Citation
Development and Assessment of a Smartphone Application for Evapotransipiration Based Irrigation Scheduling of Field Corn

Material Information

Title:
Development and Assessment of a Smartphone Application for Evapotransipiration Based Irrigation Scheduling of Field Corn
Creator:
Diamond, Justice S
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (125 p.)

Thesis/Dissertation Information

Degree:
Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Agricultural and Biological Engineering
Committee Chair:
Dukes,Michael D
Committee Co-Chair:
Migliaccio,Kati White
Committee Members:
Rowland,Diane L
Graduation Date:
5/3/2019

Subjects

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

Notes

Abstract:
Irrigation is utilized to ensure productive yields in cropping systems. Water for irrigation is typically supplied by extracting groundwater through wells or surface water withdrawals from lakes, reservoirs, or rivers. Technological developments for extracting groundwater have led to an increase in groundwater withdrawals for irrigation. However, excessive pumping of groundwater for irrigation has caused reductions in river and spring flows in the Suwannee River Basin. Therefore, irrigation best management practices (BMPs) were developed and assessed to reduce groundwater usage for irrigation of field corn in Live Oak, Florida and Camilla, Georgia. The objectives of this study were to develop, calibrate, and assess a smartphone application for evapotranspiration-based (ET) irrigation scheduling for field corn. To quantify this, four irrigation treatments (calendar-based, corn app, soil moisture sensor (SMS), and non-irrigated) at Live Oak, and three irrigation treatments (corn app, smart sensor array (SSA), and checkbook) at Camilla were compared for their effects on marketable grain yield across different nitrogen fertilization regimes. The corn app treatment utilizes ET-based irrigation scheduling to fulfill exact crop water requirements. The calendar-based and checkbook methods are traditional irrigation scheduling methods that do not account for changes in soil moisture conditions in real time. The corn app achieved water savings of 43% and 56% with no significant differences in marketable grain yield compared to the calendar-based and checkbook irrigation scheduling methods, respectively, demonstrating implementation of irrigation BMPs can help reduce irrigation volumes without negatively impacting yields. ( en )
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.
Thesis:
Thesis (M.S.)--University of Florida, 2019.
Local:
Adviser: Dukes,Michael D.
Local:
Co-adviser: Migliaccio,Kati White.
Statement of Responsibility:
by Justice S Diamond.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
LD1780 2019 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

1 DEVELOPMENT AND ASSESSMENT OF A SMARTPH ONE APPLICATION FOR EVAPOTRANSP IRATION BASED IRRIGATION SCHEDULING OF FIELD CORN By JUSTICE S. DIAMOND A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTI AL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2019

PAGE 2

2 © 2019 Justice S. Diamond

PAGE 3

3 To my family, frie nds, and everyone along the way

PAGE 4

4 ACKNOWLEDGMENTS I thank my family for instilling the value of education in me. I thank my friends for providing guidance when I was in doubt. I would like to acknowledge my colleagues Jason Merrick, Maria Zamora, Juan Briceno, Marc Thomas, and M ichael Gutierrez for their invaluable help and advice. I would like to thank my advisors Michael Dukes and Kati Migliaccio for their advice and oversight as well as my committee member Diane Rowland for her expert input. I would like to thank the staff at the Suwannee Valley North Florida Research and Extension Center and Stripling Irrigation Research Park for providing the means and assistance to conduct this research. The inmates and staff at the Suwannee Correctional Institution deserve a special thank y ou for their work on this project. Finally, I would like to acknowledge the collaboration between the National Institute of Food and Agriculture and FACETS project team for their efforts in finding a balance between agriculture and resource conservation in the Suwannee, Santa Fe and Flint River basins.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ........... 4 LIST OF TABLES ................................ ................................ ................................ ...................... 7 LIST OF FIGURES ................................ ................................ ................................ .................... 9 LIST OF ABBREVIATIONS ................................ ................................ ................................ .... 11 ABSTRACT ................................ ................................ ................................ ............................. 13 CHAPTER 1 LITERATURE REVIEW ................................ ................................ ................................ ... 15 Evapotranspiration Concepts ................................ ................................ .............................. 16 Potential Evapotranspiration ................................ ................................ .................... 17 Reference Evapotranspiration ................................ ................................ .................. 17 Estimation of Reference Evapotranspiration ................................ ............................. 18 Actual Crop Evapot ranspiration ................................ ................................ ............... 19 Irrigation Scheduling ................................ ................................ ................................ ......... 21 Evapotranspiration based Irrigation Scheduling ................................ ....................... 21 Soil Moisture Sensor Based Irrigation ................................ ................................ ...... 23 Smartphone and Web based Tools for Irrigation Scheduling ................................ .... 23 2 DEVELOPMENT AND CALI BRATION OF A SMARTPHONE APP FOR EVAPOTRANSPIRATION BASED IRRIGATION SCHEDULING OF FIELD CORN ... 28 Introduction ................................ ................................ ................................ ....................... 28 Objectives and Hypothesi s ................................ ................................ ................................ . 30 Materials and Methods ................................ ................................ ................................ ....... 30 Experimental site ................................ ................................ ................................ ..... 30 Smartirrigation Corn App Model Description ................................ ........................... 31 Initial Input ................................ ................................ ................................ ....... 31 App functionality ................................ ................................ .............................. 32 Growing degree da y calculation ................................ ................................ ........ 33 Crop evapotranspiration calculation ................................ ................................ .. 35 Irrigation and rain input calculations ................................ ................................ . 36 Root depth calculation ................................ ................................ ...................... 37 Plant available soil water calculation ................................ ................................ 37 Root zone water deficit calculation ................................ ................................ ... 38 Calculation steps for corn app model ................................ ................................ 39 Soil Moisture Sensor Data Collection ................................ ................................ ...... 40

PAGE 6

6 Dat a Analysis ................................ ................................ ................................ .......... 41 Calibration ................................ ................................ ................................ ............... 42 Results ................................ ................................ ................................ ............................... 44 Model Adjustment and Calibration Res ults ................................ .............................. 44 Validation ................................ ................................ ................................ ................ 45 Discussion ................................ ................................ ................................ ......................... 45 Effect of Corn App Model Parameter Adjus tments ................................ .................. 45 Corn App Limitations ................................ ................................ .............................. 48 Soil Moisture Sensor Limitations ................................ ................................ ............. 50 Conclu sion ................................ ................................ ................................ ......................... 50 3 EVALUATION OF IRRIGATION APPLIED, YIELD, AND WATER USE EFFICIENCIES FOR DIFFERENT IRRIGATION SCHEDULING METHODS ON CORN ................................ ................................ ................................ ................................ 73 Intro duction ................................ ................................ ................................ ........................ 73 Objectives and Hypotheses ................................ ................................ ................................ . 75 Materials and Methods ................................ ................................ ................................ ....... 75 Experimenta l Site and Agronomic Practices ................................ ............................. 75 Irrigation Treatments at NFREC SV ................................ ................................ ........ 77 Corn app treatment ................................ ................................ ........................... 78 Soil moisture sensor treatment ................................ ................................ .......... 78 Calendar based treatment ................................ ................................ .................. 79 Non irrigated treatment ................................ ................................ ..................... 80 Irrigation Treatments at SIRP ................................ ................................ ................... 80 Checkbook method ................................ ................................ ........................... 80 Soil water potential sensors ................................ ................................ ............... 82 Effect of Irrigation Depth on Corn Yield ................................ ................................ .. 82 Data Analysis ................................ ................................ ................................ .......... 83 Water Use Efficiency ................................ ................................ ............................... 84 Crop Water Use Efficiency ................................ ................................ ...................... 85 Measured Soil Moisture for each Irrigation Treatment ................................ ............. 85 Res ults and Discussion ................................ ................................ ................................ ....... 86 Weather Conditions ................................ ................................ ................................ . 86 Effect of Irrigation and Fertilization on Marketable Grain Yield and IWUE at NFREC SV ................................ ................................ ................................ .............. 88 Effect of Irrigation and Fertilization on Marketable Grain Yield and IWUE at SIRP ................................ ................................ ................................ ........................ 91 Soil Moisture Comparison ................................ ................................ ....................... 93 Conclusion ................................ ................................ ................................ ......................... 94 LIST OF REFERENCES ................................ ................................ ................................ ........ 120 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ... 125

PAGE 7

7 LIST OF TABLES Table page 1 1 Field corn crop coefficients ................................ ................................ ............................ 25 1 2 Applications and web based tools for improved irrigation scheduling ............................ 26 2 1 Corn app u ser inputs ................................ ................................ ................................ ...... 66 2 2 Phenological growth stage as a function of accumulated GDD ................................ ....... 67 2 3 Crop coefficient (Kc) as a function of accumulated GDD ................................ .............. 68 2 4 Soil textures and corresponding soil water holding capacity ................................ ........... 69 2 5 Corn app model calibration results ................................ ................................ ................. 70 2 6 Calibrated corn app model validation results. ................................ ................................ . 72 3 1 Irrigation treatments at NFREC SV ................................ ................................ ............... 96 3 2 Irrigation by fertilization treatments and corresponding plot numbers ............................ 99 3 3 Field properties for both testing sites. ................................ ................................ .......... 100 3 4 NFREC irrigation treatments and descriptions ................................ ............................. 101 3 5 SIRP irrigation treatments and descriptions ................................ ................................ . 102 3 6 Estimation of corn water use for checkbook irrigation scheduling at SIRP ................... 103 3 7 Comparison of marketable grain yield, irrigation applied, IWUE RF , IWUE, and CWUE at NFREC SV ................................ ................................ ................................ . 110 3 8 Type III tests for significant interactions at NFREC SV ................................ ............... 111 3 9 Type III tests for significant interactions between fixed effects and IWUE at NFREC SV ................................ ................................ ................................ ................. 113 3 10 Comparison of marketable grain yield, irrigation applied, IWUE, and for irrigation treatments at SIRP ................................ ................................ ................................ ....... 115 3 11 T ype III tests for significant interactions between fixed effects and marketable grain yield at SIRP ................................ ................................ ................................ ............... 115 3 12 Type III tests for significant interactions between fixed effects and IWUE for the trea tments at SIRP ................................ ................................ ................................ ....... 116

PAGE 8

8 3 13 Estimation of number of days each irrigation treatment exceeded a 50% allowable depletion threshold ................................ ................................ ................................ ...... 119

PAGE 9

9 LIST OF FIGURES Figure page 2 1 User selection of field location ................................ ................................ ....................... 52 2 2 User input of field name, planting date and choice of weather da ta source ..................... 53 2 3 User selection of available soil water holding capacity ................................ ................... 54 2 4 User input of irrigation type, irrigation system eff iciency, and root zone water notification threshold ................................ ................................ ................................ ..... 55 2 5 Field selection from corn app main menu ................................ ................................ ...... 56 2 6 Corn app f ield menu ................................ ................................ ................................ ...... 57 2 7 Forecast menu provided in the corn app ................................ ................................ ......... 58 2 8 User option to send report of maintained soil water balance ................................ ........... 59 2 9 Field layout and location of SMS sensors beneath app treatment plots at NFREC SV .... 60 2 10 Residual plot s between uncalibrated corn app model and measured SMS data ............... 61 2 11 Growth staging photos for field corn ................................ ................................ .............. 62 2 12 Comparison of predicted RZWD (%) from uncalibrated and calibrated corn app model and meas ured SMS data for the months of April, June, and August ..................... 63 2 13 Residual plot and comparison of predicted RZWD (%) from validated corn app model and measured SMS data for the months of May and July ................................ ..... 64 2 14 Comparison of irrigation applications by SMS treatments and calibrated corn app model simulated for the 2018 growing season ................................ ................................ 65 3 1 Aerial view of 2018 field layout at NFREC SV. ................................ ............................ 96 3 2 Various growth stages o f corn crop observed at NFREC SV ................................ .......... 97 3 3 Aerial view of SI RP field layout ................................ ................................ .................... 98 3 4 Conditional residual plots of market grain yield for data collected at NFREC SV ........ 105 3 5 Conditional residual plots of market grain yield for data collected at SIRP .................. 106 3 6 Total monthly rainfall and temperatures recorded by the Live Oak FAWN station ....... 107 3 7 Average monthly rainfall and temperature recorded by the Camilla GAEMN station ... 108

PAGE 10

10 3 8 Precipitation events and application totals by irrigation treatment at NFREC SV ......... 109 3 9 Boxplot comparison of marketable grain yield in response to irrigation and fertilization regimes at NFREC SV ................................ ................................ .............. 112 3 10 Boxplot comparison of IWUE in res ponse to irrigation and fertilization regimes at NFREC SV. ................................ ................................ ................................ ................ 113 3 11 Precipitation events and application totals by irrigation treatment at SIRP. .................. 114 3 12 Boxplot comparison of marketable grain yield in response to irrigation and fertilization regimes at SIRP ................................ ................................ ........................ 116 3 13 Boxplot comparison of IWUE in response to in response to irrigation and fertilization regimes at SIRP ................................ ................................ ........................ 117 3 14 C omparison of soil moisture within effective root zone estimated by soil moisture probes for each irrigation treatment at NFREC SV ................................ ...................... 118

PAGE 11

11 LIST OF ABBREVIATIONS AWHC Available soil water holding capacity BMP Best management practices CWUE Crop water use efficiency DAP Days after planting DHUs Daily heat units ET c Actual crop evapotranspiration ET o Reference e vapotranspiration ET Evapotranspiration FAO Food and Agricultural Organization FAWN Florida Automated Weather Network FC Field capacity FRET Forecasted Reference Evapotranspiration GAEMN University of Georgia Automated Environmental Monitoring Netwo rk GDDs Growing degree days High High nitrogen fertilization treatment of 336 kg/ha applied at SIRP I Effective irrigation IWUE Irrigation water use efficiency IWUE RF Rain fed irrigation water use efficiency Kc Crop coefficient Low Low nitrogen fer tilization treatment of 277 kg/ha applied at SIRP MAD Maximum allowable depletion N1 Low nitrogen fertilization treatment of 112 kg/ha applied at NFREC SV N2 Medium nitrogen fertilization treatment of 224 kg/ha applied at NFREC SV

PAGE 12

12 N3 High nitrogen fert ilization treatment of 336 kg/ha applied at NFREC SV NEXRAD Net Generation Weather Radar NFREC SV Suwannee Valley North Florida Research and Extension Center NOAA National Oceanic and Atmospheric Association NSE Nash Sutcliffe efficiency NWS National Weather Service PAW Plant available water R Effective rain RDI Daily root depth increase RD i Daily root depth RMSE Root mean square error Rz initial Initial root depth Rz max Maximum root depth RZMC Root zone moisture content RZWD Root zone water de ficit SIRP Stripling Irrigation Research Park SMS Soil moisture sensor SRB Suwannee River Basin SRWMD Suwannee River Water Management District SSA Smart sensor array T base Base temperature T max Daily maximum temperature Traditional Nitrogen fertili zation treatment of 333 kg/ha applied at SIRP UF IFAS UGA University of Georgia

PAGE 13

13 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science DEVELOPMENT AND ASSESSMENT OF A SMARTPHONE APPLICATION FOR EVAPOTRANSPIRATIO N BASED IRRIGATION SCHEDULING OF FIELD CORN By Justice S. Diamond May 2019 Chair: Michael D. Dukes Cochair: Kati Miglia ccio Major: Agricultural and Biological Engineering Irrigation is utilized to ensure productive yields in cropping systems. Water for irrigation is typically supplied by extractin g groundwater through wells or surface water withdrawals from lakes, reserv oirs, or rivers. Technological developments for extracting groundwater have led to an increase in groundwater withdrawals for irrigation. However, excessive pumping of groundwater for irrigation has caused reductions in river and spring flows in the Suwann ee River Basin. Therefore, irrigation best management practices (BMPs) were developed and assessed to reduce groundwater usage for irrigation of field corn in Live Oak, Florida and Camilla, Georgia. The objective s of this study were to develop , calibrate , and assess a smartphone application for evapotranspiration based (ET) irrigation scheduling for field corn . To quantify thi s, four irrigation treatments (c alendar based, corn app, soil moisture sensor (SMS) , and non irrigated) at Live Oak , and three irri g ation treatments (co rn app, smart sensor array (SSA) , and checkbook ) at Camilla were compared for their effects on marketable grain yield across different nitrogen fertilization regimes. The corn app treatment utilizes ET based irrigation scheduling to fu lfill exact crop water requirements . The calendar based and checkbook methods are traditional irrigation scheduling methods that do not account for changes in soil moisture conditions in real time.

PAGE 14

14 The corn app achieved water savings of 43% and 56% with no significant differences in marketable grain yield compared to the calendar based and checkbook irrigation scheduling methods , respectively, demonstrating implementation of irrigation BMPs can help reduce irrigation volumes without negatively impacting yie lds.

