Article Request Page ASABE Journal Article
The Impact of Spatial Soil Variability on Simulation of Regional Maize Yield
V. Sharda, C. Handyside, B. Chaves, R. T. McNider, G. Hoogenboom
Published in Transactions of the ASABE 60(6): 2137-2148 (doi: 10.13031/trans.12374). Copyright 2017 American Society of Agricultural and Biological Engineers.
Submitted for review in March 2017 as manuscript number NRES 12374; approved for publication as part of the “Crop Modeling and Decision Support for Optimizing Use of Limited Water” collection by the Natural Resources & Environmental Systems Community of ASABE in August 2017.
The authors are Vaishali Sharda, ASABE Member, Postdoctoral Fellow, Nebraska Water Center, University of Nebraska, Lincoln, Nebraska; Cameron Handyside, Research Engineer, Earth System Science Center, University of Alabama, Huntsville, Alabama; Bernardo Chaves, Associate in Research, Irrigated Agriculture Research and Extension Center, Washington State University, Prosser, Washington; Richard T. McNider, Distinguished Professor of Science, Earth System Science Center, University of Alabama, Huntsville, Alabama; Gerrit Hoogenboom, ASABE Member, Professor, Department of Agricultural and Biological Engineering, University of Florida, Gainesville, Florida. Corresponding author: Vaishali Sharda, 152 E.J. Frick Dr., Manhattan, KS 66503; phone: 510-305-4262; e-mail: email@example.com.
Abstract. The study of climate variability and its impacts on crop production has become a continuous effort for the scientific community over the past two decades. However, the impact of spatial soil variability along with climatic factors on crop yield remains uncertain. The objective of this study was to determine the impact of soil and climatic variability on maize yield. We used Alabama as a case study because the agriculture is predominantly rainfed and there is a large variability in growing season precipitation due to the influence of climate variability signals such as the El Niño Southern Oscillation (ENSO). The cropping system model CERES-Maize of the Decision Support System for Agrotechnology Transfer (DSSAT) was used to simulate growth, development, and grain yield for maize for the top ten maize-producing counties in Alabama under rainfed conditions during dry and wet ENSO years. Maize yield simulations were compared for one prominent agricultural soil in each county, the top three prominent agricultural soils in each county, and spatially distributed SSURGO soils in each county. Simulated yields were then compared with maize yields reported by the National Agricultural Statistical Services (NASS). The simulation results showed that maize yield was impacted by both climate variability and spatial soil variability. Statistical relationships were established between crop yield, yield changes, and soil properties. This simulation study established the clear importance of soil variability in crop-climate impact studies.
Keywords.Crop Modeling, DSSAT, Database, Soil properties, Spatial variability.
Crop simulation models are valuable tools that use weather and soil data along with management and crop genetics to simulate crop growth, development, and yield. In general, they can capture the spatial variability in soils, crop management practices, and other factors that influence yield. Crop models have been used extensively to complement in-field research and to interpret the responses of agronomic systems under varied crop management situations and diverse environmental conditions (Tsuji et al., 1998). Several crop models use multipart methodologies to simulate environmental processes and increase the understanding of various cropping systems and underlying physical mechanisms (Bouman et al., 1996). Detailed data are needed for the simulation of agricultural systems to make interpretations of model-based experiments (Jones et al., 2017). Although crop models have been around for a while, there has been an increased interest in using crop models in the past decade to face the challenges of global food security and to study sustainable agricultural production systems under the stresses of increasing world population and climate change (Paola et al., 2016). Crop models, with historical weather data as input, have also been used to study the impact of the El Niño Southern Oscillation (ENSO) on yields of various crops (Hansen et al., 1998; Garcia y Garcia et al., 2006). These models are increasingly used as sources of agricultural information at different temporal and spatial scales (Hartkamp et al., 2009; Hoogenboom, 2000). However, the original models were designed to operate at a point scale (Hansen and Jones, 2000; Faivre et al., 2004; Stone and Meinke, 2005). A scientific understanding is often needed for impact assessment studies at regional, national, or global scales (Resop et al., 2012; deWit et al., 2010). Spatial variability of soil properties, topography, and climate, among other factors, leads to variability in crop production systems. This variability needs to be addressed to achieve the ultimate goal of sustainability in agricultural production (Basso et al., 2001). The concerns that exist about using crop models for assessing regional agricultural impacts (Palosuo et al., 2011) are mainly due to the accuracy of input data, especially at a spatial scale, along with a suitable representation of crop physiological processes and model parameters, which are the key factors for a simulation that can be used for providing alternative scenarios for decision making.
