapotranspiration during the prev ious time step. At the start of simulation all layer water content has same amount calculated with constant depletion level. For the four-drain process, a layer water content at time t would be a WCD4 of previous time step which is the layer water content applying irrigation, N a pplication, post N application, and rainfall (Figure 4-12c). The CWMS model enhanced the pure tipping bu cket model by computing the wetting front moving speed during the simulati on, whereas the pure tipping bucke t model assumed that water moves the entire layer depth in one time step. Th e model assumed that al l irrigation, water with N application, water with post N application and rainfall occurs at noon and has a maximum of 12 hours to move through the soil on the first day after irrigation, water with N application, water with post N application or rainfa ll. And, the model calculates th e wetting front speed (Figure 413) and the time for which the wetting front trav els a layer thickness, t hus to obtain the layer index of wetting front at the end of the day and le t unfinished drained water continues to drain at 56

PAGE 57

next day. WFSP is the layer wetting front speed, and it can be calculated by the symbol of the infiltration WFSP (Figure 4-13b) or the symbol of the hydraulic conductivity WFSP (Figure 413c). Figure 4-13d is the equa tion for hydraulic conductivity. A B C Figure 4-12. Water balance equations: A) prof ile water input amount eq uations, B) layer water input amount equation, and C) layer water content equation 57

PAGE 58

A B C D Figure 4-13. Enhanced hydraulic conductivity re lated equations in CW MS model: A) Wetting front speed equation, B) infiltration we tting front speed equation, C) hydraulic conductivity wetting front speed equation, and D) hydraulic conductivity equation Nitrogen balance For nutrient management, especially applicati on of nitrogen by irrigati on (fertigation), the CWMS model provides processes for calculating nitrogen balance followed by transformation of nitrate and ammonium. Transfor mations are composed of comp licated sub-processes and are affected by input and flow (drain from above la yer) amount of nutrient and water between layers. The model assumed that there are four nutrient flow processes by four drain events: drain by rain, pre-irrigation, during-irrig ation, and post-irrigation. After f our-drain steps, transformation processes is applied to ammoni um and nitrate following in order of volatilization, uptake of ammonium, uptake of nitrat e, and nitrification. The amount of nitrogen input into a soil profile is an equation formed with the dissolved ammonium (NH4) and nitrate (NO3) nitrogen amount of dry fertili zer by rain or irrigation and 58

PAGE 59

fertilizer amount during the fertigation. Nitroge n drain amounts (ammonium and nitrate) of a layer from the above layer are represented by sy mbols, Layer Drained Nitrate and Layer Drained Ammonium, and they are calculated by equations of soluableRate. For uptake amounts of nitrogen, passive uptake of ammonium and active/ passive uptake of nitrate are formed as an equation. Also, the model assumed that the up take process of amm onium occurred after volatilization and nitrification fo llows a nitrate uptake process, and they are utilized by using separate symbols for each stage. The amount of nitrification is represented by symbol NIT, and its equation includes the rate of maximum nitrification as a function of ammonium content (NH4NC3), maximum nitrification amount at a time (NIT Vmax), half-saturation constant of nitrification (NITkm), and soil moisture factor (NITwf). Nitrification is allowed till 80% of cu rrent ammonium content (Figure 4-14). Figure 4-14. Layer nitrification equation Figure 4-15. Accumulated profile volatili zation equation 59