PAGE 15

15 CHAPTER 1 LITERATURE REVIEW Field c orn accounts for more than 95% of total feed grain production in the United States. (USDA, 2018b) . While corn production is concentrated in the Midwest, corn is also grown in Florida and southern Georgia for grain and silage (USDA, 2018b; Wr ight, 2004) . Although field corn ac reage in Florida decreased from the 1970s 2000 , yields per acre more than doubled across the same time frame (Wright, 2004) . Advances in corn yields can be attributed to improvements in irrigation technology, increased irrigated acreage, and improved hybrids (Pioneer, 2018; Wright, 2004) . Groundwater withdrawals from the Floridan Aquifer have increased from 2,384,550 m 3 /day i n 1950 to 15,215,700 m 3 /day in 2000 (Marella and Berndt, 2005) . Agricultural withdrawals account for half of that increase and irrigated acreage in the Suwannee River Water Management District (SRWMD) is expected to increase by 40% from 2015 to 2040 (Berndt, 2014; The Balmoral Group, 2018) . Field corn and peanuts are typically grown in rotation with each other in northern Florida and southern Georgia to reduce pressure from pests, weeds, and disease (Wright et al., 2013) . Field corn and peanuts accounted for 54% of sprinkler irrigated crops in the SRWMD in 2015 (Marella et al., 2016) . Irrigation is essentia l to ensure productive corn yields by reducing the stress of dry conditions and counteracting variability from variable weather patterns and low soil water holding capacity (Porter, 2017) . However, increased competition among freshwater users and environmental regulations are negatively impacting agricultural water security in the region (Marella and Berndt, 2005) . The Upper Floridan Aquifer is a critical source for agricultural and municipal irrigation in Northern Florida and Southern Georgia (Marella and Berndt, 2005) . The Upper Floridan

PAGE 16

16 Aquifer supplies drinking water for 10 million people and supports agricultural activities worth more than $7.5 billion dollars (Berndt, 2014; Hodges et al., 2017) . Increased municipal and agricultural pumping from the Upper Floridan Aquifer has led to large regional drawdowns reducing the area that contributes water to the Suwannee River Management Distri ct and reducing flows to the Suwannee River (Grubbs and Crandall, 2007 ) . The Suwannee River has experienced a 60% decline in base flow since 1940 and more than 5180 km 2 of groundwater contribution area to the Suwannee River Basin has been reduced since predevelopment (Grubbs and Crandall, 2007) . Increased groundwater withdrawals beyond the rate of recharge threaten water resources globally b y contributing to issues such as subsidence, ecosystem loss, saltwater intrusion, economic damage to downstream communities, and changes in natural river flows (Mitra et al., 2016) . The Ogallala Aquifer, which unde rlies the Great Plains region, accounts for 30% of the total irrigated acreage in the U.S. and has experienced substantial groundwater depletion due to over pumping for corn and cattle production (Davi d et al., 2013) . An estimated 30% of the high plains aquifer has been depleted and an additional 39% will be over withdrawn over the next fifty years given existing trends (David et al., 2013) . To av oid similar circumstances that have occurred from over pumping in the Midwest, best management practices (BMPs) are being developed and assessed in the Suwannee River Basin (SRB) to reduce agricultural water usage and nutrient loading in the Floridan Aquif er. Reduced pumping in the Floridan Aquifer can be accomplished through efficient water use strategies such as irrigation scheduling. Evapotranspiration Concepts T plant transpiration can be defined as evapotranspiration (ET) (Al len et al., 1998) . Weather

PAGE 17

17 parameters that affect ET include solar radiation, air temperature, relative humidity, and wind speed (Allen et al., 1998) . Crop factors that affect ET are crop variety, phenological growth stage, and planting density (Allen et al., 1998) . Evapotranspiration can also be impacted by environmental conditions such as soil salinity, fertility, ty pe and grower practices such as pest and soil management (A llen et al., 1998) . Evapotranspiration based estimation methods vary in their complexity from the use of ET cont rollers to hand calculations. Evapotranspiration can be measured directly by determining a soil water balance using lysimeters or an energy ba lance with eddy covariance systems (Allen et al., 1998) . H owever, direct measurement of ET can be very expensive and difficult to implement (Allen et al., 1998) . Evapotranspiration can also be estimated using meteorological data and a variety of models. Potential Evapotranspiration Potential evapotranspiration is a concept that was first introduced by Penman in the 194 height, complete ground cover, and sufficient water within the soil profile (Jensen and Allen, 2016) . However, this def inition has caused confusion due to the lack of specification of which green crop to use in the calculation. Utilizing the Bowen ratio to cancel out terms related to calculations at the evaporative surface, Penman combined the aerodynamic and energy balanc e equations to develop a combination equation to estimate ET (Jensen and Allen, 2016) . This combination equation only requires meteorological measurements of air temperature, humidity, wind speed, and solar rad iation (Jensen and Allen, 2016) . Reference Evapotranspiration R eference evapotranspiration (ET o ) was introduced in the late 1970s to provide a more accurate estimation of ET by adopting a reference crop which has allowed for more consistent

PAGE 18

18 selection of crop coefficients across various locations and climates (Jensen and Allen, 2016) . In addition, Monteith added a surface resistance term into the original Penman equa tion and inserted an aerodynamic resistance criteria in place of the linear wind function term (Jensen and Allen, 2016) . Reference evapotranspiration can be defined as evaporation from a grass reference crop wi th a fixed height of 0.12 m, an assumed surface resistance of 70 s m 1 , and an albedo of 0.23 (Allen et al., 1998) . Reference evapotranspiration is determined independently of crop type, crop growth, and management practices , and provides an estimation of the evaporative demand of the atmosphere (Allen et al., 1998) . Estimation of Reference Evapotranspiration Though re ference evapotranspiration can be determined by using a variety of equations, the Penman Monteith equation has become widely accepted to calculate reference ET o for irrigation management (Jensen and Allen, 2016) . The Penman Monteith equation is a combination based equation which incorporates both temperature and radiation based reference evapotranspiration equations. This calculation requires temperature, relative humidity, solar radiation, and wind speed measur ements taken in an open area with low lying, well watered vegetation (Jensen and Allen, 2016) . For this study, ET o was calculated using the ASCE Penman Monteith Food and Agricultural Organization ( FAO 56 ) equa tion (Equation 1 1) and real time weather data from the Florida Automated Weather Network (FAWN) and University of (Allen et al., 1998) . Reference evapotranspiration provides a guideline for comparing ET o in various regions or different parts of the ye ar as well as determining evapotranspiration of different crops (Allen et al., 1998) .

PAGE 19

19 (1 1) ET o represents the evapotranspiration from a reference surface [mm d 1 ] 1 ] Rn is the net radiation at the crop surface [MJ m 2 day 1 ] G is the soil heat flux density [MJ m 2 day 1 ] T is the mean daily air temperature at 2 m height [°C] u 2 is the wi nd speed at 2 m height [m s 1 ] e s is the saturation vapor pressure [kPa] e a is the actual vapor pressure [kPa] e s e a is the saturation vapor pressure deficit [kPa] 1 ]. Actual Crop Evapotranspiration Actual crop evapotranspiration (ET c ) can be defined as the evapotranspiration from healthy, well fertilized crops, grown in extensive fields, with sufficient soil water conditions and achieving full production growth (Allen et al., 1998) . Calculating a functional estimation for actual evapotranspiration typically involves m ultiplying a reference crop evapotranspiration by the desired crop coefficient (Allen et al., 1998) . A crop coefficient or K c is a combined factor surroundings, and management applications (Allen et al., 1998) . The K c also combines the characteristics of a crop that differ entiates it from the reference crop that is used to approximate reference ET (Allen et al., 1998) . The crop coefficient varies during the growth cycle of the crop

PAGE 20

20 depending on its developmental stages (Allen et al., 1998) . With so much variability in determining the K c, there are not a lot of locally adapted K c values for most crops in Florida (Dukes et al., 1995) . One way to determine a crop coefficient is to take the ratio of crop evapotranspiration (ET c ) to reference evapotranspiration (ET o ) (Piccinni et al., 2009) . Many studies have utilized this ratio to determine locally adapted crop coefficients for corn (Djaman and Irmak, 2013; Howell et al., 2006; Kang et al., 2003; Piccinni et al., 2009) . A study by (Piccinni et al., 2009) was con ducted at the Texas AgriLife Research Center in Ulvade, TX and utilized lysimeters to determine daily corn ET c by subtracting lysimeter mass gains by lysimeter mass losses. Grass ET o was determined using direct measurements from the lysimeter and calculat ions using the ASCE Penman Monteith Equation (Piccinni et al., 2009) . A growth stage specific crop coeffici ent for corn was developed by using the ratio of lysimeter corn crop ET c to the grass lysimter ET o (Piccinn i et al., 2009) . The following growth stage specific crop coefficients were developed for corn: K c at emergence = 0.35, K c at tasseling = 1.00 and K c at black layer = 0.90 (Piccinni et al., 2009) . These value s were compared to growth stage specific K c values developed in Bushland, Texas by Howell et al. (2006), K c values reported from FAO 56, and, K c values de veloped by the Kansas state mobile irrigation lab (2014) and Allen et al., (1998) (Table 1 1). The initial corn K c value of 0.35 developed in U val de, Texas by Piccinni et al., (2009) was larger than the initial corn K c value of 0.1 reported by Howell et. a l (2006) in Bushland, Texas. Di stinct K c values developed between regions can be attributed to differences in methods in measurement, local weather conditions, and variety of the crop grown. Differences in K c values demonstrate that crop coefficients shoul d be developed regionally to promote accuracy in determining evapotranspiration and improve efficiency in irrigation management (Piccinni et al., 2009) .

PAGE 21

21 Irrigation Scheduling Irrigation scheduling decisions are made using a combination of three tools: estimating evapotranspiration to determine crop water requirements, in field observation, and soil moisture se nsors (Buhrig and Shock, 2015) prior knowledge and experience with the crop and historical weather patterns to develop a calendar based schedule (Migliaccio et al., 2010 ) . Today, numerous irrigation innovations have been developed to improve irrigation scheduling. These developments include evapotranspiration (ET) and soil moisture sensor (SMS) based irrigation which uses real time weather data and soil water measurement s, respectively, to determine exact crop water requirements (Bartlett et al., 2015; González Perea et al., 2017; Mbabazi et al., 2017; Vellidis et al., 2016) . Evapotranspiration based Irrigation Scheduling Previo us research has shown evapotranspiration based irrigation scheduling provides water savings compared to traditional techniques. A study by Lamm and Rogers (2015) simulated ET based irrigation scheduling for corn using 43 years of weather data in Colby, Kan sas and found potential water savings of 212 mm for a 25.4mm/4 days irrigation capacity system compared to non science based irrigation scheduling. Utilizing the previous five days of estimated crop evapotranspiration (ET c ) to schedule irrigation for avoca do production in south Florida, Mbabazi et al., (2017) obtained 62 67% water savings compared to traditional time based methods. Evapotranspiration based irrigation scheduling methods often incorporate a soil water balance model to estimate crop water re quirements (Allen et al., 1998; Kisekka et al., 2014; Vellidis et al., 2016; Zamora Re and Dukes, 2017) . A study conducted in Live Oak, FL achieved

PAGE 22

22 42 and 39% water savings for field corn compared to a traditional calendar based irrigation scheduling method by utilizing a soil water balance approach that calculates daily ET c to estimate soil moisture and schedule irrigation in 2015 and 2016, respectively (Zamora Re and Dukes, 2016) . Utilizing daily calculated ET c to predict crop water use and a soil water balance that accounts for measured precipita tion and irrigation to estimate plant available soil water, a study by Vellidis et al., (2016) attained water savings of 76, 40, and 23% for cotton production compared to a checkbook method that does not account for environmental conditions in 2013, 2014 a nd 2015, respectively. Water savings utilizing evapotranspiration based irrigation scheduling can also be achieved through replacing a percentage of ET c demand. In a study by El Wahed and Ali (2013), the effects of three ET based irrigation scheduling meth ods (100%, 85%, and 70% replacement of ET c ), two types of irrigation systems (drip and sprinkler), and five mulching treatments were compared and assessed for water use efficiency in corn. This study found the highest water use efficiency and yields for th e treatment that replaced 100% of actual evapotranspiration under a drip irrigation system with 20 tons per hectare of manure incorporated within the surface layer of soil. However, the limited irrigation treatment that replaced 85% of ET c was found to pro duce comparable grain yield while saving 15% of water applied and thus improving net profit (El Wahed and Ali, 2013) . A study conducted in the Texas High Plains found that irrigating at 75% of ET c demand resulted in similar corn yields and improved water use efficiency by approximately 10% for two drought tolerant AQ UAmax corn hybrids compared to an irrigation treatment that replaced 100% of ET c demand (Hao et al., 2015) .

PAGE 23

23 Soil Moisture Sensor Based Irrigation Soil water sensors are inserted in the soil within the crop r oot zone and utilize soil water content as feedback to control irrigation (Grabow et al., 2012) . A study by Zamora Re and Dukes (2017) achieved 53 and 43% water savings with no significant differences in corn yield among treatments using soil moisture sensor (SMS) based irrigation compared to a calendar based irrigation scheduling method in 2015 and 2016, respectively (Zamora Re and Dukes, 2017) . A study conducted at University of N ebraska Lincoln at experiment stations found water savings of 34 and 32% with little to no reduction in corn yield using a soil moisture sensor based irrigation regime in comparison to traditional farmer irrigation management in 2005 and 2006, respectively (Irmak et al., 2012) . Farmer managed fields relied on traditional visual observations, calendars, and per sonal experience, whereas the soil moisture sensor based fields utilized pre determined soil water depletion thresholds checked with soil moisture sensors and crop phenology predicted by crop simulation models using real time and historical weather data (Irmak et al., 2012) . Smartphone and Web based Tools for Irrigation Scheduling The increased use of inter net enabled smart devices has allowed for the development of a suite of Smartirrigation applications which provide easy to use tools to improve irrigation scheduling ( https://smartirrigationapps.org ) ( Migliaccio et al., 2016) . The corn app generates a root zone water deficit using daily estimated crop evapotranspiration (ET c ), root depth, and soil water holding capacity. Reference evapotranspiration is calculated using the ASCE Penman Monteith FAO 56 e quation (Allen et al., 1998) and real time weather data fr om the Florida

PAGE 24

24 Monitoring Network (GAEMN). Numerous apps and web based tools have been developed throughout the world to improve irrigation scheduling for a variety of cro ps. (Table 1 2) While ET based irrigation scheduling has been shown to provide water savings and improve water use efficiency for a variety of agronomic crops in comparison to non science based irrigation scheduling methods, the question of how well the Sm artirrigation corn app performs in this regard remains. To address this question, this thesis presents a series of soil moisture deficits and ability to improve irrigation scheduling compared to traditional techniques. A smartphone application for ET based irrigation scheduling for field corn was developed. A comparison between daily corn app root zone soil moisture deficit predictions and measured soil moisture sensor data was performed. ET based irrigation scheduling for field corn was conducted to determine if water savings were possible without negatively impacting marketable grain yields. Application depths , yields and water use efficiencies for different ir rigation scheduling methods (app, SMS, calendar based, and no irrigation) and (app, SSA, checkbook) were compared.

PAGE 25

25 Table 1 1. Various f ield corn crop coefficients . Field corn crop coefficients developed by (A) Piccinni et. al., (2009) in Uvalde, Texas, (B) Allen et al., (1998) for FAO 56, (C) Howell et al., (2006) in Bushland, Texas (D) K State Research & Extension Mobile Irrigation Lab 2014 for humid and moderate Growth Stage K c (A) K c ini 0.35 K c mid 1.00 1.20 K c end 0.90 (B) K c ini 0.30 K c mid 1.20 K c end 0.35 (C) K c ini 0.1 K c mid 1.10 (D) K c ini 0.25 K c mid 1.05 K c end 0.55

PAGE 26

26 Table 1 2 . Applications and web based tools for improved irrigation scheduling App Name and Developer Scheduling Tool Utilized Water Savings by Netafim (2017) Combines inputs such as crop growth stage, soil type, and germination duration with weather data to develop a drip irrigation schedule for corn. Has not been assessed for pot ential water savings yet. by University of Nebraska (Yang et al., 2017) Online app for corn and soybean that predicts irrigation needs based on available s oil water for the crop, phenological growth stage, and potential for crop water stress at present and for the future. Currently being assessed at research plots and growers fields for potential water savings. by (González Perea et al., 2017) Takes in climate data, soil type, and information related to irrigation system to produce daily/weekly required irrigation times for strawberry production. Achieved water savings from 11 33% compared to growers traditional methods in several commercial farms Sm artirrigation Cotton App by (Vellidis et al., 2016) Pulls meteorological d ata from closest weather station in FAWN or GAEMN networks to calculate ET o . ET c is then calculated using a crop coefficient as a function of phenological growth stage. The app produces a root zone water deficit (RZWD) percent based on a daily soil water d eficit divided by plant available soil water. The user decides when to irrigate but a 40% RZWD is recommended to trigger irrigation. Improved water use efficiency and applied less water compared to traditional checkbook methods across three years in conve ntional and conservation tillage field trials. Water Irrigation Scheduling for Efficient Application by (Bartlett et al., 2015) Evapotranspiration based irrigation tool that utilizes the soil water balance method and data from the Colorado Agricultural Meteorolo gical Network and Northern Colorado Water Conservation District weather stations. Allows users to view soil moisture deficit, weather measurements and the ability to input applied irrigation amounts into the App. Has not been evaluated yet for potential w ater savings.

PAGE 27

27 Table 1 2. Continued App Name and Developer Scheduling Tool Utilized Water Savings Smartirrigation Avocado by (Mbabazi et al., 2017) Utilizes average of previous five days of ET c to develop irrigation schedule for avocado. Sixty two to sixty seven percent water savings were achieved with the app irrigation schedulin g methods compared to a traditional time based irrigation model. by University of North Dakota Web based tool that generates irrigation recommendations based on soil type, weather data, phenological growth stage, and daily crop water use. Has not been assessed for potential water savings but has been used for irrigation scheduling in corn. Irrigation Scheduler Mobile by (Peters et al., 2013) Daily weather data is automatically pulled from networks in 11 different states to calculate and utilize ET o . Incorporates a soil water balance model to perform check book style irrigation scheduling. Produces a one week forecast of crop water us and soil water content for irrigation scheduling. Works for a wide variety of crops such as hops, sweet and grain corn, and broccoli. Has not been assessed for potential water savings. by Kansas State University ( K State Research & Extension Mobile Irrigation Lab, 2014) ET based irrigation scheduling tool that uses real time crop and weather data to calculate daily crop water rate use. Daily crop water use is combined with soil root zone data to determine an irr igation schedule that maintains adequate soil moisture levels. Has not been assessed for potential water savings.