Several studies have indicated that soil properties have a significant impact on the spatial variability of yield (Batchelor and Paz, 1998). Crop growth is sensitive to both spatial (soil properties) and temporal (precipitation and temperature) factors (Anwar et al., 2009). Past studies have shown that the spatial scale of soils had a larger effect than climate scenarios on the spatial variability of yield (Mearns et al., 2001). Tremblay et al. (2011) reported that maize yield in wet climates is generally higher on coarse-textured soils as compared to fine-textured soils. In dry climates, higher crop yields can often be observed in clayey soils (because of higher water-holding capacity) than in sandy soils (Armstrong et al., 2009). Depending on precipitation, maize yield has been reported to be related to the clay content of the soil (Shahandeh et al., 2011). Over the years, several studies have also been conducted to study the impact of climate and other environmental factors on maize yield (Mera et al., 2006; Schlenker and Roberts, 2009). Most of these studies have been conducted at field scale, where the variability of soil parameters is not significant, in comparison to the spatial variability of larger areas or spatial scales. However, the impact of spatial soil variability and climate variability on maize yield at a regional scale remains to be determined. Therefore, understanding the spatial variability of soil parameters across crop fields, along with climate impacts, is vital for improving the application of agricultural inputs and crop yields and for improving the performance and effective utilization of crop simulation models at larger spatial scales.
The Decision Support System for Agrotechnology Transfer (DSSAT) (Hoogenboom et al., 2010; Jones et al., 2003) is one of the most widely used crop simulation models (Tsuji et al., 1998; Thorp et al., 2008) for evaluating agricultural management options. DSSAT version 4.5 comprises models for more than 28 crops that simulate crop growth, development, and yield along with management strategies that involve irrigation, fertilizer application, crop rotations, and others. DSSAT has been widely used to simulate crop water use and yield along with the development of management strategies under different soil and climate conditions (Liu et al., 2011; Soler et al., 2011; McNider et al., 2015). DSSAT has also been employed at various scales to simulate the impact of climate change on crop production (Carbone et al., 2003; Tubiello et al., 2002), to study the impact of climate variability on agriculture (Persson et al., 2009; Andresen et al., 2001), and to forecast yield (Soler et al., 2007; Bannayan et al., 2003). Most crop simulation models, such as DSSAT, require detailed soil and weather data and crop management information as inputs. The minimum input data for DSSAT include the soil profile (water-holding characteristics, texture, slope, nitrogen, organic matter, etc.), daily weather data (minimum and maximum temperature, precipitation, and solar radiation), crop management data (planting, plant population, row spacing, fertilizer application, etc.), and cultivar coefficients (Bao et al., 2017). Among these, soil information is one of the key inputs (Wu et al., 2010). However, extensive soil property information, such as soil surface information or detailed physical and chemical properties by soil horizon, is often difficult to obtain at regional scales. The detailed soil information required by a crop model is generally not provided by the various global soil databases that exist, such as the Harmonized World Soil Database (HWSD) (FAO, 2012) or the World Inventory of Soil Emission potentials (WISE) (Batjes, 2002). Another major hurdle in using a global soils database is that the raw data are not in an immediately usable format that is compatible with crop models, although some efforts have been made to convert global soil databases into usable crop model input (Gijsman et al., 2007; Romero et al., 2012). Given the spatial variability of soil, and the impact that soil properties, such as bulk density, clay content, texture, available soil moisture, organic matter (Wright et al., 1990), soil depth (Kreznor et al., 1989), and pH (Moore et al., 1993), have on crop yield, it is important to make use of detailed information on soil properties when using crop models at a regional scale.