PAGE 60

Volatilization (cumulativeVOLA) is created at the soil profile module, since volatilization occurs till 5cm soil depth from the surface. It represents a cumulative vo latilization loses over a day, which is determined by the relations hips between ammonium content (VolPNH4NIn), days counting from the volatilization starting tim e (NApplyElapse), per centage of maximum cumulative volatilization (qm), and site-specific temperature/wind parameters (Figure 4-15). Application Implementation The CWMS model is used to implement a CW MS application program for use by grower that utilizes crop, soil and w eather data. A CWMS application consists of the automatically generated simulation code and a graphic user interface. The generated simulation code is plugged into the graphic user interface w ithout further coding required. A graphic user interface allows growers to in teract with the system through three phases: the setup, the irrigation scheduli ng, and reporting. The setup phase is for configuring the cell and block information for a particular grove (Fig ure 4-16a) and for descri bing simulation site, start/end data, and resource locati on (Figure 4-16b). The simulation period is separated into an initializing period and a simulation period. For a long term simulation, an initializing period can be longer than 14 days which is the default initi alizing period for farm irrigation scheduling. The irrigation scheduling phase, shown in Figure 4-17 provides irrigation scheduling information to growers. By default, it is based on a 14-day simulation followed by a 3-day prediction period (the simulation period can be extended) to provi de immediate term recommendations on irrigation ra tes. In order to plan a st rategy on irrigation scheduling, simulation system tested for full seasons ( over 250 days per year and over 30 years). 60

PAGE 61

A B Figure 4-16. Setup Phase: A) grove /block and B) simulation period Figure 4-17. Irrigati on scheduling result 61

PAGE 62

A B Figure 4-18. Simulation result s: A) table a nd B) graph 62

PAGE 63

Detailed simulation results are provided in th e form of daily, monthly, and yearly reports (Figure 4-18). The daily report contains each la yers root length, water content, nitrate and ammonium content, and soil coefficient. Data can be browsed by selecting a specific date and profile type. The monthly report shows data for a particular month includ ing irrigation interval days and duration, evapotranspiration, crop coeffi cient, soil coefficient, water and nutrient leaching amount at 2 m depth and irrigation de pth, rain, and irrigation. The yearly report provides the monthly total value of irrigation, rain, water and nut rient leaching amount at 2 m depth and irrigation depth, and fertilizing amount. Water, nitrate, and ammonium content contained in the daily report can be displayed as a chart. Model Extension The CWMS Java code was generated auto matically by the SimulationEditor and the EquationEditor, and can be used as a software co mponent that can be used independently of the model building environment. The compiled Java code can be easily connected to other user interfaces or simulation system. For example, the Watershed Assessment Model (WAM) which is used as a part of a larger FDACS BMP simulation, adopted the CWMS model as a subcomponent of land use in citrus production. The CWMS model provides information about leached water and nutrient amounts to the host simulation system. A key issue was how easily the CWMS model could be modified for integration with WAM. A water and nitrogen balance model need ed by the WAM was required to use several new soil types, and modifications to the model were made to support these new parameters. Using the EquationEditor, new symbols for soil t ypes and parameters were created and existing equations were modified by replacing old symbol and by adding new formula, and some new equations calculating required values by the WA M were added into the model. This was all 63

PAGE 64

accomplished with less than 10 hours of work in cluding creation of required file input/output protocol by WAM (that took about 80% of total work hours). Model Performance Simulation of the CWMS model containing approximately 700 symbols and 500 equations uses different numbers of equations depending on the combination of pro cesses (e.g. use process of tipping bucket or effect of hydraulic conductivity). To test the performance of the CWMS model for the maximum number of processes, the combination includ ing hydraulic conductivity, four-drain process and moving wetting front is chosen, which consists of 330 equations. Applying dimension size to these equations, tota l number of calculation at double precision level is 330,000 for 14 days period. It takes approximately 5 seconds by a computer with Intel Pentium 1.7GHz CPU speed and 1G RAM. Calculation time for 1, 10 and 30 years are 24 seconds, 5 minutes and 15 minutes, respectively. For an ex tension for WAM which contains process of tipping bucket and moving wet front, it takes 5 minutes for 30 years simulation. In order to validate the accuracy of the mode l, Morgan (Morgan et al., 2006a) compared the observed water contents at soil depth 10, 20, 30 and 50cm from foil surface with 2 years simulation result for two different sites. Accordi ng to the validation result, R2 values were varied in the range between 0.46 and 0.75, and for soil depth 10 and 20 cm it gave 0.7 R2 average value which is higher than another two points. The validity of the model depends on the accuracy of the equations and parameters, and is impacted by the quality of the model implementation platform. Model Sensitivity Analysis Sensitivity analysis is useful method to guide model development as well as to understand model behavior when the model is under constr uction. This method is applied to the CWMS model for identifying most significant factor s to Cell Water Amount (CWA) and determining their interaction, which has been implemented at the SimulationEditor as an additional feature. 64