PAGE 28

28 CHAPTER 2 DEVELOPMENT AND CALIBRATION OF A SMARTPHONE APP FOR EVAPOTRANSPIRATIO N BASED IRRIGATION SCHEDULING OF FIELD CORN Introduction The Floridan Aquifer is an expansive groundwater system that stretches over 160,000 km 2 and encompasses most of Florida and parts of Alabama, Georgia, and South Carolina (Bellino et al., 2018) . The Floridan Aquifer system consists of the Upper Floridan Aquifer and Lower Floridan Aquifer (Bellino et al., 2018) . Groundwater withdrawals from the Floridan Aquifer have increased from 2,384,550 m 3 /day in 1950 to 15,215,700 m 3 /day in 2000 and approximately 90% of the withdrawals in 2000 came from the Upper Floridan Aquifer (Bellino et al., 2018; Berndt, 2014) . The Upper Floridan Aquifer sup plies drinking water for 10 million people and supports agricultural activities worth more than $7.5 billion dollars (Berndt, 2014; Hodges et al., 2017) . Increased municipal and agricultural pumping from the Upper Floridan Aquifer has led to large regional drawdowns reducing the groundwater contribution area and flows to the Suwannee River (Grubbs and Crandall, 2007) . The Suwannee River has experienced a 60% decline in base flow since 1940 and more than 13,416 km 2 of groundwater contribution area to the Suwannee River Basin has been reduced since predevelopment (Grubbs and Crandall, 2007) . Field corn and peanuts are typically grown in rotation with each other in northern Florida and southern Georgia to reduce pressure from pests, weeds, and disease (Wright et al., 2013) . Field corn and peanuts accounted for 54% of sprinkler irrigated crops in the Suwannee River Management District (SRWMD) in 2015 (Marella et al., 2016) . In 2015, irrigation for agriculture accounted for 54% of groundwater withdrawals in the SRWMD ( Marella and Dixon, 2018) . The

PAGE 29

29 Suwannee groundwater basin contains numerous freshwater springs that discharge into the lower Suwannee and Santa Fe Rivers (Bellino et al., 2018) . A well located at the center of the Suwannee groundwater basin recorded an annual decline of approximately 3 mm per year from 1980 2010 (Bellino et al., 2018) . Increased groundwater withdrawals beyon d the rate of recharge threaten water resources globally by contributing to issues such as subsidence, ecosystem loss, saltwater intrusion, economic damage to downstream communities, and changes in natural river flows (Mitra et al., 2016) . Irrigation scheduling decisions are made using a combination of three tools: in field observation, utilizing evapotranspiration calculations to estimate crop water requirements and soil moisture sensors (Buhri g and Shock, 2015) . Irrigation scheduling can increase irrigation efficiency by decreasing runoff, deep percolation, and soil evaporation losses as well as managing soil water content to reduce ET during less water demanding crop growth stages (Howell, 1996) and experience with the crop and historical weather patterns to develop a ca lendar based schedule (Migliaccio et al., 2010) . Evapotranspiration (ET) and soil moisture sensor (SMS) based irrigation utilize real time weather data and soil water measurements, respectively, to determin e exact crop water requirements (Bartlett et al., 2015; González Perea et al., 2017; Vellidis et al., 2016) . The increased use of internet enabled smart devices has allowed for the development of a suite of Smartirrigation applications which provide easy to u se tools to improve irrigation scheduling ( https://smartirrigationapps.org ) (Migliaccio et al., 2016) . Evapotranspiration based irrigation scheduling has been shown to reduce water usage and improve w ater use efficiency for corn production (Lamm and Rogers, 2015; Xue et al., 2017; Zamora Re and Dukes, 2017) . A

PAGE 30

30 model for a smartphone application that utilizes ET based irrigation scheduling for field corn was dev eloped and tested in this study. The Smartirrigation corn app generates a root zone water deficit using daily estimated actual evapotranspiration (ET c ), root depth, and available water holding capacity. Reference Evapotranspiration (ET o ) is calculated usin g the ASCE Penman Monteith FAO 56 equation (Allen et al., 1 998) and real time weather data from the Florida Monitoring Network (GAEMN). Objectives and Hypothesis The objectives in this study were to (1) Develop a model for a sma rtphone application for irrigation scheduling in field corn (2) compare daily outputs from the corn app model with measured soil moisture sensor data and (3) calibrate corn app parameters to impr ove the accuracy of the model in predicting root zone soil mo isture deficits . It is hypothesized that the corn app model produces a Nash Sutcliffe efficiency value greater than zero when compared with mea sured soil moisture sensor data. Materials and Methods Experimental site This study was conducted in Live Oak , FL, USA, at the of Food and Agricultural Sciences (UF IFAS) Suwanee Valley North Florida Research and Extension Center (NFREC SV) (Figure 2 9). The soils at this site are characterized as sandy (95.9% sand, 2.3% silt and 1.8% clay) (USDA, 2018). The experimental design for the NFREC field site was a randomized complete block organized in a split plot. This design include d four

PAGE 31

31 replicates or blocks. Each plot was approximately 12 m long x 6 m wide separated by 6 m alleys. An alley of 12 m is included between the blocks (Figure 2 5). There were four irrigation (app, SMS, calendar based, and no irrigation) and three fertilization treatments (N1 112 kg N/ha; N2 224 kg N/ha; N3 336 kg N/ha) replicated four times across tw o fields. An additional three fertilization treatments (N0 0 kg N/ha, N4 448 kg N/ha, N5 560 kg N/ha) were tested on SMS plots and replicated four times across two fields. Prior to planting field corn on March 5, 2018, all plots were strip tilled wi th a Orthman 1tRIPr row unit to create 12 rows per plot (Orthman, 2018) . The Pioneer 1870 YHR/BT field corn variety was planted with 76 cm row spacing , 17 cm seed spacing, and a desired plant population of 81250 plants/ha on March 8, 2018. Each irrigation treatment received two initial nitrogen granular applications at the beginning of the growing season followed by four liquid side dress applications l eading up to tasseling. S martirrigation Corn App Model D escription Initial i nput The Smartirrigation corn app generates a daily root zone water deficit using a soil water balance model. The Smartirrigation corn app is available to download through the a pp store for Apple and most Android mobile devices. Once downloaded, the corn app requires some initial input from the user (Table 2 1). This initial input is used in the model to calculate plant available water, which is translated into a root zone water deficit. The following steps describe initial user input for the corn app: 1. Tap or drag the pin on the map to the desired field location (Figure 2 1) 2. Name the field (Figure 2 2)

PAGE 32

32 3. Specify planting date (Figure 2 2) 4. Select weather data source. Users can choos e between land based weather station data from a land based Florida Automated Weather Network (FAWN) or Georgia Automated Environmental Weather Network (GAEMN) weather station or gridded weather data from the Dark Sky application and Forecasted Reference E vapotranspiration (FRET) (NWS, 2018) (Figure 2 2) 5. Choose primary so il texture for the field (Figure 2 3) 6. Modify available soil water holding capacity if it differs from default values associated with selected soil texture. (Figure 2 3) 7. Specify irrigation system type. Users can choose between high and low pressure overhead sprinklers (Figure 2 4). 8. Specify irrigation system efficiency. A default value is set to 85% (Figure 2 4) 9. Choose the desired irrigation application rate (in/event) (Figure 2 4) 10. Select a target root zone water deficit notification threshold. This threshold changes based on the estimated crop growth stage. A default threshold of 50% is used from emergence to V12 vegetative stage and then switches to 33% from tasseling to R3 reproductive stage. The user will receive a notification on their mobile device when the root zone water d eficit exceeds 40% for the 50% target and 20% for the 33% target (Figure 2 4). App functionality After the user has specified the necessary initial inputs, a field is created using the initial input information and is available to vi ew from the main menu. Users can create as many fields as they like (Figure 2 5). When the desired field is selected from the main menu, a field menu is

PAGE 33

33 displayed which contains a variety of information and functions. Information displayed on the selec ted field menu includes a ten day summary of the root zone water deficit, accumulated growing degree days, and crop growth stage (Figure 2 5). In addition, field information such as the sprinkler type, planting date, soil type, and irrigation rate are displaye d (Figure 2 5). Functions available from the field menu include options to edit the initial input for the selected field and the ability to add applied irrigation or observed rain for the current or past nine days (Figure 2 6). A summary of current conditi ons, hourly forecast for temperature and rain, and projected weekly forecast for temperature, relative humidity, rain and wind using weather data from the Dark Sky application is available to view by selecting the forecast icon in the top right corner of t he selected field menu (Figure 2 7). The Dark Sky application is described in detail below. A summary of the maintained daily soil water balance is available to download at any time by selecting the download icon in the top right corner of the selected fie ld menu (Figure 2 8). The report is sent as a CSV file to the users email address. The Smartirrigation corn app model parameters are described below. Growing degree day c alculation Based on user location specifications within Georgia or Florida, daily wea ther data are obtained from a land based Florida Automated Weather Network (FAWN) or Georgia Automated Environmental Weather Network (GAEMN) weather station or gridded Dark Sky weather data. Dark Sky is a web based and smartphone application that converts data from National Weather Service (NWS) Doppler radar stations into images where algorithms are able to predict rainfall based on the characteristics of those images (Markowitz, 2013) . The Dark Sky application also provides past daily weather data through a global application programming interface (API) (Dark Sky, 2018) . The Dark Sky API is supported by an aggregate of various

PAGE 34

34 weather sources such as ( NOAA ) Global precipitation forecasting system which is backed by (Dark Sky, 2018) . Daily weather data include : 1. altitude (m) 2. latitude (radians) 3. average temperature (°C) 4. max temperature (°C) 5. minimum temperature (°C) 6. average relative humidity (%) 7. total solar radiation (MJ/m 2 ) 8. average wind speed (m/s) 9. average dew point temperature (°C) Weather data are used to calc ulate daily heat units (DHUs) and reference evapotranspiration (ETo). Daily Heat Units (DHUs) are calculated using Eq. 2 1 with a base temperature (T base ) of 10 °C (Neild et al., 1987). Two temperature constraints are used in the DHU calculation for corn. Daily maximum temperatures (T max ) greater than or equal to 30 °C are set to 30 °C (Neild et al., 1987). Daily minimum temperatures below T base are set to T base (Neild et al., 1987). If daily mean temperatures are less than T base , DHU is equal to zero (N eild et al., 1987). (2 1)

PAGE 35

35 Growing degree days (GDDs) are the accumulated DHUs after planting. DHUs are calculated and cumulated until harvest. GDDs can be used to predict the phenological stage of a plant. Theoreti cal values of accumulated GDDs as a function of phenological growth stage were used to develop the relationships in Table 2 2 (Neild and Newman, 1990) . The GDDs determined in Tables 2 2 and 2 3 were calculated using daily maximum and minimum temperature. Ini tial and maximum K c values were determined using KanSched for humid moderate wind areas (K State Research & Extension Mobile Irrigation Lab, 2014) . Based on an analysis of 2015 to 2017 data from the Live Oak FAWN weather sta tion and observations taken at the NFREC SV during corn production, relationships between GDD and K c were developed (Table 2 3) (Zamora Re, 2019) . Table 2 4 is used in the corn app to select the K c for the crop ET calculation. Crop evapotranspiration calculation The amount of water lost to transpiration and evaporation, or evapotranspiration (ET), can be estimated using the crop coefficient (K c ) method with ET o . Thus, crop evapotranspiration (ET c ) can be estimated using the following equation: (2 2) where ET c and ET o are in units of mm/day. Reference evapotranspiration is determined using daily weather data from the selected FAWN or GAEMN station or gridded data from the Dark

PAGE 36

36 Sky weather a pplication and the Penman Monteith FAO 56 equation (Allen e t al., 1998) . When FAWN/GAEMN weather data is unavailable due to station or network failures, Forecast Reference Evapotranspiration (FRET) data from NOAA are used, and ETo (mm) is already calculated. Forecast Reference Evapotranspiration values are determ ined using National Weather Service forecasted temperature, relative humidity, wind, and cloud cover as well as the Penman Monteith reference evapotranspiration equations to estimate the depth of water in mm that would transpire from a short canopy referen ce grass (NWS, 2018) . Irrigation and rain input calculations Effect ive irrigation (I, mm) is calculated by multiplying the irrigation applied by the irrigation effectiveness factor. An irrigation effectiveness factor is set at a default of 85% but is user adjustable (Figure 2 3) (I rmak et al., 2011; Vellidis et al., 2016) . This factor accounts for sprinkler irrigation losses such as drift and droplet evaporation, plant interception, net canopy evaporation, and soil evaporation (Irmak et al., 2011) . An irrigation effectiveness factor of 85% corresponds to a low pressure sprinkler and an irrigation effectiveness factor of 75% corresponds to a high pressure sprinkler to account of for water droplet evaporation and drift (Vellidis et al., 2016) . If daily rainfall is greater than 4.0 mm, effective rainfall (R) is calculated by multiplying daily total rain by an effectiveness percentage of 90 %. Canopy interception, canopy evaporation, runoff, and rainfall measurement variability are accounted for in the 10% loss of total daily rainfall (Vellidis et al., 2016) . If daily rainfall is less than 4.0 mm, effective rain (R) equals zero. Daily rainfall events less than 4.0 mm are unlikely to have a noticeable effect on soil moisture (Vellidis et al., 2016) . Thus, water inputs to the system are calcu lated as I plus R in units of mm.

PAGE 37

37 Root depth calculation The first step in the soil water balance is to determine the root depth since it increases as plants grow. At planting, root depth or initial root depth (Rz initial ) is set to 8 cm. A maximum root de pth (Rz max ) was defined as 61 cm. Daily root depth increase (RDI) is calculated using Eq. 2 3 (Allen et al., 1998) . Eighty percent crop cover is theoretically achieved 43 days after the initial planting date (Allen et al., 1998) . Thus , the daily increase in root depth is the same until 61 cm is reached. Considering that 80% cover is achieved in 43 days, the relationshi p for daily root depth increase (RDI, cm) is: (2 3) (2 4) Each day root depth (RD i root depth (cm) is less than Rz max then, (2 5) Otherwise, RD i equals Rzmax. Any value for RD i greater than Rzmax, is assigned Rzmax. Plant available soil water calculation The plant available water (PAW) is equal to the available soil water holding capacity (Table 4) multiplied by root depth on the current day (Eq. 2 6). Available soil water holding

PAGE 38

38 capacity (AWHC) is defined as difference between field capacity and permanent wilting point (USDA NRCS, 2008) . (2 6) For each day after planting, when new soil depth (or a greater root depth) is added in the of root depth is de scribed in Eqs. 2 3 2 5. When RD is greater than or equal to Rz max , PAW is set to A WHC multiplied by Rz max . The user provides the soil texture as an input (Table 2 4). The soil texture is a ssociated with available soil water holding capacity (Table 2 4) . The user can input a specific available soil water holding capacity if it differs from the value assigned from Table 2 4 (Figure 2 4) (USDA NRCS, 2005) . Root zone water deficit calculation The soil water balance in this model is determined by calculating a root zone water deficit (RZWD). The initial root zone water deficit (mm) is set to zero (DAP, RZWD=0). For each timestep (or day), a root zone soil water deficit (RZWD) is determined by subtracting effective irrigation plus rain (I+R) from the actual evapotranspiration (ET c ). If the RZWD is less tha n zero, RZWD is set to zero. Each day, the root zone water deficit is determined by adding the 7 are in mm.

PAGE 39

39 (2 7) The Root Zone Water Deficit (RZWD, %) is determined using Eq. 2 8. (2 8) The app sends the user a notification when the 50% RZWD threshold is close to or has been met. A 50% water deficit is used to schedule irrigation becau se it corresponds to depletion of 50% of PAW which is a threshold commonly used to avoid plant stress and schedule irrigation for corn (USDA NRCS, 2005) . The app also sends users a notification when a 33% RZWD threshold is met during the reproductive growth stages. Crop water demand is highest during the reproductive growth stages so a sufficient amount of water should be applied during this time to ensure productive yields (Porter, 2 017) . Calculation steps for corn app model Step 1. Determine phenological stage based on GDD as described in Tables 2 and 3. Step 2. Determine Kc as described in Table 3. Step 3. Determine ET o for the day using selected weather data and Penman Monteith F AO 56 equation. Step 4. Calculate ET c for the day using Kc from Step 2 and ET o from Step 3.