Along with the major maize-producing states in the U.S. Corn Belt, Alabama contributes to making the U.S. the world’s largest producer and exporter of maize (NASS, 2015). Over the past few decades, maize yields in Alabama have been fluctuating, partly due to extreme weather events (NASS, 2015) and partly due to competition from the irrigated agriculture in the western part of the U.S. (McNider et al., 2015). Various studies have revealed how maize growth and yields have been impacted by climate variability, especially the El Niño Southern Oscillation (ENSO) (Sharda et al., 2012) in the southeastern U.S. (Hansen et al., 1998; Legler et al., 1999). ENSO is a coupled ocean-atmospheric phenomenon that occurs in the equatorial Pacific Ocean and can be categorized into three phases, i.e., El Niño with warm sea surface temperatures (SST), La Niña with cool SST, and neutral, based on an index derived from observed SST (Hansen et al., 1998). No specific research has studied the effect of climate variability along with spatial soil variability in maize yields at a regional scale and provided a solution for ENSO-impacted regions. Because soil variability can be very high in the spatial range of 0.1 to 10 km (Wassenaar et al., 1999), understanding its effect on crop yield is a critical component of site-specific management. One of the objectives of this study was to convert the soil profiles available through the Web Soil Survey (NCSS, 2013) into DSSAT crop model format for easy access and application. These soils were then used to investigate the effects of spatial soil variability on maize yields in Alabama. The specific question addressed in this article is how crop growth and yield are related to spatial soil distribution and climate variability. Therefore, the objectives of this study were to study the impact of spatial soil variability on maize yield at a regional scale along with the interaction with the climate variability signal ENSO.
Materials and Methods
DSSAT version 4.5, Crop Estimation through Resources and Environmental Synthesis (CERES) Maize (Ritchie et al., 1998; Jones et al., 1986), was used to simulate maize yields with a variety of climate inputs for the period 1982 to 2011. CERES-Maize calculates crop growth and simulates water and nitrogen balance at a daily time step by simulating processes of soil water, nutrient, and plant growth, along with developmental processes for the formation of final crop yield and yield components (Ritchie et al., 1998). The model simulates six phenological stages for a maize plant, with each stage controlled by weather factors, plant genetics, and other environmental factors such as water, sunlight, atmospheric gases, etc. Air temperature is used to simulate plant development through growing degree days (GDD), a measure of accrued growing season temperature greater than a defined baseline (8°C in CERES-Maize) (Glotter et al., 2016). Genetic coefficients are used to determine genotype-specific characteristics of maize growth. Leaf area index (LAI), radiation, factors for yield reduction due to non-optimum temperatures, and water stress factors are used to calculate potential dry matter production (Quirring and Legates, 2008). The simulations were conducted for the top ten maize-producing counties in Alabama (NASS, 2015), including Lawrence, Limestone, Colbert, Lauderdale, Madison, Jackson, DeKalb, Talladega, Morgan, and Marshall (fig. 1). The DSSAT v4.5 model was run for 30 years, from 1982 to 2011, in seasonal analysis mode (Thorton and Hoogenboom, 1994). Seasonal analysis can be used to compare the variability in crop performance based on different weather conditions and management practices over multiple years (Tsuji et al., 1998). Simulations were set up for the most dominant soil in each county, for the three most dominant soils in each county and for spatially distributed SSURGO soils in each county. The model was run for each of the ten counties for the entire 30-year study period. The annual average simulated yields were grouped according to ENSO phases as defined by the El Niño 3.4 index (Sharda et al., 2012) to study the impact of ENSO-induced climate variability on yields.
The input data required to run DSSAT include daily maximum and minimum temperatures, precipitation, and solar radiation, along with soil characterization data, management information (e.g., planting date, row spacing, and seeding density), and hybrid-specific genetic coefficients to simulate yield.
Daily maximum and minimum air temperature and precipitation data for a 30-year period of record (1982-2011) across Alabama were obtained from station data available from the National Climate Data Center (NCDC, 2012) via the Center for Ocean-Atmospheric Prediction Studies (COAPS, 2012). One weather station was selected for each county. Solar radiation was generated at each station using a weather generator (WGENR; Hodges et al., 1985) adjusted for the southeastern U.S. by Garcia y Garcia et al. (2008).
The Natural Resource Conservation Service (NRCS) National Cooperative Soil Survey (NCSS, 2013) Soil Survey Geographic (SSURGO) database provides soil data for more than 95% of the counties of the conterminous U.S. The NCSS soils database is a collaborative effort of many federal, regional, and state agencies along with private institutions across the U.S. All these agencies work together to inspect, record, categorize, understand, distribute, and publish information about U.S. soils. This database consists of land capability classification categories comprised of soil classes and subclasses as defined by the National Soil Survey Handbook (NSSH) (table 1). Based on these classes and subclasses, initial criteria were developed for selecting one dominant agricultural soil and the top three agricultural soils for the top ten maize-producing counties Alabama (table 2).