PAGE 65

The procedure consists of two steps: a factor screening with Morris ra ndomized OAT design and a global sensitivity analysis with screened factors. At the factor screening step, a variable (e.g. CWA, a sum of water am ount in the soil cell) is selected to analyze the res ponse of the system to the CWMS models water balance processes. Variables related with the tree characteristics, water inputs (rain and irrigation) and water movement are selected as an input factor (Table 4-1). Table 4-1. Input factors related with water input and hydraulic conductivity no Symbol ID Symbol min max unit 1 Canopy Volume cv 3.2 12.6 m2 2 Emitter Flow Rate EFR 20 70 L/hr 3 Hydraulic Conductivity Parameter n HCPn 2.6 4.6 4 Initial Depletion Dini 0.09 0.19 5 Irrigation efficiency IE 79 89 % 6 Readily Available Coefficient KRA 0.03 0.13 7 Wetted Diameter WD 100 500 cm The selected variable used for tree characteristic was canopy volume (CV, volume of a citrus tree as function of tree age). Selected wa ter input variables were emitter flow rate (EFR, the volume of water discharged from the emitter w ithin a period of time), in itial depletion (Dini, the initial condition of de pletion), and irrigation e fficiency (IE, the percentage of water pumped into the irrigation system that actually gets distributed by the emitter) and wetted diameter (WD, the diameter of the irrigation emitter). The selected variables for water movement were soil hydraulic conductivity (HCPn), was assumed to same for all soil layers, and readily available coefficient (KRA, coefficient used to calculate th e readily available water content for uptake for uptake). The minimum and maximum values of factors six different variable s in Table 4-1 were applied. Six orientation matrices are generated ac cording to the Morris factor-screening design, and the respective elementary effects for 7 differe nt factors per orientation matrix are estimated 65

PAGE 66

from the simulation response for CWA. Following Morris OAT design 8 simulation configurations are generated for each of six orie ntation matrices. In each orientation matrix, the first row represents the base case (configuration) and the remaining 7 are used to determine the elementary effects for all 7 factors involved. After comparing the elementary effects of 7 factors for 6 different trajectories and the corresponding mean and variance of the distribution, factor 7, the WD, appears significantly separated from the other factors and it means that the wetted diameter dominate the simulation result. The WD increases valu e of CWA since it determine direct water input amount from irrigation event, but its effect compared with other factors was explained before this analysis. Whereas, factor 5 and 6, IE and KRA, has lower mean-variance relation value than other factors, so that model results are less sensitive those two factors were screened before sensitivity analysis. At the sensitivity analysis step, 35 complete factorial design makes 243 scenarios with the 5 factors selected by the Morris OAT screening-factor method and 3 different factor levels. The analysis of variance on the simulation resu lt of CWA was performed, which included interactions between two different factors. The results presented in Table 4-2 shows the sum of squares and the sensitivity index which is calculated by dividing th e sum of squares with the total variability. From the result of sensitivity analysis, th e CWA was apparently governed by the WD, and Dini with a significant impact on the system than other factors. The HCPn was more sensitive than the CV and the EFR. The EFR appeared as the least sensitive factor among the main factors. The WD related the water input to the amount of irrigation. Thus, wa ter balance could be affected significantly by the i rrigation amount when there was le ss rainfall. For the interaction between two factors, the interaction of the WD and the CV was more significant than others, and 66