PAGE 40

40 Step 5. Determine rooting depth for the day. Initial rooting depth or Rz initial is 3 inches. Maximum root depth (Rz max ) is defined at 24 inches. The first day is as signed the initial rooting depth. Days that follow are calculated using Eqs. 2 4 and 2 5 until Rz max is reached. Step 6. Calculate plant available soil water content (PAW) for the day using Eq. 2 6. Step 7. Calculate soil water deficit (RZWD) for the day u sing Eq. 8. If RZWD > PAW, RZWD = PAW. Step 8. Calculate root zone water deficit as a percent (RZWD %) for the day using Eq. 2 8. Step 9. T he user is notified when the RZWD ( % ) threshold approaches 50% for vegetative growth stages and 33% for reproductive stages. Soil Moisture Sensor Data C ollection SENTEK Drill and Drop MTS probes were installed within the center of the 4 th planting row in six corn app treatment plots at NFREC SV (Figure 2 5) (Sentek Pty Ltd., 2016) . Each probe consists of nine sensors that estimate volumetric water content in 10 cm increments beginning at 5 to 85 cm every thirty min utes. Volumetric s oil moisture readings from April 4 to August 16, 2018 were downloaded remotely from a website ( http://myfarm.highyieldag.com/home ). Daily estimated volumetric soil moi sture content from all six probes were averaged together and used for comparison against the corn app model root zone water deficit output. T he sum of averaged daily volumetric soil moisture readings from sensors located at 5 to 65 cm were divided by an effective root depth of 650 mm to estimate the percentage of water within the root zone (Eq. 2 9). A field capacity of 12.5% was determined using guidelines proposed by Zotarelli

PAGE 41

41 et al., (2010). A root zone water deficit using the ratio of the percentage of water within the ro ot zone determined from SMS data to field capacity is described in Eq. 2 10. (2 9) K (%) = Percentage of water within root zone X = Sum of averaged volumetric water content from 5 cm to 65 cm (2 10) Data A nalysis The Nash Sutcliffe efficiency (NSE) test Eq. 2 11 was used to compare the corn app oisture deficit predictions in comparison to SMS measurements (Nash and Sutcliffe, 1970) . The NSE test determines the relative magnitude of the residual variance compared to measured d ata variance (Nash and Sutcliffe, 1970) . In this case, the soil moisture sensors are the observed data and the corn app is the predicted data. Values for the Nash Sutcliffe efficiency test range from negative infinity to one and an NSE value of 1 indicates the model perfectly matches the observed data (Nash and Sutcliffe, 1970) . An NSE value of 0 indicates the corn app model predictions are as accurate as the mean of the observed data and a negative NSE value implies the observed mean of the soil moisture sensor data is a better predictor than the corn app model. (2 11) Figure 2 10 shows the uncalibrated app overestimates daily root zone water deficits in comparison to soil moisture data. This can be attributed to overest imating evapotranspiration or

PAGE 42

42 underestimating rainfall and irrigation contributions. In addition, Figure 2 10 suggests that the uncalibrated corn app model may over predict evapotranspiration during early crop development. The Kc initial period has been s hown to extend from 0 DAP to 20 DAP (Allen et al., 1998; L. Hatfield and Dold, 2018) . Adjusting the K c curve t o extend the initial K c period corresponded better with infield observations and model predictions for where the corn crop developed from emergence and initial growth stages and transitioned into rapid vegetative development (Figure 2 11).Therefore, the pe riod for K c initial was extended ten days from ending after 8 DAP to ending after 19 DAP. Calibration The resultin g corn app model after extending the initial K c period was then calibrated by adjusting three parameters to increase rainfall and irrigatio n contributions to soil moisture conditions. The corn app model parameters selected for adjustment were the threshold for counting rainfall as well as irrigation and rainfall effectiveness factors . These three parameters were selected as they contained unc ertainty in accurately determining the portion of rainfall and irrigation contributing to soil moisture conditions. This uncertainty is due to the corn app ca nopy evaporation, and droplet drift and evaporation accounted for in the irrigation and rainfall effectiveness factors . A total of forty seven possible combinations of adjusted parameters (irrigation effectiveness, rainfall effectiveness, and threshold f or counting rainfall) were assessed in this study. Irrigation effectiveness was increased 5% at a time, from 85 to 100%, rainfall effectiveness parameter was increased 5% at a time, from 90 to 100%, and the rainfall threshold

PAGE 43

43 was decreased by approximately 1 mm at a time, from 4 to 0 mm. Each combination of adjusted parameters was simulated using the corn app model spreadsheet with 2018 growing season data. Three months: April, June, and August, (71 days total) during the 2018 growing season were used for c alibration of the corn app model to better match measured soil moisture sensor data. Since the soil moisture sensors were installed in early April 2018, measured SMS data from April 4 April 31, 2018 was used for comparison against the corn app model root zone moisture deficit outputs. In addition, measured SMS data from August 1 14 2018, was used because the soil moisture sensors were removed on August 14, 2018. The root zone water deficit output from every adjusted model was compared to measured soil m oisture data using the Nash Sutcliffe efficiency test until the optimum combination of parameters was found (Table 2 6) (Nash and Sutcliffe, 1970) . In addition, a root mean square erro r (RMSE) test was used to determine the adjusted model that minimized the error between predicted root zone water deficits and measured SMS data (Table 2 6) (Equation 2 12). (2 12) The calibrated app model that produced the highest NSE and lowes t RMSE values when compared with measured SMS data was validated for the months of May and July (61 days total) during the 2018 growing season. In addition, the calibrated and validated corn app model was simulated through the entire 2018 growing season to determine if water savings were achieved in comparison to the uncalibrated corn app model. Using the same weather data and methods for irrigation scheduling as the corn app treatment for the 2018 growing season, the calibrated model was simulated in a spr eadsheet for the 2018 growing season. Irrigation was not scheduled for

PAGE 44

44 weekends and major holidays and the first three initial irrigation applications for watering in fertilizer and pesticide were included. Results Model Adjustment and C alibration R esults A Nash Sutcliffe efficiency and RMSE of 1.69 and 0.14, respectively, were determined for the comparison between the uncalibrated corn app model and SMS data, which indicated the mean of the observed SMS data was a better predictor of soil moisture condi tions in comparison to measured SMS data than the corn app model. Extending the K c initial period produced a model that achieved an NSE and RMSE value of 0.67 and 0.14, respectively, which are improved values in comparison to uncalibrated app model when c ompared with measured SMS data for the calibration data set. The results of comparisons between root zone water deficits predicted by each calibrated corn app model and measured SMS data are summarized in Table 2 6. Thirty three rainfall and nine irrigati on events totaling 265 and 91 mm, respectively, occurred on the corn app treatment plots at NFREC SV during the three months chosen for calibration. The two models that resulted in the highest NSE of 0.38 and lowest RMSE value of 0.09 when compared with me asured SMS data shared the adjusted parameters of irrigation effectiveness set at 95% and threshold for counting rainfall set at 0 mm but differed in that one model had rainfall effectiveness set at 90% and the other at 100% (Table 2 6) . Both of the highes t performing calibrated models were selected for validation because they minimized the error between daily root zone water deficits predicted by the app and measured SMS data. An NSE value of 0.38 indicates the calibrated

PAGE 45

45 models predict daily root zone wat er deficits better than the average of measured SMS data would when compared with measured SMS data (Figure 2 12). Validation The months of May and July (61 days total) during the 2018 growing season were used for validation of the calibrated corn app m odels developed in this study. Thirty four rainfall and t welve irrigation events totaling 342 mm and 122 mm, respectively, occurred on the corn app plots at the NFREC SV for the months of May and July during the 2018 growing season. When compared with meas ured SMS data, the selected calibrated corn app model with the rainfall effectiveness set at 90% produced a higher NSE value and lower RMSE than the calibrated model with rainfall effectiveness set at 100% (Table 2 6). The calibrated corn app model with ra infall effectiveness set at 90% produced a NSE value of 0.72 and RMSE value of 0.06 when compared with measured SMS data (Figure 2 13). Values for the Nash Sutcliffe efficiency test range from negative infinity to one and a Nash Sutcliffe efficiency of 0.6 5 or greater indicates an acceptable model (Ritter and Muñoz Carpena, 2013) . The uncalibrated corn app model applied 283 mm of irrigation for the 2018 growing season. Both c alibrated models applied 40% less water than the uncalibrated corn app model with approximately 171 mm total when simulated through the 2018 growing season using the same methods for irrigation scheduling (Table 2 6). Discussion Effect of Corn App Model P a rameter A djustments Overall, this study found that increasing the rainfall and irrigation effectiveness percentage and decreasing the threshold for counting rainfall resulted in a calibrated model that

PAGE 46

46 more closely matched measured SMS data in comparis on to the uncalibrated corn app model. These results suggest that irrigation and rainfall may have a more appreciable effect on soil moisture conditions than previously assumed by the uncalibrated corn app model. The soils at NFREC SV are primarily sand , w hich generally have high infiltration rates with little to no runoff. Runoff was one loss that was accounted for in the rainfall effectiveness percent. The uncalibrated corn app model assumed that rainfall less than 4 mm did not contribute to soil moisture conditions. However, for every iteration of calibration where the threshold for counting rainfall was set to 0 mm, the resulting model produced high NSE and low RMSE values relative to the other iterations where it was set higher than 0 mm, demonstrating that rainfall events as small as 1 mm may contribute to soil moisture conditions. Evapotranspiration losses are generally lower on days when rainfall occurs. This is due to cloud cover, lower average temperatures, and higher relative humidity. Cloud cover and lowered temperatures reduce evaporation from the soil and higher humidity re duces transpiration from plants. Reduced evapotranspiration on days when rainfall occ urs is another factor that may contribute to improved models when the rainfall threshold is set 0 mm and rainfall effectiveness is set to 100%. One loss that was accounted for in all three parameters was canopy interception. Rainfall and irrigation are partitioned in the corn canopy into throughfall and stemflow (Marco et al., 2015) . Throughfall is the portion of water that makes it directly to the ground, while stemflow is the portion of water captured by the corn canopy and transferred down to the soil via the stem (Marco et al., 2015) . A study by Marco et al., (2015) m easured throughfall and stemflow with inter r ow buckets and stemflow collectors and estimated net evaporation loss from the corn canopy to be less than 10%. Considering that net canopy evaporation may be less than 10%, the

PAGE 47

47 irrigation and rainfall effective ness and threshold for counting rainfall parameters may have been over accounting for this loss. Other studies have considered an irrigation effectiveness factor of 85% for low pressure sprinkler and an irrigation effectiveness factor of 75% corresponds for high pressure sprinkler to account of for water droplet evaporation and drift (Vellidis et al., 2016) . The plots at NFREC SV were irrigated with a Valley two span linear end feed 8000 system equipped with low pressure sprinkler nozzles and a variable rate irrigation package. Increasing the irrigation effectiveness from 85 to 9 5% generally produced calibrated models that corresponded better with measured SMS data, suggesting that drift and droplet evaporation may have not been as substantial of a loss as previously assumed. When compared with measured SMS data, the calibrated mo dels where irrigation effectiveness was set at 100% and the threshold for counting rainfall was set at 0 mm produced lower NSE values compared to the calibrated models with irrigation effectiveness set at 95% and the threshold for counting rainfall at 0 mm . The lowered NSE values from increasing the irrigation effectiveness to 100% suggest that irrigation application losses such as drift and droplet evaporation and net canopy evaporation may be accounted for when setting the irrigation effectiveness to 95%. Both models chosen for validation produced an NSE value of greater than 0.65 when compared with measured SMS data (Table 2 6). The high NSE values produced from the validation results suggest that the calibrated models outputs correspond closely to measur ed SMS data. However, approximately 108 mm more of cumulative rainfall and irrigation occurred during the period chosen for validation in comparison to the calibration period. Improved NSE and RMSE values for the validation data set s uggest the app may per form better under wetter conditions.

PAGE 48

48 On average, the SMS treatments applied 141 mm of water for the 2018 growing season. When simulated through the 2018 growing season, the calibrated corn app models chosen for validation applied approximately 171 mm of w ater, which was 40% less water applied than the 283 mm total applied for the corn app treatment for the 2018 growing season. On average, the SMS treatments applied 141 mm of water for the 2018 growing season . When simulated through the 2018 growing season, the calibrated corn app model applied a similar am ount and frequency of irrigation water with 171 mm total compared to the SMS irrigation treatment for the 2018 growing season. There were no significant differences in marketable grain yield between the co rn app and SMS treatments for the 224 kg/ha and 336 kg/ha nitrogen regimes , suggesting that the calibrated corn app model could achieve similar yields by applying less water (Figure 3 9 ). Corn App L imitations veness in predicting daily soil moisture conditions that correspond to measured SMS data is that it calculates daily root zone water Utilizing the previous days soil wa ter balance causes a lag in the apps predictions for the current day. So why The corn app will not register new rainfall and irrigation events or daily evapotranspiration until the following day, whereas soil moisture sensors are able to record fluctuations in soil moisture conditions in real time. This is evident from the staggered peaks and valleys in root zone soil moisture deficits when comparing the 11) . In addition, the app tends to over predict root zone soil moisture deficits on the days when large rainfall events occur (Figure 2 11). predictions of root zone soil moisture deficits on days when large rainfall events occur could be due to the fact the SMS register the rainfall event in real time, whereas the app does not register

PAGE 49

49 the event until the following day. Soil moisture sensors show a more gradual shift in soil moisture conditions with peaks and valleys frequently occurring the day prior to th corresponding peaks and valleys (Figure 2 11). Peaks and valleys in Figure 2 11 demonstrate high and low root zone water deficits, respectively. The Corn App is also limited in that it estimates daily evapotranspiration, root depth, and utilizes t heoretical crop coefficients that have not been locally adapted. While the FAO 56 Penman Monteith equation is a widely accepted method for calculating reference evapotranspiration, it is inherently limited in that it provides an estimation of evapotranspir ation that may not completely reflect actual conditions. Furthermore, the corn app does not use locally adapted crop coefficient for field corn. Locally adapted crop coefficients are often needed to provide an accurate estimation of crop evapotranspiration . Extending the K c initial period resulted in a higher NSE and lower RMSE value than the uncalibrated corn app model when compared with measured SMS data. The Corn App also estimates effective root depth which is used to calculate plant available water. Th e effective root depth is the depth at which roots extract water from the soil. While corn root depth can extend beyond 60 cm, the maximum effective root depth used in the Corn App is 60 cm. This depth corresponds to effective corn root depths reported in other studies with similar site conditions (Evans et al., 1996; USDA NRCS, 2006) . In addition, the Corn App estimates a root zone water deficit for an entire field where there is often variability in soil type and topography. One of the most critical limitations to the corn app is that it uses weather information from a FAWN or GAEMN station or gridded dark sky weather data to obtain rainfall dat a. Rainfall is particularly variable in the southeastern U.S. and users located far away from a FAWN or GAEMN station may receive different amounts of rainfall than what was recorded

PAGE 50

50 (Zhang et al., 2018) . Ideally, farmers who use the app will have a rain gauge and record rainfall depths to be input into the app. Soil Moisture Sensor L imitations While there were twenty four corn app treatment plots total, only six soil moisture predictions (Figure 2 5). In addition, the sensors were only inserted in blocks 2, 3 and 4 further reducing the sample size of measured SMS data to draw from. One limitation to using measured soil moisture sensor data as values to calibrate the corn app model is that they estimate volumetric water content only in the area where they are inser ted. This reduces the ability of the sensors to provide complete range of measurements over larger areas. Soil moisture measured by sensors may not reflect soil moisture conditions across the whole field. The soil moisture sensors were also not calibrated for the 2018 growing season reducing their accuracy in providing volumetric moisture content. Soil moisture sensors also require technical expertise to implement and use correctly. Conclusion In this study, a smartphone application for ET based irrigat ion scheduling in field corn was developed and calibrated to improve its accuracy in predicting root zone water deficits . The corn app model was calibrated and validated by comparing daily root zone water deficit values from measured SMS data within the sa me plots for the 2018 growing season. The u ncalibrated corn app mod el produced a NSE value of 1.45 , indicating the corn app model outputs corresponded poorly to measured SMS data. In comparison to the uncalibrated corn app model, a djusting the K c curve to reflect measured and predicted growing conditions produced a model

PAGE 51

51 with higher and lower NSE and RMSE values of 0. 67 and 0.14, respectively, when compared with measured SMS data. I ncreasing the amounts of irrigation and rainfall accounted for in the cor n app model by setting th e irrigation effectiveness to 95%, rainfall effectiveness to 9 0 % , and threshold for counting rainfall to 0 mm, produced a calibrated and validated model that achieved NSE values of 0.38 and 0.73, respectively, when compared with me asured SMS data. Only 61 days of the 2018 growing season were used for validation of the calibrated corn app model. The calibrated corn app model should be validated in different fields across multiple years for entire growing seasons to improve its accur acy. In addition, locally adapted crop coefficients should be used in the corn app model to improve its accuracy in predicting crop evapotranspiration.

PAGE 52

52 Figure 2 1 . User selection of field location

PAGE 53

53 Figur e 2 2 . User input of field name, planting date and choice of weather data source

PAGE 54

54 Figure 2 3 . User selection of available soil water holding capacity

PAGE 55

55 Figure 2 4 . User input of irrigation type, irrigation system efficiency, irrigation application rate, and root zone water notification threshold

PAGE 56

56 Figure 2 5 . Field selection from corn app main menu

PAGE 57

57 Figure 2 6 . Field menu that displays sprinkler type, planting date, soil type, and irrigation rate information. A summary of a 10 day water balance, growth stage and accumulated growing degree days is also displayed

PAGE 58

58 Figure 2 7 . F orecast menu provided in the corn app

PAGE 59

59 Figure 2 8 . User option to send report of maintained soil water balance

PAGE 60

60 Figure 2 9 . Aerial f ield layout and location of SMS sensors beneath app treatment plots at NFREC SV. The gold stars represent the soil moisture sensors inserted within the corn app plots. (I1 = Calendar based, I2 = Corn App, I3 = SMS, and I5 = Non irrigated) Photo courtesy of author.

PAGE 61

61 Figure 2 10 . Resid ual plots of SMS data vs. corn app model. A) Residuals plot of observed root zone water deficit from soil moisture sensor data vs. predicted root zone water deficit from uncalibrated corn app model. B) Residuals plot of observed root zone water deficit fro m soil moisture sensor data vs. predicted root zone water deficit from calibrated corn app model

PAGE 62

62 Figure 2 11 . The photo on the left was taken on March 21, 2018 (13 DAP) and shows the corn crop e merging and developing the first leaf collar. The photo on the right was taken April 4, 2018 (27 DAP) and shows the corn crop with three leaf collars with the fourth leaf beginning to emerge from the whorl. Photos courtesy of author .