To account for the spatial soil variability (Sharda et al., 2013), the soil profiles for each county were selected using information provided by the USDA-NRCS website (http:// websoilsurvey.nrcs.usda.gov/app/WebSoilSurvey.aspx). All soil properties were contained in tabular data sets consisting of 56 tables with various data. The electronic database includes many soil properties, such as percentages of sand, silt, and clay, organic carbon content, color, drainage, slope, etc. (table 3). DSSAT requires detailed information on soil properties that includes soil layer information (depth and thickness, texture, bulk density, stone percentage, saturated hydraulic conductivity, etc.) as well as soil surface information (albedo, runoff curve number, soil fertility factor, and drainage coefficient). However, not all properties were available to meet the minimum data requirements for DSSAT (Hoogenboom et al., 2012), such as the lower limit of plant extractable water (SLLL), drained upper limit (SDUL), and saturated soil water content (SAT). These properties were estimated using methods provided in the DSSAT documentation (Gijsman et al., 2002). Several other soil parameters that were not available in the SSURGO database were set to their DSSAT default values, e.g., soil albedo (SALB) was set to 0.13 (brown). Three tables from the SSURGO tabular dataset were used for processing: the MapUnit table identifies the map units included in the referenced legend, the Component table lists the map unit components identified in the referenced map unit and selected properties of each component, and the Horizon table lists the soil horizons. Each of these tables has a unique key identifier, and all the tables are connected through “one-to-many” relationships. The unique key identifiers (chkey, cokey, and mukey) were used to query the data. The tabular data from SSURGO were then queried and joined with spatial data (fig. 2) using ArcGIS 10.1 (ESRI, 2012).
Table 1. Land capability classification categories (from NSSH Part 622; NCSS, 2013).
Capability class definition
Capability class is the broadest category in the land capability classification system. Class codes I, II, III, IV, V, VI, VII, and VIII are used to represent both irrigated and non-irrigated land capability classes.
Class I soils have slight limitations that restrict their use.
Class II soils have moderate limitations that reduce the choice of plants or require moderate conservation practices.
Class III soils have severe limitations that reduce the choice of plants or require special conservation practices, or both.
Class IV soils have very severe limitations that restrict the choice of plants, require very careful management, or both.
Class V soils have little or no hazard of erosion but have other limitations, impractical to remove, that limit their use mainly to pasture, range, forestland, or wildlife food and cover.
Class VI soils have severe limitations that make them generally unsuited to cultivation and that limit their use mainly to pasture, range, forestland, or wildlife food and cover.
Class VII soils have very severe limitations that make them unsuited to cultivation and that restrict their use mainly to grazing, forestland, or wildlife.
Class VIII soils and miscellaneous areas have limitations that preclude their use for commercial plant production and limit their use to recreation, wildlife, or water supply or for esthetic purposes.
Capability subclass definition
Capability subclass is the second category in the land capability classification system. Class codes e, w, s, and c are used for land capability subclasses.
Subclass e is made up of soils for which the susceptibility to erosion is the dominant problem or hazard affecting their use. Erosion susceptibility and past erosion damage are the major soil factors that affect soils in this subclass.
Subclass s is made up of soils that have soil limitations within the rooting zone, such as shallowness of the rooting zone, stones, low moisture-holding capacity, low fertility that is difficult to correct, and salinity or sodium content.
Subclass w is made up of soils for which excess water is the dominant hazard or limitation affecting their use. Poor soil drainage, wetness, a high water table, and overflow are the factors that affect soils in this subclass.
Subclass c is made up of soils for which the climate (the temperature or lack of moisture) is the major hazard or limitation affecting their use.