PAGE 67

interactions with the HCPn had relatively larger value than other interactions because it contributed to increase water amount in deep soil layer. The CV affected rainfall into the system by blocking direct rain, and limited am ount of rain could reach to soil. Table 4-2. Sensitivity analysis result includi ng main effects and two-factor interactions Effects SS (x106) Sensitivity Index Effects SS (x106) Sensitivity Index cv 7,915 0.00102 cv*WD 7,847 0.00101 EFR 805 0.00010 EFR*HCPn 493 0.00006 HCPn 2,084 0.00027 EFR*dini 82 0.00001 dini 120x103 0.01545 EFR*WD 171 0.00002 WD 7.64x106 0.98102 HCPn*dini 63 0.00001 cv*EFR 145 0.00002 HCPn*WD 914 0.00012 cv*HCPn 993 0.00013 dini*WD 99 0.00001 cv*dini 19 0.00000 residuals 5,919 Sensitivity analysis result provides informati on about impact factors and related factors impacts. It may be useful to reconfigure model parameter and to create other models using these symbols. In Chapter 4, the CWMS model developed by Morgan et al. (Morga n et al., 2006a; Morgan et al., 2006b) is implemented using ontology-based simulation methodologies and tools covered in Chapter 3. Soil geometry structure is designed with 4 concepts, soil bloc k, soil cell, soil profile, and soil layer and their dimensions are defined wi th index concepts describing array size. With the existing mathematical models symbols and equations are defined and entered into the ontology, and they formed a model base contai ning symbols and equations and structuring relation between them. Model performance is te sted under two different simulation conditions, and using a sensitive analysis tool critical input factor s and the associated factor relations are revealed for water amount in the soil system Through these processes the CWMS model is created and executed efficiently by the graphic interface program. 67

PAGE 68

CHAPTER 5 SUMMARY AND FUTURE WORK The following methodologies were develope d utilizing ontology-based simulation techniques to build mathematical models. 1) The EquationEditor includes a symbol dictionary for ente ring symbols appearing in the equations along with their definiti ons and units. Symbols are defined by specific concepts. Equation are rendered visually using classic mathematical notation, but internally a hier archical data structure (t ree) is used for storing operators and symbols. The equal operator is the root node of the equation tree. Operators (like + and -) used in the equa tion become a node in the tree with child nodes being additional operat ors or symbols. The Equation Object Model (EOM) used in the Equation Editor is a collec tion of basic objects which represent information describing a mathematical e xpression and defining data type of attributes, and which communicate with ontology-based database sy stem to retrieve data. 2) The SimulationEditor incorporates the EquationEditor and is designed to represent the structure of dynamic systems using graphic elements. The SimulationEditor also contains facilities for automatica lly generating and running simulations and providing reports. The SimulationEditor provi des a graphic user interface to create and maintain a simulation system; a structure design interface for the simulation system, a simulation control interface, a simulation result reporting interface, and some additional interfaces including a math markup language generator and a statistical model analyzer. 68

PAGE 69

These methodologies were applied to develop a model of the Citrus Water Management System. Approximately 700 symbols and 500 equa tions are conceptualized and stored in ontology database using the SimulationEditor an d EquationEditor. A Java program for running the simulation was generated automatically from the modeling environment and incorporated into both a stand-alone application for grower and WAM. Th e modeling environment provides adequate tools to create and modify models without any programming knowledge. From the results, it can be concluded that ontology-based simulation offers a significant improvement in the methodology for building, publishing, and managing model. For the next step, it wi ll be useful to study model reusabil ity within the existing model base of the CWMS model. Issues on model reusability are related with the scale of models in the problem domain, which requires the consistency of the temporal and spatial scale to guaranty compatibility in models (Leon et al., 2002). For a similar scale domain problem with the CWMS model, such as models using different soil profiles, limitation placed on reusing these submodels are that they have rigi d and sophisticate connections with other processes such as a hydraulic conductivity which has strict requirement fo r spatial scale. On the other hand, from an extension study of the CWMS model, WAM, in which two different models cooperate independently keeping the intern al scale of the model may cause significant inefficiency of simulation time. There are needs for an effici ent way to organize and classify models and processes within a model base and for a flexible way to constructing a set of simulation model for different project and modeler by switching with a different process. Diverse level taxonomic categories may help the model base to be handled more intuitively. The data reusability could follow the model reus ability. Especially, relational databases are important sources of data for simulation model, but it is expensive to identify existing data, to 69