PAGE 63

63 Figure 2 12 . C omparisons of the corn app and SMS treatments for the calibration period. A) Comparison of predicted RZWD (%) from uncalibrated corn app model and measured SMS data for the months of April, June, and August. B) Comparison of predicted RZWD (%) by calibrate d corn app model and measured SMS data for the mo n ths of April, June, and August

PAGE 64

64 Figure 2 13 . Comparison of the calibrated corn app and SMS for the validation period. A) Residual plot of predicted RZWD (%) from validated c orn app model and measured SMS data for the months of May and July. B) Comparison of predicted RZWD (%) from validated corn app model and measured SMS data for the months of May and July.

PAGE 65

65 Figure 2 14 . Comparison of irri gation applications by SMS (I3 N and I3 S) irrigation treatments and calibrated corn app model simulated for the 2018 growing season

PAGE 66

66 Table 2 1 . User i nputs provided for each location where the corn app will simulate so il water conditions Input Description Location User selects location of farm Name Provide a specific name for the field Planting date Date crops are planted and after which heat units are accumulated as a measure of plant growth Select weather data so urce User will have the option of using weather station data from FAWN/GAEMN or using Dark Sky/FRET data Soil texture Soil texture will include sand, clay, silt. Each soil texture is associated with a particular available water holding capacity (cm/cm) ( Additional textures found in Table 4.) Available water holding capacity The available water holding capacity can be adjusted if it differs from the default value associated with the chosen soil texture Planting date Date crops are planted and after which heat units are accumulated as a measure of plant growth Irrigation system efficiency User can select efficiency of their specific irrigation system; a default will be set to 85% Irrigation rate (in/event) User can input the amount they irrigate per ev ent Root zone water deficit target A default will be set to 50% from emergence to V12 and reduced to 33% from tasseling to R3. Both of these values are user adjustable.

PAGE 67

67 Table 2 2 . Phenological growth stage as a funct ion of accumulated GDD Accumulated GDD (GDU, °C) Phenological Growth Stage Emergence V1 V2 V3 V5 V6 V8 V9 V12 Tasseling R1 R2 R3 R4 R5 R6 Maturity Harvest

PAGE 68

68 Table 2 3 . Crop coefficient (Kc) as a function of accumulated GDD Accumulated GDD (GDU, °C) Kc 0.25 ) & 1.05 0.55 0

PAGE 69

69 Table 2 4 . Soil textures and corresponding soil water holding capacity represented as a fraction by soil volume Soil Texture Soil Water Holding Capacity (mm/mm) Sand 0.06 Fine Sand 0.07 Loamy Sand 0.13 Sandy Loam 0.1 Loam 0.18 Silt Loam 0.21 Clay Loam 0.2

PAGE 70

70 Table 2 5 . Corn app model calibration results Irrigation effectiveness (%) Rainfall effectiveness (%) Rainfall threshold (mm) Nash Sutcliffe Efficiency Value RMSE 85 95 3.8 0.58 0.14 85 100 3.8 0.49 0.14 85 90 2.5 0.21 0.12 85 95 2.5 0.12 0.12 85 100 2.5 0.04 0.11 85 90 1.3 0.01 0.11 85 95 1.3 0.09 0.11 85 100 1.3 0.16 0.10 85 90 0 0.16 0.10 85 95 0 0.23 0.10 85 100 0 0.29 0.09 90 90 3.8 0.50 0.14 90 95 3.8 0.42 0.13 90 100 3.8 0.35 0.13 90 90 2.5 0.07 0.12 90 95 2.5 0.00 0.11 90 100 2.5 0.07 0.11 90 90 1.3 0.13 0.10 90 95 1.3 0.19 0.10 90 100 1.3 0.25 0.10 90 90 0 0.26 0.10 90 95 0 0.31 0.09 90 100 0 0.35 0.09 95 90 0.15 0.36 0.13 95 95 0.15 0.30 0.13

PAGE 71

71 Table 2 5. Continued Irrigation effectiveness (%) Rainfall effectiveness (%) Rainfall threshold (mm) Nash Sutcliffe Efficiency Value RMSE 95 100 0.1 5 0.25 0.12 95 90 0.1 0.03 0.11 95 95 0.1 0.09 0.11 95 100 0.1 0.14 0.10 95 90 0.05 0.21 0.10 95 95 0.05 0.26 0.10 95 100 0.05 0.30 0.09 95 90 0 0.38 0.09 95 95 0 0.36 0.09 95 100 0 0.38 0.09 100 90 0.15 0.27 0.13 100 95 0.15 0.22 0.12 100 1 00 0.15 0.18 0.12 100 90 0.1 0.10 0.11 100 95 0.1 0.14 0.10 100 100 0.1 0.18 0.10 100 90 0.05 0.26 0.10 100 95 0.05 0.29 0.09 100 100 0.05 0.31 0.09 100 90 0 0.35 0.09 100 95 0 0.36 0.09 100 100 0 0.37 0.09

PAGE 72

72 Table 2 6 . Calibrated corn app model validation results. Water applied is the result of simulating the calibrated corn app model through the 2018 growing season . Calibrated app #1 has parameter factors of an irrigation effectiveness set to 95%, rainfall effectiv eness set to 90%, and threshold for counting rainfall set to 0.00 mm. Calibrated app #2 has parameter factors of an irrigation effectiveness set to 95%, rainfall effectiveness set to 100%, and threshold for counting rainfall set to 0.00 mm. Water Applied (mm) Water Savings (%) NSE RMSE Uncalibrated App 283 1.69 Calibrated App #1 171 40 % 0.72 0.06 Calibrated App #2 171 40 % 0.68 0.07

PAGE 73

73 CHAPTER 3 EVALUATION OF IRRIGATION APPLIED, YIELD, AND WATER USE EFFICIENCIES FOR DIFFERENT IRRIGATI ON SCHEDULING METHODS ON CORN Introduction Agriculture accounts for 80% of the United States consumptive water use (USDA, 2018a) . Irrigation for agriculture is the application of water to soil in order to supply sufficient moisture for plant growth, provide crop risk reduction in case of drought, soil temperature cooling, dilution of salts in the soil, and softening tillage pans (Israelsen, 1950) . In 2012, corn accounted for approximately 25% of total harvested U.S. irrigated acreage (USDA, 2018a) While corn is typically grown in the Midwest, it is also grown in the Southeast for grain and silage production (Wright, 2004) . While the climate in the eastern United States is mostly humid, irrigation is often needed for crops (Jamison and Beale, 1958) . The widespread availability, easy access, and minimal system requirements to extract groundwater has led to an increase in groundwater used for irrigation (Scanlon et al., 2012) . Groundwater withdrawals from the Floridan Aquifer have increased from 2,384,550 m 3 /day in 1950 to 15,215,700 m 3 /day in 2000 (Berndt, 2014) . Agricultural withdrawals account for half of that increase and irrigated acreage in the Suwannee River Water Management District (SRWMD) is expected to increase by 40% from 2015 to 2040 (Berndt, 2014; The Balmoral Group, 2018) . Field corn and peanuts are typically grown in rotation with each o ther in Northern Florida and Southern Georgia. Field corn and peanuts accounted for 54% of sprinkler irrigated crops in the Suwannee River Management District (SRWMD) in 2015 (Marella et al., 2 016) .

PAGE 74

74 The Suwannee groundwater basin is host to numerous freshwater springs that are present along the lower Suwannee and Santa Fe Rivers (Bellino et al., 2018) . A distinct hydraulic connection exists between the Floridan aquifer and Suwannee groundwater basin (Grubbs and Crandall, 2007) . Increased freshwater withdrawals in the Suwannee groundwater basin has affected spring and rive r flows to a point that they are not meeting established minimum flows or temporary flow constraints (SRWMD, 2010) . In addition, climate projections have indicated increased temperatures and seasonal droughts may increase freshwater demand, particularly for agriculture where droughts could likely occur during the growing season (Bellino et al., 2018) . To reduce further alterations in natural spring flows and mitigate the potential impacts of climate change, best management practices (BMPs) are being developed and assessed in the Suwannee River Basin (SRB) to reduce agricultural water usage and nutrient loading in the Floridan Aquifer. Reduced pumping in the Floridan Aquifer can be a ccomplished through efficient water use strategies such as irrigation scheduling. Previous research has shown evapotranspiration and soil moisture sensor (SMS) based irrigation scheduling provides water savings compared to traditional techniques. A study c onducted in Live Oak, FL achieved 42 and 39% water savings for field corn compared to a traditional calendar based irrigation scheduling method by utilizing a soil water balance approach that calculated daily ET c to estimate soil moisture and schedule irri gation in 2015 and 2016, respectively (Zamora Re and Dukes, 2016) . A study conducted a t the University of Nebraska Lincoln achieved water savings of 34% and 32% with little to no reduction in corn yield using a soil moisture sensor based irrigation regime in comparison to traditional farmer irrigation management in 2005 and 2006, respective ly (Irmak et al., 2012) .

PAGE 75

75 Objectives and Hypotheses The objectives in this study were to compare applicat ion depths, yields, and water use efficiencies for four irrigation scheduling methods (app, SMS, calendar based, and no irrigation) across three nitrogen fertilization regimes (N1 112 kg/ha, N2 224 kg/ha, and N3 (336 kg/ha) . The same objectives were appl ied in a separate study with three irrigation scheduling methods (app, SSA, checkbook) across three nitrogen fertilization regimes (Low 277 kg/ha, Traditional 333 kg/ha, and High 336 kg/ha) . It is hypothesized that the app, SMS, and non irrigated treatment s will have greater water use efficiencies than the calendar based treatment. It is also hypothesized that the app and soil water potential sensors (SSA) will have greater water use efficiencies than the checkbook method. In addition, each irrigation tre atment at NFREC SV (app, SMS, calendar based, non irrigated) was compared using measured soil moisture sensor data. It is hypothesized that the app and SMS treatments will have lower average soil moisture in comparison to the calendar based treatment throu ghout the growing season. Materials and Methods Experimental S ite and Agronomic P ractices This study was conducted in Live Oak, FL, USA, at the of Food and Agricultural Sciences (UF IFAS) Suwanee Valley North Florida Re search and Extension Center (NFREC SV). The soils at this site were primarily sand (95.9% sand, 2.3% silt and 1.8% clay) and consist ed of the following soil types : Blanton Foxworth Alpin complex (48.7%), Chepley Foxworth Albany (31.6%) and Hurricane, Alban y and Chipley soils (19.6%)

PAGE 76

76 (USDA, 2018). The experimental design for the NFREC field site was a randomized complete block organized in a split plot (Figure 3 1). This design includes four replicates or blocks. Each plot is approximately 12 m long x 6 m wi de separated by 6 m alleys. An alley of 12 m is included between the blocks (Figure 3 1). There were four irrigation (app, SMS, calendar based, and no irrigation) and three fertilization treatments (N1 112 kg N/ha; N2 224 kg N/ha; N3 336 kg N/ha) rep licated four times across two fields. Prior to planting field corn on March 5, 2018, the plots in both fields were strip tilled with a Orthman 1tRIPr row unit creating 30 cm strips (Orthman, 2018) . Pioneer 1870 YHR/BT field corn variety was planted with 76 cm row spacing, 17 cm seed spacing, and a desired pla nt population of 8125 0 plants/ha on March 8, 2018. Each irrigation treatment received two initi al nitrogen granular applications at the beginning of the growing season followed by four liquid side dress applications leading up to tasseling. An initial nitrogen granular application of 31, 39, and 50 kg/ha were applied on April 5, 2018 to the N1, N2, and N3 plots, respectively. A second nitrogen granular application of 31 and 50 kg/ha were applied on April 18, 2018 to the N2 and N3 treatments, respectively . For the N1 treatment, one nitrogen liquid side dress applications of 39 kg/ha was applied on A pril 30, 2018. For the N2 treatment, four nitrogen liquid side dress applications of 31 kg/ha were applied on April 30, 2018, May 4. 2018, May 10, 2018, and May 15, 2018. For the N3 treatment, four nitrogen liquid side dress applications of 50 kg/ha were applied on April 30, 2018, May 4. 2018, May 10, 2018, and May 15, 2018. Growth stages were determined in both fields at NFREC SV biweekly when 50% or more of the field exhibited a particular growth stage (Figure 3 2). Tasseling was observed on May 20 21, 2018. (UGA) Stripling Irrigation Research Park (SIRP). The soil at SIRP is classified as Lucy loamy

PAGE 77

77 sand with 0 to 5 percent slopes. The experimental design at SIRP is a randomized complete block organized in a split plot (Figure 3 3). This design contains three blocks or replicates. Field corn was grown on one field containing twenty seven plots with three ir rigation (app, checkbook, and SSA ) and three fertilization tre atments, which are a combination of application rate and method (low 277 kg/ha, traditional 333 kg/ha, and high 336 kg/ha) replicated three times (Table 3 2). Each plot is approximately 15 m long and consists of eight rows. Eight row buffers separate each plot within a row and 18 m alleys are located between plots. While the traditional and high fertilizer treatments have similar application rates, they differed by application techniques. The traditional fertilizer treatment utilized a preplant granula r of 56 kg/ha on March 7, 2018 , pop up application of 50 kg/ha at planting on March 29, 2018 , and a granular application of 224 kg/ha on May 7, 2018 . The high fertilizer treatment consisted of a preplant granula r application of 56 kg/ha on March 7,2018, po p up application of 50 kg/ha on March 29,2018 , and four in season fertigation applications of 57 kg/ha . A preplant granular of 56 kg/ha , pop up of 50 kg/ha at planting , and four in season 35 kg/ha fertigation applications were conducted for the low fertil ization treatment. The four in season applications for the low and high fertilization treatments were applied on May 9, May 16, May 23, and May 30 during the 2018 growing season. Field properties for both testing sites are described in Table 3 3. The corn field was strip tilled using a Kelley strip till rig with 36 cm wide strips and 91 cm spacing from center to center (KMC, 2018) . The same Pioneer 1970 hybrid variety used at NFREC SV was pla nted at SIRP on March 29, 2018 with a desired seed population of 80,450 plants/ha. Irrigation T reatments at NFREC SV Four irrigation treatments (app, SMS, calendar based, and no irrigation) were as sessed at NFREC SV in this experiment (Table 3 4). Irrigation scheduling for each treatment began 30

PAGE 78

78 days after planting (DAP). Prior to 30 DAP, every treatment received three 10 mm irrigation applications to water in fertilizer and herbicide. Rainfall was monitored using the Live Oak FAWN weather station. A default irrigation application of 10 mm was applied to each treatment that required irrigation. In addition to written records maintained by the staff at NFREC SV, irrigation application depths for each treatment were recorded in a spreadsheet. Corn app treatment Irrigation for the corn app was scheduled using a daily estimated root zone water deficit. This is calculated daily by subtracting effective irrigation plus rain (I+R) from the actual evapotra nspiration (ET c ). Crop evapotranspiration was determined by multiplying ET o by a crop coefficient based on growing degree days (GDD). Reference evapotranspiration was calculated by utilizing the ASCE Penman Monteith FAO 56 equation with daily weather data from the Live Oak FAWN weather station. At the beginning of each week, an irrigation schedule was produced based on forecasted ET o for the next five days and projected rainfall. Irrigation events of 10 mm were scheduled for the days when the root zone wate r deficit approached the 50% theshold. The schedule was adjusted depending on large forecasted rainfall events or a sufficient rainfall event lowered the deficit below the threshold. The corn app model was run in a spreadsheet for the 2018 growing season w hile the mobile app was under development. In addition to irrigation and reference evapotranspiration, rainfall and temperature data from the Live Oak FAWN weather station were input into the spreadsheet model daily. Soil moisture sensor treatment SENTE K Drill and Drop MTS probes were installed within the center of the 4 th planting row in each I3 (SMS) plot (48 total) (Sentek Pty Ltd., 2016) . Each probe consists of nine sensors

PAGE 79

79 that estimate volumetric water content in 10 cm increments beginning at 5 cm to 85 cm every thirty minutes. Using the SENTEK probes, moisture content of the soil was monitored and irrigation was determined using the maximum allowable depletion (MAD) and field capacity (FC) points to refill the soil profile with irrigation. Soil moisture readings were downloaded remotely from a website ( http://myfarm.highyieldag.com/home ). A field capacity of 12.5 % was determined for the NFREC SV field site according to guidelines proposed by Zotarelli et al., (2013). The I3 (SMS) irrigation treatment was managed independently in the North and South fields in this experiment. For each probe, the sum of volumetric water content measured by each sensor from 5 cm to 65 cm was divided by the total effective root depth of 650 mm to determine the percentage of water within the root zone. An effective root depth of 650 mm was chosen because it corresponded to the lowest depth where evapotranspiration fluctuations were occurring in the majority of the sensors. The lowest quartile , of probes underlying SMS plots, of percentages of soil moisture content within the root zone determined for each field was used to determine whether irrigation was needed. When the lowest quartile of probes showed the percentage of water within the effective root zone was approaching or had met the MAD of 6.25 %, an irrigation application of 10 cm was scheduled for the following day. Calendar based treatment 31 days after planting (DAP), a target amount of 30 mm/week was established and was made up of rain or irrigation with the exception that rain events had to be 6 mm or larger. For 40 59 DAP a 40 mm/week target was established and irrigation applications were skipped if 13 19 mm/week

PAGE 80

80 of rainfall occurred. Two irrigation applications we re skipped if >19 mm of rain occurred. For 60 105 DAP a 50 60 mm/week irrigation target was used unless 13 25 mm of rain occurred the day prior to a scheduled irrigation. Two irrigations were skipped if >25 mm of rain occurred the day prior to a scheduled irrigation. Finally, around 105 DAP at full dent stage, weekly irrigation targets were reduced to 40 mm/week for one week and 20 mm/week for two weeks until irrigation was terminated when the corn was observed to be physiologically mature and the kernels h ad formed a black layer at 134 DAP. Non irrigated treatment Besides the initial three irrigation applications to water in fertilizer and herbicide, no irrigation was applied to this treatment. Irrigation T reatments at SIRP Three irrigation treatments ( app, checkbook, and SSA) were assessed at SIRP in this experiment (Table 3 5). Rainfall was monitored using the Camila GAEMN weather station. A default irrigation application of 12.7 mm was applied to each treatment that required irrigation. The app treatm ent conducted at SIRP was the same as the treatment described for NFREC SV above. However, weather data for the app treatment conducted at SIRP was obtained from the Camilla GAEMN weather station. Checkbook method The checkbook method uses a theoretical moisture balance and keeps a record of water inputs and outputs to maintain a balanced quantity of water for crop growth. Information needed for the checkbook method include soil type, soil available water holding capacity, daily water

PAGE 81

81 use of corn, and a rain gauge in close proximity to the growing field. The checkbook treatment differs from the corn app treatment described in chapter 2 in that is uses historical averages of evapotranspiration rather than a daily estimation, which may result in under or ov er irrigation during drier or wetter years. Weekly crop water use is a function of days after planting (Table 3 6) (Migliaccio et al., 2016) . The following steps describe scheduling irrigation for corn at 70 DAP at SIRP using the checkbook method: Step 1. Determine the plant available water by multiplying the available water holding capacity by the effective root depth. Assuming the effective root depth is 61 c m and available water holding capacity of 0.08 cm/cm for the Lucy loam sand at SIRP, pla nt available water is equal to 49 mm. Step 2. From Table 3 6, corn crop water use is 8.1 mm per day at 70 DAP. Step 3. Using a 50% MAD, 25 mm of irrigation is needed to replace the water lost if no precipitation occurs. Step 4. Assuming an irrigation effi ciency of 85% for the linear high pressure sprinkler system used at SIRP, the total irrigation required is . Step 5. The frequency of irrigation was determined by dividing the amount of replacement water by the daily water use. For example: Step 6. As a result, 25 mm of water needs to be applied every 3 days in order to maintain soil moisture within the 50% available water threshold for corn that is 70 days old.