Table 2. Top ten maize-producing counties of Alabama with dominant agricultural soil types. County Soil 1 Soil 2 Soil 3 Lawrence Decatur silty clay loam Cumberland silty clay loam Etowah silty loam Madison Decatur silty clay loam Abernathy silt loam Dickson silt loam Lauderdale Dickson silt loam Decatur silty clay loam Choccolocco silt loam Limestone Cookeville silty loam Dickson cherty loam Decatur silt loam Colbert Decatur silt loam Chenneby silt loam Etowah silty clay loam Jackson Hanceville loam Lindside silt loam Holston sandy loam DeKalb Capshaw silt loam Hartsells sandy loam Leadville silty loam Talladega Decatur silty clay loam McQueen silt loam Wickham fine sandy loam Morgan Sequatchee sandy loam Waynesboro sandy loam Cumberland clay loam Marshall Hartsells sandy loam Monongahela sandy loam Albertville silty loam
The desired soil properties from the database were then exported to tables in Microsoft Excel to form input tables for SAS (SAS, 2010) code written to convert these profiles into DSSAT-compatible format. The input tables were preprocessed to account for some soil attributes that were not defined for certain soil profiles. The SSURGO profiles were then converted to DSSAT soil profiles (fig. 3) using the SAS code.
ArcMap 10.1 (ESRI, 2012) was used to create maize yield maps for each county. Crop data layer (CDL) (NASS, 2015) data for the top ten maize-producing counties in Alabama were downloaded, and the SSURGO spatial soil data layer was then clipped to the corn-growing areas in each county obtained from the CDL data. The yield maps were then created by assigning yield values obtained by running DSSAT for each of the soils present in each county.
Table 3. Soil properties available from NCSS database required for DSSAT (adapted from Wu et al., 2010). DSSAT
Definition Units SITE Site name - COUNTRY Country name - LAT Latitude degrees LONG Longitude degrees SLSOURCE Soils data source - SLTX Soil texture - SLDESCRIP Soil description - SCSFAM Soil family - SLDP Soil depth cm SLDR Soil drainage rate fraction per day SLRO Runoff curve number - SLPF Soil fertility factor 0 to 1 SLB Depth until base of layer cm SLCL Clay % SLSI Silt % SLOC Soil organic carbon concentration % SLHW pH in water - SCEC Soil cation exchange capacity cmol(+) per kg
Management and Planting Data
The crop management practices for the simulations were based on recommendations from the Alabama Cooperative Extension Service of Auburn University (Glass and van Santen, 2011). Rainfed practices were simulated for planting dates ranging from March 28 to March 31, depending on the location of the county. The planting dates were selected based on personal communications with extension experts at Auburn University and considering historic planting date information available on the NASS website. The plant population was 6.2 plants m-2, row spacing was 0.61 m, and 171 kg ha-1 of nitrogen fertilizer was applied as ammonium nitrate equally divided over three applications. Only weather and soils data were allowed to vary from one simulation to another in order to isolate the crop model response to weather and soil type.
Figure 2. Flowchart for conversion of SSURGO soils data into format compatible with DSSAT input files (adapted from Sheshukov et al., 2011). Figure 3. Example of DSSAT soil profile created from SSURGO data. This profile is for Durham sandy loam soil in Elmore county, Alabama.
Yield and phenology in CERES-Maize are determined by six genetic coefficients, and the process of calibration aims at obtaining rational estimates of these coefficients by comparing simulated data with observed data. The cultivar coefficients of the CERES-Maize model include thermal time from seedling emergence to the end of the juvenile phase (P1), the extent to which development is delayed for each hour increase in photoperiod above the longest photoperiod at which development proceeds at a maximum rate (P2), the thermal time from silking to physiological maturity (P5), the maximum possible number of kernels per plant (G2), the kernel filling rate during the linear grain filling state and under optimum conditions (G3), and the interval in thermal time (degree days) between successive leaf tip appearances (PHINT). Because the application of DSSAT was at a regional scale in this study, a general cultivar that characterizes the cultivars grown in Alabama was desired. A methodology was developed by McNider et al. (2015) to calibrate a maize cultivar for regional application of DSSAT in the southeastern U.S. Generalized likelihood uncertainty estimation (Beven and Binley, 1992) was used to modify the genetic coefficients to reduce the difference between simulated and observed yields and anthesis dates. Statistical parameters, including the coefficient of determination and root mean square error (RMSE), were used to calibrate McCurdy 84aa, a DSSAT default medium- to full-season maize cultivar, so that it would mimic the target cultivar. Dyna-gro 58K02 was selected as the target cultivar because it was being grown at the Tennessee Valley Research and Extension Center (TVREC) in Belle Mina, Alabama, which is operated by Auburn University Extension Service. The model was calibrated using six years (2004-2009) of field trial data for Dyna-gro 58K02 grown at TVREC and validated for the years 2000-2011. Statistically, the model performed well in simulating the measured maize yields (for a detailed description of the calibration procedure, refer to McNider et al., 2015). Calibrated values of the CERES-Maize model are given in table 4.