PAGE 70

determine the exact format of data and to use data in a model. Th ere has been research to convert directly relational databases di rectly to ontology using a mapping language (Barrasa et al., 2004) and to transform/service various heterogeneous database sources including database as ontologies (TopBraid, 2003). It is difficult to provide a formal way to use databases as ontologies since security and restriction le vel of data varies and case depends on the domain. Therefore, it will be useful to study sharing databases as a form of ontology for a simulation model in the agriculture domain to build an efficient si mulation model network. This study can include methods to publish/search databases, negotiate/c ommunicate automatically to get wanted data, and provide a protocol for networking within th e domain. Finally, a visual environment will be required to enhance the reusability of models and data and ontology since they have a complicated relationship and structure. There are some studies on solving complexity problem in displaying ontology visually (Bosca et al., 2005; TouchGraph, 2005), but it needs to develop methods focusing on the rela tions in model and data. 70

PAGE 71

LIST OF REFERENCES Alam, F. M. and K. R. McNaught, 2004, Using Morriss Randomized OAT Design as Factor Screening Method for Developing Simulation Meta models, Proc. Of the 2004 Winter Sim. Conf. Vol. 1:930-938 Athanasiadis, I. N., A. E. Rizzoli, M. Donatell i and L. Carlini, 2006, Enriching software model interfaces using onotlogy-based tools, iEMSs, Burlington, Vermont, July 2006 Ausbrooks, R., S. Buswell and D. Carlisle, 2003, Mathematical Markup Language (MathML) Version 2.0, http://www.w3.org/TR/MathML Barrasa, J., O. Corcho and A. Gomez-Perez, 2004, R2O, an Extensible and Semantically Based Database-to-ontology Mapping Language, Second Workshop on Semantic Web and Databases (SWDB2004). Toronto, Canada. August 2004 Beck, H. W., 2007, Lyra ontology management system, http://orb.at.ufl.edu/ObjectEditor/index.html Beck, H. W., L. G. Albrigo and S. Kim, 2004, DISC citrus planning and scheduling program, Proceeding of the Seventh International Sympos ium on Modelling in Fruit Research and Orchard Management: 25-32 Benjamin, P. C., M. Patki and R. J. Mayer, 2006, Using ontologies for simulation modeling, Winter Simulation Conference 2006: 1151-1159 Booch, G., J. Rumbaugh and I. Jacobson, 1997, The Unified Modeling Language user guide, Addson-Wesley Bosca, A., D. Bonino and P. Pellegrino, 2005, Onto Sphere: more than 3D ontology visualization tool, SWAP 2005, the 2nd Italian Semantic Web Workshop, Trento, Italy, December 14-16, 2005, CEUR Workshop Proceedings Buswell, S., O. Caprotti, D. P. Carlisle, M. C. Dewar, M. Gaetano and M. Kohlhase, 2004, The OpenMath Standard 2.0, http://www.openmath.org/standard/om20-2004-06-30/ Cuske, C., T. Dickopp and S. Seedorf, 2005, JO ntoRisk: An Ontology-based Platform for Knowledge-based Simulation Modeling in Financ ial Risk Management, European Simulation and Modeling Conference 2005 Donatelli, M., G. Bellocchi and L. Carlini, 2006a, Sharing knowledge vi a software components: models on reference evapotranspiration, Europ. J. Agronomy Vol. 24(2): 186-192 Donatelli, M., G. Bellocchi and L. Carlini, 2006b, A software component for estimating solar radiation, Environmental Modelli ng and Software Vol. 21(3): 411-416 71