PAGE 82

82 Soil water potential sensors The soil water potential sensors implemented at SIRP were components of the University of Georgia Smart Sensor A rray (Vellidis, 2011) (UGA SSA). The UGA SSA c onsists of a (V ellidis, 2011) . Each node includes a circuit board, radio transmitter, three Watermark soil moisture potential sensors, and temperature sensors (Vellidis, 2011) . The soil moisture potential sensors are a wireless system that measures soil water tension (kPa) at twenty, forty, and sixty centimeters. A weighted average of so il water tension in all nine SSA plots was calculated daily. When the weighted average approached 25 30 kPa, irrigation was scheduled for the following day. A 25 30 kPa weighted average threshold for irrigation was determined as the maximum allowable depletion for field corn in field experiments conducted at SIRP. Effect of Irrigation Depth on Corn Y ield Daily irrigation applications were recorded in a spreadsheet for each treatment at both sites. Irrigation applied was measured as a depth in mm . Harvest at NFREC SV was conducted on August 16 and 17, 2018 when the corn was observed to be physiologically mature (black layer). Corn yield measurement was conducted on the 6 and 7 planting rows beginning three meters inside each plot to c ounter border effects. An overall length of 6 meters within each harvest row was selected for data analysis. Within the specified 6 meters in both harvest rows, plants were counted and corn ears were hand harvested and placed in sacks. The ears were then r ecorded for total wet weight after removing the husk. Using an electric sheller, the total ears per plot were shelled, placed in sacks, and weighed. Finally, average grain moisture content was determined by measuring the moisture content percentage of thre e shelled grain samples from

PAGE 83

83 each bag with a John Deere Moisture Check Plus . Final corn marketable grain yield (kg/ha) at 15.5% moisture content was calculated for each treatment using the method for hand harvested shelled corn (Lauer, 2002) . Harvest at SIRP was conducted on August 28 th , 2018 when the corn was observed t o be physiologically mature. A total of 15 m in the fourt h and fifth planting rows were selected for data analysis. Yield and moisture for each plot were determined for the two selected harvest rows using the UGA variety trial combine. Dry matter was calculated for each plot by multiplying the weight of harveste d corn by the percentage of dry matter in each sample. The dry matter weight was then divided by 0.855 to obtain the grain yield at a marketable moisture weight of 15.5%. Finally, dry matter weight was divided by the plot area to obtain the yield per plot. Data A nalysis The effects of fertilization and irrigation treatments on grain marketable yield (kg/ha) and irrigation water use efficiency ( IWUE ) (kg/m 3 ) were determined using the GLIMMIX procedure in the Statistical Analysis Software (SAS) application . This procedure provides estimation and statistical interpretation of general linear mixed effect models. The general linear mixed model assumes random effects are normally distributed. The random effects assessed in this study were the replication and fi eld, while the fixed variables were fertilization and irrigation treatments. A Tukey honest significant difference test was used to determine significant difference in mean grain market yield in response to irrigation and fertilization treatments. Assumpti ons of the linear mixed effect model were assessed using the GLIMMIX procedure for marketable grain yield data collected at both NFREC SV and SIRP (Figure 3 3; Figure 3 4). The errors in both experiments were found to be independent, normally distributed, and have constant variance.

PAGE 84

84 The random dispersion in the residual plots showed that a linear model was appropriate to assess fertilization and irrigation effects on market grain yield. Water Use E fficiency Irrigation water use efficiency (IWUE) for each t reatment at NFREC SV and SIRP was determined as the ratio of irrigated crop yield to irrigation water applied (Mandal, 2001) (Equation 3 1). (3 1) IWU E = irrigation water use efficiency (kg/m 3 ) Y i = economic yield of irrigated crop (kg/ha) IR i = depth of i rrigation water applied (m 3 /ha) In addition, rain fed irrigation water use efficiency (IWUE RF ) for each treatment at NFREC SV was determined as the ratio of irrigated crop yield minus rain fed crop yield to irrigation water applied (Irmak et al., 2011) (Equation 3 2). Since there was no rain fed treatment at SIRP, this calculation was only applied to the treatments at NFREC SV. (3 2) IWUE RF = irrigation water use efficiency (kg/m 3 ) Y i = marketable grain yield of irrigated crop (kg/ha) Y r = marketable grain yield of rainfed crop (kg/ha) IR i = depth of irrigation water applied (m 3 /ha)

PAGE 85

85 Crop Water Use E fficiency Crop water use efficiency (CWUE) for each irrigation treatment at both sites was calculated as the ratio of marketable grain yield to crop evapotranspiration (Equation 3 3). Crop evapotranspiration was estimated in this study using the Pe nman Monteith FAO 56 equation and theoretical crop coefficients (Irmak, 2015) . (3 3) CWUE = irrigation water use effic iency (kg/m 3 ) Y i = marketable grain yield for each irrigation treatment (kg/ha) ET c = crop evapotranspiration (m 3 /ha) Measured Soil Moisture for each Irrigation T reatment Daily volumetric moisture content readings measured by soil moisture probes underly ing each irrigation treatment at NFREC SV from April 4 th , 2018 through August 16 th , 2018 were downloaded remotely from a website. A total of six soil moisture probes installed within the corn app and non irrigated treatment plots, eighteen probes installed within the calendar based treatment plots, and twenty four soil moisture sensors installed within the SMS treatment plots were used for comparison. Each probe consists of nine sensors that estimate volumetric water content in 10 cm increments from 5 cm t o 85 cm. Volumetric water content estimated by each sensor is reported in mm of water per cm of soil. The volumetric water content estimated by each sensor located from 5 cm to 65 cm was multiplied by a 10 cm depth to determine a total moisture content (m m) within each sensor measuring zone. The sum of moisture content measured from each sensor from 5 cm to 65 cm was used to estimate total moisture content within the root zone (RZMC) (Equation 3 4). An average moisture content for each treatment

PAGE 86

86 was determ ined by taking an average of total moisture content determined for each soil moisture probe. (3 4) In addition, the number of days each irrigation treatment exceeded the 50% maximum allowabl e depletion threshold from April 4, 2018 to August 14, 2018 was estimated in this study. For each probe, this metric was determined by first adding the volumetric moisture content estimated by each sensor from 5 cm to 65 cm. The sum of volumetric moisture content from each sensor was then divided by the effective root depth of 650 mm to determine a percentage of water within the root zone. The percentage of available water for each probe was calculated by dividing the percentage of water within the root zon e by the field capacity of 12.5 % determined using guidelines proposed by Zotarelli et al. (2013). Daily average available moisture content for each treatment was determined by taking the average available moisture content determined for the probes corres ponding to each irrigation treatment. Finally, the number of days where average available moisture content exceeded 50% MAD were counted and summed for each irrigation treatment. Results and Discussion Weather C onditions Seventy five precipitation even ts totaling 635 mm were recorded by the Live Oak FAWN station at the NFREC SV during the 2018 g rowing season. Compared to a 30 year average from 1981 2010, above average rainfall was observed in April, May, and July. (Figure 3 6) (U.S. Climate Data, 2018) . Corn was planted on March 8, 2018 and received 31 mm of

PAGE 87

87 rainfall thr oughout the rest of March. While rainfall occurred 23 days in March, it was a relatively low amou nt of rainfall compared to a 30 year average , from 1981 2010, of 133 mm for the month of March in Live Oak, Florida (U.S. Climate Data, 2018) . Below average rainfall occurred in June with 64 mm less rainfall occurring than a 30 year average from 1981 2010 (Figure 3 6). Compared to a 30 year average from 1981 2010 , monthly temperatures were within approximately 1 °C throughout the growing season with the exception of higher temperatures observed in the month of March where average temperature were 2.5 °C greater (Figure 3 6 ) (U.S. Climate Data, 2018 ) . A total of 924 mm of rainfall was recorded by the Camilla GAEMN station located at SIRP for the 2018 growin g season. In comparison to a 30 year average for Camilla, Georgia, above average rainfall was recorded by the Camilla GAEMN rain gauge in April with 128 mm total , May with 131 mm total , June with 228 mm total , and July with 218 mm total (Figure 3 7). Approximately 39 mm more rainfall was recorded in April by the Camilla GAEMN station with 128 mm total in comparison to a 30 year average from 1981 2010. Approximately 93 mm and and 68 mm more rainfall was recorded by the Camilla GAEMN station for the months of June and July, respectively, in comparison to a 30 year average from 1981 2010. Approximately 191 mm of rainfall was recorded at SIRP from Au gust 1 st , 2018 until harvest on August, 28 th 2018, which was approximately 62 mm higher than a 30 year average for the month of August in Camilla, Georgia (U.S. Climate Data, 2018) . In comparison to a 30 year average from 1981 2010, monthly temperatures recorded by the Camilla GAEMN station were within 1 °C for March, April , July, and August and between 1 to 2 °C higher in May and June (Figure 3 7).

PAGE 88

88 Effect of I rrigation and F ertilization on Marketable Grain Y ield and IWUE at NFREC SV I rrigation scheduling for corn at the NFREC SV site began April 8, 2018 (31 DAP) and was t erminated on July 20, 2018 (134 DAP). The calendar based treatment applied the most irrigation water with 497 mm total (Table 3 7). The corn app treatment applied 43% less water than the calendar based treatment with a total of 283 mm (Figure 3 8). On aver age, the SMS treatment applied 72% less water than the calendar based treatment (Figure 3 8). The water savings achieved using improved irrigation scheduling techniques in this study agree with the ones reported in a similar study conducted in Live Oak, Fl , at the NFREC SV in 2015. The study conducted at NFREC SV in 2015 achieved water savings of 42% and 53% with no significant differences in marketable grain yield compared to a calendar based treatment by utilizing ET based and SMS based irrigation schedul ing, respectively (Zamora Re, 2019) . The type III test for fixed effects determined a significant interaction between irrigation and fertilization on marke table grain yield (Table 3 8). There were no significant differences in marketable grain yield between the calendar based, corn app, and SMS irrigation treatments across the N2 and N3 nitrogen regimes, demonstrating similar yields can be achieved by apply ing less water (Figure 3 9). However, marketable grain yields for the calendar based and corn app irrigation treatments were statistically lower than the SMS irrigation treatment for the N1 nitrogen fertilization regime, which indicates that nitrogen may h ave been a limiting factor considering applying more water did not increase yields in this case. Marketable grain yields were significantly lower for the non irrigated treatment in comparison to the calendar based for the N3 nitrogen fertilization regime, which indicates water may have been a limiting factor considering applying mor e water improved yields for this interaction.

PAGE 89

89 As expected, the Corn App and SMS irrigation treatments improved irrigation water use efficiency by 45% and 74% in comparison to the calendar based treatment, respectively (Table 3 7). A significant interaction between fertilization and irrigation on IWUE was determined in this study , which indicates that IWUE changes in response to fertilization and irrigation (Table 3 9). Irrigation water use efficiency was statistically lower for the calendar based treatment in comparison to the app and SMS treatments across all nitrogen fertilization regimes, demonstrating that applying more water significantly reduced the efficiency of the system ( Figure 3 10) . While the calendar based treatment achieved the highest yields for the N3 (336 kg/ha) treatment, IWUE was significantly lower in comparison to the app across the N2 and N3, and the SMS treatment across all three nitrogen fertilization regimes , demonstrating the app and SMS treatments are more efficient irrigation scheduling methods despite lower amounts of nitrogen applied (Figure 3 10) . In addition, t he SMS treatment produced the highest average water use efficiency and cr op water use effici ency of 8.2 and 2.2 kg m 3 , respectively (Table 3 7). A study conducted in Nebraska utilized soil moisture sensor based irrigation scheduling for corn production and achieved an average water use efficiency and crop water use efficiency of 20 and 2.75 kg m 3 , respectively, (Irmak et al., 2012) . Higher irrigation and crop water use efficiencies achieved with S MS based irrigation scheduling by Irmak et al., (2012) could be explained by increased available water holding capacity at the fields tested in Nebraska due to larger percentages of silt, clay, and organic matter in comparison to the primarily sandy soil a t our site. Corn grows optimally in a well drained, medium textured soil with high water holding capacity, which fits the description of the predominantly silt loam and silty clay loam soils at the study sites in Nebraska (Irmak et al., 2012; McClure, 2009 ) . However, it is important to note that the values for irrigation water use efficiency calculated for this study and the study conducted in

PAGE 90

90 Nebraska are typically higher than other values reported in literature due to not accounting for a non irrigated t reatment. The IWUE RF values calculated for the treatments at NFREC SV are lower than the IWUE calculations that do not account for the non irrigated treatment (Table 3 7). For example, the IWUE RF value for the corn app treatment was 0.3 kg m 3 , which is g reater than the IWUE value for the corn app treatment of 3.8 kg m 3 . The calendar based treatment produced the lowest IWUE and CWUE of 2.1 and 1.9 kg m 3 , respectively. The corn app produced an average IWUE and CWUE of 3.8 and 2.0 kg m 3 , respectively . A similar study conducted by Djaman et al., (2018) reported similar average IWUE and CWUE values of 1.74 and 1.53 kg m 3 , respectively, utilizing ET based irrigation scheduling for field corn at the New Mexico State University Agricultural Science Center a t Farmington. While irrigation often provides insurance for productive corn yields, no significant differences in market grain yield were observed between the non irrigated and other irrigation treatments for the N2 (224 kg/ha) nitrogen regime (Figure 3 9 ). Above average rainfall for the majority of the 2018 growing season at NFREC SV may have reduced the additional beneficial effects of irrigation on grain yield in comparison to a non irrigated treatment. Rainfall distribution concentrated at, or near, th e highest points of crop water requirements may reduce irrigation demand. Water demand in corn is highest during the tasseling stage and a sufficient amount of water should be applied around this time to help achieve yield goals (Table 3 6) (Porter, 2017) . Tasseling was observed in both fields on May 21 and 22, 2018 (Figure 3 2). Thirteen rainfall events occurred the week prior, durin g, and after tasseling for a total of 150 mm, which could explain the higher than expected yields for the non irrigated treatment. In addition, total rainfall exceeded crop evapotranspiration estimated for the 2018 growing season, which may have reduced ir rigation demand. Further research is needed to determine how the

PAGE 91

91 corn app and SMS based irrigation methods perform during years that are generally drier or with sparsely distributed rainfall. In comparison to the SMS treatment, average marketable grain yie lds were significantly lower for the calendar based and app treatments across the N1 (112 kg/ha) fertilization rate, respectively (Figure 3 9). Nutrient loss from over application of water at the beginning of the season for the lowest nitrogen application rate may have reduced yields for the corn app and calendar based treatment, which applied water more frequently around fertilization events in comparison to the SMS treatment. The soils at NFREC SV are primarily sand with a low water holding capacity of 0. 07 cm/cm. Soils with low water holding capacity pose higher risk for nitrate leaching than finer textured soils, which could explain the significantly lower yields for the more frequently irrigated calendar based and corn app treatments in comparison to th e SMS treatment. Effect of I rrigation and F ertilization on M arketable Grain Y ield and IWUE at SIRP For the experiment conducted at SIRP, the checkbook method applied the most irrigation water for a total of 325 mm (Table 3 10 ). The corn app and SSA treatm ents applied 57 and 59 % less water with no significant differences in marketable grain yield compared to the checkbook method, respectively (Table 3 10; Figure 3 12 ). As expected, the corn app and soil water potential sensor irrigation scheduling methods i mproved water use efficiency by 56% and 58% with no significant differences in marketable grain yield, respectively (Table 3 10) . A similar study conducted in 2014 and 2015 at the E.V. Smith Research and Extension Center (EVSREC) in Shorter, Alabama achiev ed similar water savings of 69 and 53% , respectively, by utilizing soil moisture sensor based irrigation scheduling in comparison to a checkbook method without a significant reduction in yields (Filho, 2016)