Table 4. Cultivar coefficients for the CERES-Maize model. Coefficient Definition Units Min. Max. Calibrated
P1 Thermal time from seedling emergence to end of juvenile phase. °C days 110 458 277 P2
Extent to which development is delayed for each hour increase in photoperiod above the longest photoperiod at which development proceeds at a maximum rate.
day h-1 0 3 0.048 P5 Thermal time from silking to physiological maturity. °C days 390 1000 673.7 G2 Maximum possible number of kernels per plant. kernel plant-1 248 990 683.1 G3 Kernel filling rate during the linear grain filling state and under optimum conditions. mg d-1 4.4 16.5 15.94 PHINT Interval in thermal time between successive leaf tip appearances. °C days 30 75 38.90
ENSO Phases and Niño 3.4 Index
The National Oceanic and Atmospheric Administration (NOAA) Niño 3.4 index was used for definition of the ENSO phase. This index is based on a three-month, extended reconstruction sea surface temperature analysis (ERSST.v2), which includes average SST anomalies in the Niño 3.4 region (5° N to 5° S and 170° W to 120° W) of the Pacific (Sharda et al., 2012). This region has been reported (Trenberth and Hoar, 1996) to be the main area where sea level pressure and temperature anomalies are well correlated. If the Niño 3.4 index is more than +0.5°C, the event is termed El Niño; if the index is less than -0.5°C, the event is termed La Niña. A neutral phase is defined when the Niño 3.4 index is between +0.5°C and -0.5°C. The NOAA Climate Prediction Center assigns the ENSO phase to a three-month period based on the SST anomaly, and this information is available on the NOAA website for 1950 to the present (NOAA, 2015). For ENSO phase assignment, we used growing season (March through October) ENSO information for 1982 to 2011, which was available through the NOAA Climate Prediction Center website. For the growing season, the ENSO phase over the entire period was considered. For example, for SST anomaly index values of +0.5°C or greater for a consecutive period of six months, the ENSO year of March through October was categorized as an El Niño year. Similarly, for SST anomaly values of -0.5°C or less, the year was classified as a La Niña year. If there was no phase change during these months, that phase was assigned to the year. Otherwise, that year was deleted from the analysis. Based on this method, the seasons for the entire period of 30 years (1982-2011) were classified as El Niño, La Niña, or neutral (table 5).
Table 5. Number of years in each El Niño Southern Oscillation (ENSO) phase during the 30-year study period (1982-2011). ENSO Phase Number of Years El Niño 6 La Niña 7 Neutral 12 Deleted 5
Statistical criteria commonly used for evaluating crop models (Bao et al., 2017; Anothai et al., 2008) were selected for this study. These include root mean square error (RMSE) and index of agreement (d) or d-stat (Willmott et al., 2012), given by equations 1 and 2, respectively:
where n is the number of observations, Pi is the simulated value for the ith measurement, Oi is the observed value for the ith measurement, and Pi' = Pi - O and Oi' = Oi - O, where O is the mean of all observations. The lower the RMSE value is and the closer the value of d-stat is to 1, the better the simulation.
To determine the differences in harvested yield within the different counties, climate, and soil conditions in the model application, we performed an analysis of variance (ANOVA) and least significant difference (LSD) using PROC GLM in SAS (SAS, 2010). Different years were considered repeats in this analysis of variance.
Results and Discussion
Simulated maize yields were compared with observed yields (table 6) for all ten counties and for the different soil inputs. In general, the average annual yield varied significantly (p < 0.05) when using one soil type, three top soils, and SSURGO soils for each county. The yield was highest in Limestone County (7,136 kg ha-1) and lowest in Marshall County. The difference between simulated and observed yields was less than 10% of the observed yield. The model overestimated maize yield for almost all simulations, which was expected because actual field limitations, such as weeds, insects, etc., are not modeled in DSSAT and CERES-Maize. The value of d-stat ranged from 0.67 to 0.98, indicating the high ability of the model in simulating yield under all given conditions. The RMSE ranged from 421 to 2,268 kg ha-1. The highest value of d-stat was 0.98, which was for SSURGO soil simulations in DeKalb and Talladega counties. For all ten counties, the use of spatially distributed SSURGO soils, instead of using the most prominent soil in the county or the top three agricultural soils in the county, produced better results.