PAGE 72

Eitzinger, J., M. Trnka, J. Hosch, Z. Zal ud and M. Dubrovsky, 2004, Co mparison of CERES, WOFOST and SWAP models in simulating soil water content during growing season under different soil conditions, Ecologi cal Modelling Vol. 171(3): 223-246 Ewert, F., H. Van Keulen, M. K. Van Ittersum, K. E. Giller, P. A. Leffelaar and R. P. Roetter, 2006, Multi-scale analysis and modelling of natu ral resource management, Proceedings of the iEMSs, Burlington, Vermont, July 2006 Fishwick, P. A. and J. A. Miller, 2004, Ontol ogies for Modeling and Simulation: Issues and Approaches, Proceeding of 2004 Winter Simulation Conference, Vol. 1:251-256 Forrester, J. W., 1971, World Dynamics, Cambridge, MA: Productivity Press: 144 Furmento, N., A. Mayer, S. McGough, S. Newhouse, T. Field and J. Darlington, 2001, Optimisation of component-based applications within a grid e nvironment, Proceedings of the 2001 ACM/IEEE conference on Supercomputing, Denver, CO, November 2001 Green, W. H. and G. Ampt, 1911, Studies of soil physics, part 1.-the flow of air and water through soils, J. Agricultural Science Vol. 4: 1-24 Gruber, T. R., 1995, Toward Principles for the Design of Ontologies Used in Knowledge Sharing, International Journal of Hu man Computer Studies Vol. 45:907-928 Guarino, N., 1997, Understanding, building and us ing ontologies, Int. J. Human-Computer Studies Vol.46, 293-310 Haan, C. T., H. P. Johnson and D. L. Brakensiek, 1982, Hydrologic modeling of small watersheds, ASAE Monograph No. 5:533 Islam, A. S. and M. Piasecki, 2004, A Stat egy for Web-Based Modeling of Hydrodynamic Processes, EM2004 June 13-16 Ittersum, M. K. v., F. Ewert, T. Heckelei, J. Wery, J. Alkan Olsson, E. Andersen, I. Bezlepkina, F. Brouwer, M. Donatelli, G. Flichman, L. Olsson, A. E. Rizzoli, T. van der Wal, J. E. Wien and J. Wolf, 2008, Integrated assessm ent of agricultural systems A component-based framework for the European Union (SEAMLESS), Agricu ltural Systems, Vol. 96(1-3):150-165 Jones, J. W., B. A. Keating and C. H. Porter, 2001, Approaches to modular model development, Agricultural Systems 70: 421-443 Jurisica, I., J. Mylopoulos and E. Yu, 2004, Ontologies for Knowledge Management: An Information Systems Perspective, Knowledge and Information Systems Vol. 6: 380-401 Knublauch, H., D. Oberle, P. Tetlow and E. Wa llace, 2006, A Semantic Web Primer for ObjectOriented Software Developers, W3 C Working Group Note 9 March 2006 72