PAGE 92

92 The type III test for fixed eff ects indicated there were no significant differences in marketable grain yield between any of the irrigation and fertilization treatments (Table 3 11) . Comparable grain yields produced for every treatment could be explained by testing similar nitrogen appl ication rates as well as above average rainfall and no non irrigated treatment. The traditional (334 kg/ha) and high (336 kg/ha) nitrogen treatments applied similar amounts of nitrogen but differed in the methods of application. In addition, the low (277 kg/ha) nitrogen treatment was relatively high compared to the IFAS recommended rate for irrigated corn of 235 kg/ha. Similar nitrogen application rates can reduce variability in grain yield between irrigation regimes. In comparison to the corn app and SMS treatments for the study conducted at NFREC SV during the 2018 growing season, IWUE and CWUE values were slightly higher for the corn app and UGA SSA treatments at SIRP. A significant interaction between irrigation and IWUE was determined in this study, w hich indicates IWUE changes in response to irrigation (Table 3 12). In comparison to the app and SSA irrigation treatments, the checkbook treatment had significantly lower IWUE, indicating the app and SSA treatments applied water more efficiently Figure 3 13). Higher IWUE values achieved at SIRP in comparison to the NFREC SV were the result of higher average marketable grain yields and less water applied for the corn app and soil moisture treatments. The field tested at SIRP had a slightly higher available soil water holding capacity of 0.08 cm/cm in comparison to our field with an estimated available soil water holding capacity of 0.07 cm/cm. While the difference between available water holding capacities at both sites is small, the soils at SIRP may be ab le to hold more PAW , which can reduce irrigation demand and crop water stress. An additional 289 mm of total rainfall for the 2018 growing season was recorded at SIRP in comparison to NFREC SV. The corn app treatment conducted at

PAGE 93

93 SIRP applied 143 mm less water in comparison to the corn app treatment for NFREC SV. Additional rainfall at SIRP may have reduced irrigation demand, which could explain the higher irrigation water applied for the corn app treatment conducted in comparison to NFREC SV. The CWUE va lues achieved by the corn app at SIRP and NFREC SV were 3.4 kg/m 3 and 1.2 kg/m 3 , respectively (Table 3 10) . Total crop evapotranspiration estimated at SIRP was approximately 88 mm less than total crop evapotranspiration estimated at NFREC SV for the 2018 g rowing season. This difference in estimated crop evapotranspiration helps explain the larger CWUE values achieved by the corn app at SIRP. Less total crop evapotranspiration estimated at SIRP can be attributed to greater amounts of rainfall and number of r ainfall days in comparison to NFREC S V for the 2018 growing season. Soil Moisture C omparison The calendar based irrigation treatment had an estimated average soil moisture content within the effective root zone of 80 cm, which was 8 cm, 11 cm, and 19 cm greater than the average soil moisture content estimated for the corn app, average SMS, and non irrigated trea tments, respectively (Table 3 13; Figure 3 14 ). As expected the corn app and SMS treatments had lower moisture content within the effective crop root zone than the cale ndar based treatment (Table 3 13; Figure 3 14 ). Despite lower average soil moisture content for the corn app and SMS treatments, there were no significant difference in marketable grain yield in comparison to the calendar based trea tment across all fertilization regimes. A 50% maximum allowable depletion was used to schedule irrigation to avoid plant water stress that may negatively impact yields. As expected, the non irrigated treatment experienced the most days (12 total) where s oil moisture within the effective root zone was estimated to exceed the 50% maximum allowable depletion threshold (Table 3 13 ). A significant

PAGE 94

94 difference in marketable grain yield (P value =0.00) was observed between the I1 (calendar based) and I5 (non irri gated) treatments for the N3 (336 kg/ha) fertilization rate. (Table 3 9). The calendar based treatment was estimated to experience 0 days of potential water stress from exceeding the 50% MAD threshold, demonstrating that applying irrigation water can allev iate plant stres s and improve yields (Table 3 13 ). On average, the SMS treatment was estimated to experienced 2 days of potential water stress with no significant difference in marketable grain yield in comparison to the calendar based treatment, demonstra ting soil moisture sensor based irrigation scheduling can effectively manage crop water stress without negatively impacting ma rketable grain yield (Table 3 13 ; Figure 3 9). Conclusion The objectives in this study were to compare irrigation application d epths and IWUE for four irrigation treatments (calendar based, corn app, SMS, and non irrigated) at NFREC SV and three irrigation treatments (corn app, SSA, and checkbook) at SIRP, and compare measured soil moisture using SMS data collected within plots at NFREC SV . The corn app irrigation treatment applied a total depth of 283 mm and 141 mm of water at NFREC SV and SIRP respectively, which was 43% and 57% lower than what the traditional calendar based and checkbook treatments applied. No significant differ ences in marketable grain yield were determi ned for the corn app treatment in comparison to the traditional calendar based and checkbook irrigation scheduling methods across all nitrogen fertilization regimes , demonstrating utilizing improved ET based irri gation scheduling methods can help reduce irrigation applied volumes without reducing yields. The SMS treatment achieved water savings of 72% with a total of 141 mm of irrigation water applied in with no significant differences in yield in comparison to th e calendar -

PAGE 95

95 based treatment, which applied 497 mm. of total water. The UGA SSA treatment applied a total irrigation depth of 135 mm, resulting in water savings of 58% in comparison to the checkbook treatment, which applied 325 mm total. Water savings achi eved by SMS and SSA treatments with no significant differences in marketable grain yield in comparison to the traditional calendar based and checkbook treatments, respectively demonstrate irrigation BMPs such as SMS based irrigation scheduling can reduce i rrigation water while maintaining yields. The corn app and SMS treatments should be tested for multiple years at multiple sites to realize their full potential in improving irrigation scheduling. The average estimated soil moisture content within the eff ective root zone for the corn app, SMS, and non irrigated non irrigated, corn app, and average SMS treatments was 72 cm, 69 cm, and 61 cm, correspondingly, which were all lower than the calendar based treatment with an estimated average soil moisture conte nt of 80 cm for the 2018 growing season (Table 3 13) . Despite lower average soil moisture content for the corn app and SMS treatments, no significant differences in marketable grain yield were determined in comparison to the calendar based treatment. Compa ring a verage moisture content from measured SMS data for each treatment was l imited in that there was an uneven number of soil moisture probes used for each treatment and the sensors may not provide a completely accurat e measurement considering they provid e an estimate of volumetric moisture content.

PAGE 96

96 Figure 3 1 . Aerial view of 2018 field layout at NFREC SV . Photo courtesy of author. Table 3 1 . Irrigation treatments at NFREC SV Irrigation tre atment Corresponding plot label Calendar Based I1 Smartirrigation Corn App I2 Soil Moisture Sensor (SMS) I3 No irrigation I5

PAGE 97

97 Figure 3 2 . Various growth stages of corn crop observed at NFREC SV. A) This picture was taken on April 12, 2018 (14 DAP) and shows the corn crop in the V3 V5 vegetative growth stage. B) This Picture was taken on May 1, 2018 (33 DAP) and shows the corn crop in the V8 V10 vegetative growth stage. C) This photo was taken on May 22, 2018 ( 54 DAP) and shows the corn crop has begun tasseling. D) This picture was taken on July 10, 2018 (103 DAP) and shows a corn ear beginning to dent. Photos courtesy of author .

PAGE 98

98 Figure 3 3 . Aerial view of SIRP field layout. Numbers 1 9 on the left hand side represent variable rate irrigation zones for the Newton Lateral irrigation system. Plots are labelled by the plot number (111 139) followed by a dash and the irrigation and fertilization regime (1 9). Photo courtesy of author.

PAGE 99

99 Ta ble 3 2 . Irrigation by fertilization treatments and corresponding plot numbers 2018 Corn Treatments Number App x High N (336 kg/ha) 1 App x Traditional (333 kg/ha) 2 App x Low N (277 kg/ha) 3 Checkbook x High N (336 kg/h a) 4 Checkbook x Traditional (333 kg/ha) 5 Checkbook x Low N (277 kg/ha) 6 UGA SSA x High N (336 kg/ha) 7 UGA SSA x Traditional (333 kg/ha) 8 UGA SSA x Low N (277 kg/ha) 9

PAGE 100

100 Table 3 3 . Field properties for both testing sites. Field properties Site 1 (SVAEC) Site 2 (SIRP) Latitude 30.31353 N 31.28042° N Longitude 82.90122 W 84.30011° W Elevation (m) 44 48 Varieties Pioneer 1870 Hybrid Pioneer 1870 Hybrid Planting date March 8, 2018 March 29, 2018 Plot length (m) Number of rows per plot 12 12 15 8 Spacing between plots (m ) 6 6 Number of irrigation treatments 4 3 Irrigation system Two span Valley Linear End feed 8000 Newton lateral S oil water holding capacity (cm/cm ) 0.07 0.08

PAGE 101

101 Table 3 4 . NFREC irrigation treatments and descriptions Treatment Description I1 (Calendar based) practices. Consists of zero irrigation for the first 30 days after plantin g (DAP) unless severe windy conditions caused blowing sand to damage the plants. Beginning on 31 DAP, a target amount of 30 mm/week was established and was made up of rain or irrigation with the exception that rain events had to be 6 mm or larger. For 40 5 9 DAP a 40 mm/week target was established. Irrigation applications were skipped if 13 25 mm/week rainfall occurred and two irrigations were skipped if >25 mm of rain occurred. For 60 105 DAP a 50 60 mm/week irrigation target was used unless 13 25 mm of rai n occurred the day prior to a scheduled irrigation. Two irrigations were skipped if >25 mm of rain occurred. Finally, around 105 DAP at full dent stage, weekly irrigation targets were reduced to 40 mm/week for one week and 20 mm/week for another week until final irrigation was terminated at 115 DAP. I2 (App) Irrigation was determined using a theoretical root zone water deficit. This is calculated daily by subtracting effective irrigation plus rain (I+R) from the actual evapotranspiration (ET c ). ET c is det ermined by multiplying ET o by a crop coefficient based on growing degree days (GDD). ET o is calculated by utilizing the ASCE Penman Monteith FAO 56 equation with daily weather data from the Live Oak FAWN weather station. I3 (SMS) Using the SENTEK probes, volumetric moisture content of the soil was monitored and irrigation was determined using the maximum allowable depletion (MAD) and field capacity (FC) points to refill the soil profile with irrigation according to guidelines proposed by (Zotarelli et al. 2013) I5 (NO): non irrigated plots

PAGE 102

102 Table 3 5 . SIRP irrigation treatments and descriptions Treatment Description 1,2,3 (App) Irrigation was determined using a theoretical root zone water deficit. This is calculated daily by subtracting effective irrigation plus rain (I+R) from the actual evapotranspiration (ET c ). ET c is determined by multiplying ET o by a crop coefficient based on growing degree days (GDD). ET o is calculated by utili zing the ASCE Penman Monteith FAO 56 equation with daily weather data from the Live Oak FAWN weather station. 4,5,6 (Checkbook Method) This method schedules irrigation by applying the maximum expected weekly crop water use minus recorded precipitation. W eekly crop water use is a function of weeks after planting. (Table 3 6) (Migliaccio et al., 2016) . 7,8,9 (SSA) The sensors implemented were the University of Georgia Smart Sensor Array (UGA SSA). This network of soil moisture potential sensors ar e a wireless system that measures soil water tension at twenty, forty, and sixty centimeters. A 25 30 kPa soil water tension was used as the irrigation triggering threshold.

PAGE 103

103 Table 3 6 . Estimation of corn water use f or checkbook irrigation scheduling at SIRP Growth stage Days after planting Crop water use (cm/day) Emergence 0 7 8 12 0.07 0.13 Two leaves expanded (V2) 13 17 18 22 0.18 0.23 Four to six leaves expanded (V4 V6) 23 27 28 32 33 36 0.30 0.36 0.43 Six to eight leaves (V6 V8) 47 41 42 45 0.48 0.53 Ten to twelve leaves (V10 V12), ear shoots developing, bottom two leaves lost 46 50 51 54 0.58 0.64 Twelve to sixteen leaves (V12 V16), rapid elongation of top two ear shoots 55 59 60 64 0.69 0.74 Tasseling 6 5 69 0.79 Pollination and silk emergence 70 74 75 79 0.81 0.84 Blister stage 80 84 0.84 Milk stage 85 89 0.86 Early dough 90 94 0.86

PAGE 104

104 Table 3 6. Continued Growth stage Days after planting Crop water use (cm/day) Dough stage 95 99 0.84 Early dent 1 00 104 0.76 Dent 105 109 0.69 Early black layer 110 114 0.61 Black layer (maturity) 115 119 0.53

PAGE 105

105 Figure 3 4 . Conditional residual plots of market grain yield for data collected at NFREC SV. A. Residual plot B. Res iduals distribution C. QQ plot of residuals D. Box and whisker plot of residuals A. B . C . D .

PAGE 106

106 Figure 3 5 . Conditional residual plots of market grain y ield for data collected at SIRP. A. Residual plot B. Residuals distribution C. QQ plot of residuals D. Box and whisker plot of residuals

PAGE 107

107 Figure 3 6 . Total monthly rainfall and temperatures recorded by the Live Oak FAWN station in comparison to a thirty year average from 1981 2010

PAGE 108

108 Figure 3 7 . Average monthly rainfall and temperature recorded by the Camilla GAEMN station in comparison to a thirty year average from 1981 2010

PAGE 109

109 Figure 3 8 . Precipitation events and application to tals by irrigation treatment at NFREC SV

PAGE 110

110 Table 3 7 . Comparison of marketable grain yield, irrigation applied, IWUE RF , IWUE, and CWUE between irrigation treatments at NFREC SV Irrigation Treatment Marketable Gra in Yield Irrigation applied Water savings IWUE RF IWUE Increase in IWUE CWUE (kg/ha) (mm) (m 3 /ha ) (mm ) (%) (kg/m 3 ) (kg/m 3 ) (%) (kg/m 3 ) Calendar based (I1) 10191 497 4970 0 0 0.1 2.1 0 1.9 Corn app (I2) 10662 283 2830 214 43% 0.3 3.8 45% 2.0 SMS North Field (I3) 11799 121 1210 376 76% 1.56 9.8 79% 2.2 SMS South Field (I3) 11280 161 1610 336 68% 0.85 7.0 70% 2.1 SMS Average (I3) 11540 141 1410 356 72% 1.15 8.2 74% 2.2 Non irrigated (I5) 9913 29 290 468 94% 1.9

PAGE 111

111 Table 3 8 . Type III tests for significant interactions between fixed effects and marketable grain yield for the treatments at NFREC SV Fixed Effect F Value Pr > F Irrigation 1.91 0.16 Fertilization 117.61 <0.0001 Irrigation*Fertilization 7.68 <0.0001

PAGE 112

112 Figure 3 9 . Boxplot comparison of marketable grain yield in response to irrigation and fertilization regimes at NFREC SV (I1 = calendar based, I2 = corn app, I3 = SMS, and I5 = non irrigated) (N1= 112 kg/ha, N2 = 224 kg/ha, and N3 = 336 kg/ha). Box plots with the same letter represent treatments with marketable grain yields that are not significantly different from one another. The line in the middle of each box plot represents the median marketable gr ain yield for each irrigation and fertilization treatment. The box itself represents the interquartile range of marketable grain yield in response to each treatment and the space between the box and upper and lower whiskers represent the upper and lower qu artiles, respectively.

PAGE 113

113 Table 3 9 . Type III tests for significant interactions between fixed effects and IWUE for the treatments at NFREC SV Fixed Effect F Value Pr > F Irrigation 1.91 <0.0001 Fertilization 117.61 <0. 0001 Irrigation*Fertilization 7.68 0.02 Figure 3 10 . Boxplot comparison of IWUE in response to irrigation and fertilization regimes at NFREC SV (I1 = calendar based, I2 = corn app, I3 = SMS) (N1= 112 kg/ha, N2 = 224 kg/ha , and N3 = 336 kg/ha). Box plots with the same letter represent treatments with IWUE that are not significantly different from one another. The line in the middle of each box plot represents the median marketable grain yield for each irrigation and fertili zation treatment. The box itself represents the interquartile range of marketable grain yield in response to each treatment and the space between the box and upper and lower whiskers represent the upper and lower quartiles, respectively.