Table 6. Average observed and simulated maize yield and statistics, d-stat and RMSE, for all the locations and different soil type simulations. County Observed
Simulated Yield (kg ha-1) d-stat RMSE 1 Soil/
SSURGO 1 Soil/
SSURGO 1 Soil/
SSURGO Lawrence 5963 7110 7093 6684 0.79 0.87 0.92 1909 1556 1197 Limestone 5858 6982 7136 6194 0.75 0.80 0.83 2268 2040 1777 Colbert 6278 6222 6637 7036 0.82 0.84 0.89 1706 1554 1292 Lauderdale 5684 6788 6492 6373 0.80 0.83 0.90 1731 1539 1205 Madison 5866 5973 6564 6496 0.71 0.67 0.84 1938 2130 1407 Jackson 5570 6018 5866 5819 0.86 0.96 0.92 2041 1071 1553 DeKalb 5455 5694 5539 5537 0.97 0.97 0.98 482 538 421 Talladega 5180 5702 5478 5460 0.96 0.97 0.98 778 647 537 Morgan 5427 5797 5867 5635 0.92 0.94 0.94 933 819 811 Marshall 5326 5445 5554 5716 0.80 0.86 0.93 1166 951 701
To further explore the findings of this study, a graphical analysis of simulated versus observed yields was conducted (fig. 4) for all counties. The yield averages for the entire period of study were analyzed for each county and for the three types of soil simulation. The number of soils in each county ranged from 7 to 36, with Jackson County having the most spatial soil variability and DeKalb having the least, with only seven different kinds of agricultural soils. An interesting observation from this analysis was that for most of the counties, especially the top maize-producing counties such as Lawrence and Limestone, using just one soil or the top three soils for the simulations resulted in higher overestimation of yield as compared to the simulations that used spatially distributed SSURGO soils, which were closer to the 1:1 line. The simulated yields for DeKalb, Talladega, and Morgan counties were very close to the observed yields for all soil types. For Lauderdale County, all soil conditions resulted in overestimation of maize yield, with the simulated yield for SSURGO soils being the least overestimated. For Madison County, all three types of soil simulation showed comparatively scattered yields with poor fit. Overall, using the SSURGO soils resulted in improved maize yield simulation for most of the counties based on the statistical parameters, i.e., index of agreement and root mean square error (table 6).
As stated previously, the ten counties that were included in this study have large spatial soil variability. Simulated maize yield maps for the study area were created using the maize crop mask (NASS, 2015) and SSURGO soil maps (NCSS, 2013) to visualize the impact of spatial soil variability on maize yield. An example for Lawrence County (fig. 5) illustrates the impact of soil variability on maize yield established in this study. The maize yield map shows that certain soils are more productive, with higher simulated maize yields (dark green), as compared to soils that result in very low yields (red).
ENSO Impact on Yield
The differences in simulated maize yield during the three ENSO phases for the 30 growing seasons (1982-2011) were statistically significant (p < 0.05) for each of the three simulation sets conducted using one soil for each county, three soils for each county, and all the SSURGO soils in each county when analyzed separately. Maize yields during El Niño (6,762 kg ha-1) years were, on average and for all three soil conditions, significantly higher (p < 0.05) than maize yields during La Niña (5,687 kg ha-1) and neutral years (6,176 kg ha-1) averaged over all the study locations. With all the maize production in these top ten maize-producing counties being rainfed, seasonal variations in precipitation and temperature that are attributable to ENSO impact maize yields. Because most of the major maize-producing counties in Alabama are concentrated in the northern part of the state (fig. 1), these results are in line with a study by Sharda et al. (2012), who found a high probability that El Niño brings wetter and cooler conditions during the maize growing season to this part of the state. Critical growth stages that have a greater water demand, such as flowering and silking (Mourtzinis et al., 2016), were most likely impacted and had higher yields during El Niño growing seasons when wet and cooler conditions prevailed. The influence of different weather patterns associated with ENSO could be seen in the simulated yields averaged across all locations for the simulations representing the spatially distributed SSURGO soils (fig. 6), showing inherent yield variability across the maize-producing region. During the 30-year period of the study, large year-to-year variations in maize yields were observed. The La Niña yield averaged 5,654 kg ha-1, ranging from 3,237 to 8,167 kg ha-1, whereas the El Niño average yield was nearly 1,650 kg ha-1 higher than the La Niña average yield. The effect of precipitation on maize yield was also impacted by soil type, e.g., the mostly silt clay loam soils in Lawrence and Limestone counties have better water-holding capacity as compared to the sandy soils of Talladega County. The differences in maize yield among El Niño, La Niña, and neutral years were statistically significant (p < 0.05) for each of the three soil conditions.