PAGE 73

Lacy, L. and W. Gerber, 2004, Potential modelin g and simulation applications of the web ontology language OWL. WSC '04: Proceedings of the 36th conference on Winter simulation, Winter. Leon, L. F., D. Lam, S. Hamilton, N. Crookshank, D. Bonin and D. Swayne, 2002, Multi-model integration in decision support system: a technical user interface approach for watershed and lake management scenarios, Proceeding of 2002 iEMSs Vol. 3: 306 Lu, H.-Y., C.-T. Lu, M.-L. Wei and L.-F. Ch an, 2004, Comparison of Different Models for Nondestructive Leaf Area Estimation in Taro, Agron J Vol. 96(2): 448-453 MathType, 1996, Design Science, http://www.dessci.com/en/products/mathtype/ Mein, R. G. and C. L. Larson, 1973, Modeling in filtration during a steady rain, Water Resour. Res. Vol. 9(2): 384-394 Microsoft, 2003, Microsoft Office Equation Editor, http://office.microsoft.com/enus/word/HP051902471033.aspx Miller, J. A., G. T. Baramidze, A. P. Sheth and P. A. Fishwick, 2004, Investigating ontologies for simulation modeling. Simulation Sym posium, 2004. Proceedings. 37th Annual. Morgan, K. T., T. A. Obreza and J. M. S. Sc holberg, 2006a, Characterizi ng citrus tree root distribution in space and time, J. Am Soc. Hort. Sci. Vol. 131: 149-156 Morgan, K. T., T. A. Obreza, J. M. S. Sc holberg, L. R. Parsons and T. A. Wheaton, 2006b, Citrus water uptake dynamics on a sandy Florida Entis ol, Soil Sci. Soc. Am. J. Vol. 70(1): 90-97 Morris, M. D., 1991, Factorial Sampling Plans fo r Preliminary Computational Experiments, Technometrics Vol. 32(2) Muetzelfeldt, R. and J. Massheder, 2003, The Sim ile visual modelling environment, Europ. J. Agronomy Vol. 18: 345 Noy, N. F. and D. L. McGuinness, 2001, Technical Report KSL_01_05, Ontology Development 101: A Guide to Creating Your First Ontol ogy, Stanford Knowledge Systems Laboratory Park, M. and P. A. Fishwick, 2005, Integrat ing Dynamic and Geometry Model Components through Ontology-Based Interface, Simulation Vol. 81(12): 795-813 Peart, R. M. and R. B. Curry, 1998. Agricultur al systems modeling and simulation. New York, Marcel Dekker. Raubel, M. and W. Kuhn, 2004, Ontology-based task simulation, Spatial Cognition and Computation Vol. 4: 15-37 73

PAGE 74

Reddy, V. and V. Anbumozhi, 2004, DEVE LOPOMENT AND APPLICATION OF CROP SIMULATION MODELS FOR SUSTAINABLE NATURAL RESOURCE MANAGEMENT, International Agricultural Engineering Conference (IAEC) Richards, L. A., 1931, Capillary conduction th rough porous mediums, Physics Vol. 1: 313-318 Rizzoli, A. E., M. Donatelli, R. Muetzelfeldt, T. Otjens, M. G. E. Sevensson, F. v. Evert, F. Villa and J. Bolte, 2004, SEAMFRAME, A Proposal fo r an Integrated Modelling Framework for Agricultural Systems, Proc. of the 8th ESA Congress: 331-332 Rumbaugh, J., M. Blaha, W. Premerlani, F. Eddy and W. Lorensen, 1991, Object-oriented modeling and design, Englewood Cli ffs, New Jersy: Prentice Hall Scholten, H., A. Kassahun, J. C. Refsgaard, T. Kargas, C. Gavardinas and A. J. M. Beulens, 2007, A methodology to support multidisciplin ary model-based water management, Environmental Modelling & Software Vol. 22(5): 743-759 Steed, M., 1992, Stella, a simulation construc tion kit: cognitive process and educational implications, The Journal of Computers in Mathematics and Science Teaching Vol. 11(1): 39 TopBraid, 2003, TopQuadrant White Paper, http://www.topquadrant.com TouchGraph, 2005, TouchGraph White Paper, http://www.touchgraph.com/ Wallach, D., D. Makowski and J. W. Jones, 2007, Working with Dynamic Crop Models, Elsevier 74

PAGE 75

BIOGRAPHICAL SKETCH Yunchul Jung, hailing from Ulsan, Republic of Korea, finished his schooling from Haksung High School. He studied agricultural e ngineering and acquired a bachelors degree from Seoul National University, Seoul. He is currently working toward completion of his masters degree program in agricult ural and biological e ngineering at the University of Florida.