PAGE 114

114 Figure 3 11 . Precipitation events and application totals by irrigation treatment at SIRP.

PAGE 115

115 Table 3 10 . Comparison of marketable grain yield, irrigation applied, IWUE, and for irrigation treatments at SIRP Table 3 11 . Type III tests for significant interactions between fixed effects and marketable grain yield for the treatments at SIRP Fixed Effect F Value Pr > F Irrigati on 0.08 0.92 Fertilization 0.54 0.59 Irrigation*Fertilization 0.38 0.82 Irrigation Treatment Marketable Grain Yield Irrigation applied Water savings IWUE Increase in IWUE CWUE (kg/ha) (mm) (m 3 /ha) (mm) (%) (kg/m 3 ) (%) (kg/m 3 ) Corn app 15251 140 1400 1850 57% 10.9 56% 3.4 Checkbook 15484 325 3250 0 0 4.8 0% 3.5 SWS 15404 135 1350 1900 59% 11.4 58% 3.4

PAGE 116

116 Figure 3 12 . Boxplot comparison of marketable grain yield in response to irrigation and fertilization regimes at SIRP (L = 277 kg/ha, T = 334 kg /ha, and H = 336 kg/ha). Box plots with the same letter represent treatments with marketable grain yields that are not significantly different from one another. The line in the middle of each box plot represents the median marketable grain yield for each i rrigation and fertilization treatment. The box itself represents the interquartile range of IWUE in response to each treatment and the space between the box and upper and lower whiskers represent the upper and lower quartiles, respectively. Table 3 12 . Type III tests for significant interactions between fixed effects and IWUE for the treatments at SIRP Fixed Effect F Value Pr > F Irrigation 1.91 <0.001 Fertilization 117.61 0.55 Irrigation*Fertilization 7.68 0.63

PAGE 117

117 Figure 3 13 . Boxplot comparison of IWUE in response to in response to irrigation and fertilization regimes at SIRP (L = 277 kg/ha, T = 334 kg/ha, and H = 336 kg/ha).. Box plots with the same letter represent treatments with IWUE that are not significantly different from one another. The line in the middle of each box plot represents the median marketable grain yield for each irrigation and fertilization treatment. The box itself represents the interquartile range of IWUE in response to each treatment and the space between the box and upper and lower whiskers represent the upper and lower quartiles, respectively.

PAGE 118

118 Figure 3 14 . Soil moisture comparisons for each irrigation treatment at NFREC SV . A) Daily c omparison of soil moisture within effective root zone estimated by soil moisture probes for each irrigation treatment at NFREC SV from April 4, 2018 to August 14, 2018. B) Comparison of average daily soil moisture content estimated within root zone for eac h irrigation treatment at NFREC SV from April 4, 2018 to August 14, 2018.

PAGE 119

119 Table 3 13 . Estimation of number of days each irrigation treatment exceeded a 50% allowable depletion threshold according to measured SMS data Irrigation treatment Average Soil Moisture within Effective Root Zone (cm) Number of days exceeded 50% allowable depletion I1 (Calendar based) 80 0 I2 (App) 72 0 I3 N (SMS) 73 2 I3 S (SMS) 64 2 I5 Non irrigated 61 12

PAGE 120

120 LIST OF REFERENCES 2017. Netafim's New Mobile App Combines Field, Climate and Weather Data to Forecast the Watering Needs of Corn Crops. Busine ss Wire, Inc. Allen, R., Pereira, L., Raes, D., Smith, M., 1998. FAO Irrigation and drainage paper No. 56. pp. 1 79, 89 99, 103 129, 161 169. Bartlett, A.C., Andales, A.A., Arabi, M., Bauder, T.A., 2015. A smartphone app to extend use of a cloud based irrigation scheduling tool. Computers and Electronics in Agriculture 111, 127 130. Bellino, J.C., Kuniansky, E.L., O'Reilly, A.M., Dixon, J.F., 2018. Hydrogeologic setting, conceptual groundwater flow system, and hydrologic conditions 1995 2010 in Florida and parts of Georgia, Alabama, and South Carolina, Scientific Investigations Report. Reston, VA, pp. 1 3, 11 15, 1 15. Berndt, M.P., 2014. Water quality in the upper Floridan aquifer and overlying surficial aquifers, Southeastern United States, 1993 2010, Circular: 1355. Reston, Virginia : U.S. Department of the Interior, U.S. Geological Survey, 2014., pp. 11 12, 33 34 . Acquisition and Irrigation Scheduling. American Society of Agricultural and Biological Engineers, pp. 1 6. Dark Sky, 2018. Dark Sky weather sources. David, R.S., Pa ul, J.B., Xiaoying, Y., Scott, A.S., Stephen, M.W., Michael, D.A., 2013. Tapping unsustainable groundwater stores for agricultural production in the High Plains Aquifer of Kansas, projections to 2110. Proceedings of the National Academy of Sciences of the United States of America(37), 14. Djaman, K., Irmak, S., 2013. Actual Crop Evapotranspiration and Alfalfa and Grass Reference Crop Coefficients of Maize under Full and Limited Irrigation and Rainfed Conditions. Journal of Irrigation & Drainage Engineering 139(6), 433 446. Dukes, M.D., Zotarelli, L., Liu, G.D., Simonne, E.H., 1995. Principles and Practices of Irrigation Management for Vegetables. http://edis.ifas.ufl.edu/pdffiles/cv/cv10700.pdf . El Wahed, M.H.A., Ali, E.A., 2013. Effect of irrigation systems, amounts of irrigation water and mulching on corn yield, water use efficiency and net profit. Agricultural Water Management, 64. Evans, R., Cassel, D., Sneed, R.E., 1996. Soil, Water and Cr op Characteristics Important to Irrigation Scheduling. Filho, L.C.D.F.J., 2016. Evaluation of Irrigation Scheduling Methods and Nitrogen Fertilization Effect on Corn Production in Alabama Soil and Environmental Sciences. Auburn University.

PAGE 121

121 González Pere a, R., Fernández García, I., Martin Arroyo, M., Rodríguez Díaz, J.A., Camacho Poyato, E., Montesinos, P., 2017. Multiplatform application for precision irrigation scheduling in strawberries. Agricultural Water Management 183, 194 201. Grabow, G.L., Ghali, I.E., Huffman, R.L., Miller, G.L., Bowman, D., Vasanth, A., 2012. Water Application Efficiency and Adequacy of ET Based and Soil Moisture Based Irrigation Controllers for Turfgrass Irrigation. Journal of Irrigation & Drainage Engineering 139(2), 113 123. Grubbs, J.W., Crandall, C.A., 2007. Exchanges of water between the upper Floridan aquifer and the lower Suwannee and lower Santa Fe rivers, Florida, Professional paper: 1656 C. Reston, Va. : U.S. Geolo gical Survey, 2007., pp. 24 26. Hao, B., Xue, Q., Mar ek, T., E. Jessup, K., Becker, J., Hou, X., Xu, W., Bynum, E., Bean, B., Colaizzi, P., Howell, T., 2015. Water Use and Grain Yield in Drought Tolerant Corn in the Texas High Plains. Hodges, A.W., Rahmani, M., Court, C., 2017. Economic Contributions of Agri culture, Natural Resources, and Food Indust ries in Florida in 2015. p. 18. Howell, T., Evett, S., Tolk, J., S. Copeland, K., A. Dusek, D., Colaizzi, P., 2006. Crop Coefficients Developed at Bushland, Texas for Corn, Wheat, Sorghum , Soybean, Cotton, and Alf alfa. Howell, T.A., 1996. Irrigation scheduling research and its impact on water use. American Society of Agricultural Engineer St. Joseph, MI, pp. 21 33. Irmak, S., 2015. Interannual Variation in Long Term Center Pivot Irrigated Maize Evapotranspiration a nd Various Water Productivity Response Indices. II: Irrigation Water Use Efficiency, Crop WUE, Evapotranspiration WUE, Irrigation Evapotranspiration Use Efficiency, and Precipitation Use Efficiency. Journal of Irrigation and Drainag e Engineering 141(5), 04 014069. Irmak, S., J. Burgert, M., S. Yang, H., Cassman, K., T. Walters, D., R. Rathje, W., Payero, J., Grassini, P., S. Kuzila, M., J. Brunkhorst, K., E. Eisenhauer, D., Kranz, W., VanDeWalle, B., M. Rees, J., Zoubek, G., Shapiro, C., Teichmeier, G., 2012 . Large Scale On Farm Implementation of Soil Moisture Based Irrigation Management Strategies for Increasing Maize Water Productiv ity. pp. 1 3. Irmak, S., Odhiambo, L., Kranz, W., E. Eisenhauer, D., 2011. Irrigation Efficiency and Uniformity, and Crop Water Use Efficiency. pp. 1 8. Israelsen, O.W., 1950. Irrigation principles and practices . New York, Wiley <1950> 2d ed. Jamison, V.C., Beale, O.W., 1958. Irrigation of corn in the Eastern United States. [electronic resource], Agriculture handbook / United Stat es Department of Agriculture: no. 140. [Washington, D.C.] : U.S . Dept. of Agriculture, [1958]. Jensen, M.E., Allen, R.G., 2016. Evaporation, Evapotranspiration, and Irrigation Water Requirements, Manuals and Reports on Engineering Practice: v. 70. Reston : American Society of Civil Engineers, 2016. 2nd ed., pp. 3 16.

PAGE 122

122 K State Research & Extension Mobile Irrigation Lab, 2014. "KanS ched" (v3.1.5). Kang, S., Gu, B., Du, T., Zhang, J., 2003. Crop coefficient and ratio of transpiration to evapotranspiration of wi nter wheat and maize in a semi humid region. Agricultural W ater Management 59(3), 239 254. Kisekka, I., Aguilar, J., Lamm, F., Rogers, D., 2014. Using Soil Water and Canopy Temperature to Improve Irrigatio n Scheduling for Corn. pp. 2 4. KMC, 2018. 6700 Ser ies Rip/Strip. L. Hatfield, J., Dold, C., 2018. Climate Change Impacts on Corn Phen ology and Productivity, p. 108. Lamm, F.R., Rogers, D.H., 2015. THE IMPORTANCE OF IRRIGATION SCHEDULING FOR MARGINAL CAPACITY SYSTEMS GROWING CORN. Applied Engineering in Ag riculture 31(2), 261 265. Lauer, J., 2002. Meth ods for Calculating Corn Yield. Mandal, K.G., 2001. Effect of irrigation regimes and nutrient management on soil water dynamics, evapo transpiration and yield of wheat ( Triticum aestivum) in Vertisol. Marco, M ., Nicola Dal, F., Lucia, B., Francesco, M., 2015. Effect of Incident Rainfall Redistribution by Maize Canopy on Soil Moisture at the Crop Row Scale. Water, Vol 7, Iss 5, Pp 2254 2271 (2015)(5), 22 54. Marella, R.L., Berndt, M.P., 2005. Water withdrawals a nd trends from the Floridan aquifer system in the southeastern United States, 1950 2000. [electronic resource], U.S. Geological Survey circular: 1278. Reston, Va. : U.S. Dept. of the Interior, U.S. Geological Survey ; Denver, CO : For sale by U.S. Geologic al Survey, Information S ervices, 2005., pp. 6 7, 20 24. Marella, R.L., Dixon, J.F., 2018. Data tables summarizing the source specific estimated water withdrawals in Florida by water source, category, county, and water management district, 2015: U.S. Geological Survey data release. Marella, R.L., Dixon, J.F., Berry, D.R., 2016. Agricultural irrigated land use inventory for the counties in the Suwannee River Water Management District in Florida, 2015, Open File Report. Reston, VA, pp. 18 19. Markowitz, E., 2013. Predicting the weather -down to the minute: the entrepreneurs behind Dark Sky, an app that tells you exactly when the rain will start. Mansueto Venture s LLC on behalf of Inc., p. 96. Mbabazi, D., Migliaccio, K.W., Crane, J.H., Fraisse, C., Zotare lli, L., Morgan, K.T., Kiggundu, N., 2017. An irrigation schedule testing model for optimization of the Smartirrigation avocado app. Agricult ural Water Management 179, 1 3. McClure, A., 2009. Planting Corn for Grain in Tennessee

PAGE 123

123 Migliaccio, K.W., Morga n, K.T., Vellidis, G., Zotarelli, L., Fraisse, C., Zurweller, B.A., Andreis, J.H., Crane, J.H., Rowland, D.L., 2016. SMARTPHONE APPS FOR IRRIGATION SCHEDULING. pp. 291 301. Migliaccio, K.W., Schaffer, B., Crane, J.H., Davies, F.S., 2010. Plant response to evapotranspiration and soil water sensor irrigation scheduling methods for papaya production in south Florida. Agricultural Water Management 97, 1452 1460. Mitra, S., Srivastava, P., Singh, S., 2016. Effect of irrigation pumpage during drought on karst aqu ifer systems in highly agricultural watersheds: example of the Apalachicola Chattahoochee Flint river basin, southeastern USA. Hydroge ology Journal 24(6), 1565 1582. Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models part I A discussion of principles. Journ al of Hydrology 10(3), 282 290. Neild, R.E., Newman, J.E., 1990. Growing Season Characteristics and Requirements in the Corn Belt. NWS, 2018. Experimental Forecast Reference EvapoTranspiration (FRET) Information. https://www.weather.gov/cae/fretinfo.html . Orthman, 2018. 1tRIPr. Peters, T.R., Hoogenboom, G., Hill, S., 2013. Simplified Irrigation Scheduling on a Smart Phone or Web Browser. pp. 1 34. Piccinni, G., Ko, J., Marek, T., Howell, T., 2009. Determination of growth stage specific crop coefficients (KC) of maize and sorghum. Agricultural Water Management 96, 1698 1704. Pioneer, 2018. M anaging Corn for Greater Yield. Porter, W., 2017. A Guide to Corn Production in Georgia. pp. 18 22 . Ritter, A., Muñoz Carpena, R., 2013. Performance evaluation of hydrological models: Statistical significance for reducing subjectivity in goodness of fit assessments. J ournal of Hydrology 480, 33 45. Scanlon, B.R., Faunt, C.C., Longu evergne, L., Reedy, R.C., Alley, W.M., McGuire, V.L., McMahon, P.B., 2012. Groundwater depletion and sustainability of irrgation in the US High Plains and Central Valley. Proceedings of the National Academy of Sciences of the United States of America 109(2 4), 9320 9325. Sentek Pty L td., 2016. Sentek Drill & Drop. SRWMD, 2010. Wate r Supply Assessment. pp. 42 45. The Balmoral Group, 2018. Florida Statewide Agricultural Irrigation Demand Estimated Agricultural Wate r Demand, 2015 2040. pp. 7 8. U.S. Climate D ata, 2018. Climate data for Live Oak, Fl. https://www.usclimatedata.com/climate/live oak/florida/united states/usfl0280 .

PAGE 124

124 USDA, 2018a. I rrigation & Water Use. pp. 1 2. USDA, 2018b. USDA. https://www.ers.usda.gov/topics/crops/corn/background/ . (Accessed 4/09/2018 2018). USDA NRCS, 2005. Chapter 9 Irrigation Water Management NJ652.09. pp. 15 16. US DA NRCS, 2006. TABLE NJ 3.4 Effective Root Zone Moisture Extraction Depth in Unrestricted Soils (Top 5 0% of the rootzone). USDA NRCS. USDA NRCS, 2008. Soil Quality Indicators. Vellidis, G., 2011. THE UNIVERSITY OF GEORG IA SMART SENSOR ARRAY. pp. 1 2. Velli dis, G., Liakos, V., Andreis, J.H., Perry, C.D., Porter, W.M., Barnes, E.M., Morgan, K.T., Fraisse, C., Migliaccio, K.W., 2016. Development and assessment of a smartphone application for irrigation scheduling in cotton. Computers and Electronics in Agricul ture 127 , 1 11. Wright, D., 2004. Field corn production guide. University of Florida Cooperative Extension Service, Institute of Food and Agricultural Sciences, EDI S,, Gainesville, Fla., pp. 1 4. Wright, D.L., Tillman, B., Small, I.M., Ferrell, J.A., Dufau lt, N., 2013. Management and Cultural Practices for Peanuts. pp. 1 2. Xue, Q., Marek, T.H., Xu, W., Bell, J., 2017. Irrigated Corn Production and Management in the Texas High Plains. Journal of Contemporary Water R esearch & Education 162(1), 31. Yang, H., Samani, B., Specht, J., 2017. CornSoyWater: Real Time Aid for Corn and Soybean Irrigation Decisions. pp. 1 3. Zamora Re, M., 2019. IRRIGATION AND NITROGEN BEST MANAGEMENT PRACTICES IN CORN PRODUCTION, Agricultural and Biological Engineering. University of Flor ida. Zamora Re, M., Dukes, M.D., 2017. Irrigation scheduling using real time soil moisture data in corn production, 2017 ASABE Annual International Meetin g. ASABE, St. Joseph, MI, p. 1. Zamora Re, M.I., Dukes, M.D., 2016. Corn Irrigation and Fertilize r Use BMPs versus Conventional Growing Practi ces, a Two Year Overview. p. 6. Zhang, M., Leon, C.d., Migliaccio, K., 2018. Evaluation and comparison of interpolated gauge rainfall data and gridded rainfall data in Florida, USA. Hydrological Sciences Journal 63(4), 561 582.

PAGE 125

125 BIOGRAPHICAL SKETCH Justice Diamond was born and raised in Tallahassee, F L . After graduating high school, Justice attended the University of Florida where he e arned a Bachelor of Science in agricultural operations m anagement with a c oncentration in sustainable crop production. During his time as an undergraduate, he worked on a research project that focused on maximizing terpene yields in slash pine. In addition, he worked in the precision agriculture lab and helped develop tools for early detection of citrus greening as well as methods to determine citrus and strawberry yields through the use of machine learning. It was during this time that he gained a passion for sustainable agriculture and the possibilities of implementing technolo gy to improve it. With the desire to continue his education, management in the D epartment of Agricultural and Biological Engineering, which focused on improving irrigation scheduling methods through th e use of a smartphone application for evapotranspiration based irrigation scheduling of field corn. This program combined his passions for sustainable agriculture and technology. The results demonstrated evapotranspiration based irrigation scheduling can a pply 43% and 56% less water than traditional practices, which can help conserve water and reduce the impacts of irrigation withdrawals on the Upper Floridan in agricultural o perations m anagement in the spring of 2019. Justice hopes to share and apply what he has learned to promote sustainable agriculture and resource conservation.