As an example, in Lawrence County, simulated maize yields were significantly higher during El Niño years than during neutral and La Niña years when using spatially distributed SSURGO soils (fig. 7). When using the three most dominant agricultural soils, El Niño yields were still higher than La Niña yields. However, when using just one soil for the entire county, the results contradicted this yield trend. This can be attributed to the fact that the most prominent soil in the county may not be the most productive soil. Similar results were found in most of the counties studied, with SSURGO soil simulations indicating higher El Niño yields than La Niña yields, which is in agreement with other studies conducted in the region (Mourtzinis et al., 2016; Persson et al., 2009). The rainfall variability associated with different ENSO phases impacts the soil moisture condition and, along with other weather variables, has a compounding effect on crop yield. These results highlight the importance of recognizing the interacting effects of climate variables and physical conditions, such as soils, rather than relying on only one source of yield variability in risk analysis or yield forecasting. Given the strong connection between ENSO phases and climate variables in the southeastern U.S., seasonal ENSO forecasts can be used in designing adaptation strategies to mitigate climate-induced effects on crop yields (Challinor et al., 2014). The results from this study demonstrate that maize production in the region could be better managed based on the projected ENSO phase. Several studies have highlighted similar benefits of using ENSO information to improve maize production in the region, e.g., selection of management practices, such as early or late planting dates, based on the ENSO forecast to avoid crop stress during important growth stages (Mourtzinis et al., 2016), and maize-based ethanol production (Persson et al., 2009), among others.
Figure 6. Maize yield for 1982-2011 averaged across the study locations using spatially distributed SSURGO soils. Horizontal red line represents average yield for El Niño years, green line represents average yield for neutral years, black line represents average yield for all years, and blue line represents average yield for La Niña years.
Summary and Conclusions
Figure 5. Spatial representation illustrating impact of soil variability on maize yield for Lawrence County, Alabama.
The DSSAT CERES-Maize model was successfully evaluated and validated to simulate maize yield responses to climate and soil variability in the top ten maize-producing counties in Alabama. The seasonal analysis feature of DSSAT was used to simulate maize yields for the 30-year study period to determine how yields are impacted by ENSO, an interannual climate variability signal prominent in the southeastern U.S. The model prediction accuracy was satisfactory across the years for the region studied, although the model generally overestimated maize yields for all counties and soil conditions. The effect of climate variability and ENSO phase on maize yield agreed with previous studies that have been conducted in the region and that have demonstrated the effects of climate variability on crop production in the southeastern U.S. (Hansen et al., 1998; Legler et al., 1999; Persson et al., 2009; Mourtzinis et al., 2016). The impacts of ENSO on maize yields in the region also suggest that using ENSO forecasts to predict maize yields could enhance maize production in the southeastern U.S. Because most agricultural land generally in the southeastern U.S., and specifically in northern Alabama, is rainfed, these results also highlight the relationship between crop yield and soil conditions.
Spatial variability of the model responses was shown to be strongly affected by the soil variability within the region when simulating yields for the 30-year study period. Using SSURGO soils as input to simulate maize yield was statistically better for almost all counties as compared to using the most dominant soil or using the top three soils for each county, which resulted in higher overestimations of yield. Maps created by overlapping soil type, maize mask, and yield simulations illustrated the spatial variability in maize yield that was attributable to soil variability. These maps allow us to examine yields in different areas and determine how soils and soil properties impact production. There is large spatial soil variation in the region, and because many agronomic decisions start with the soil, the results of this study can be helpful in meeting the challenges of sustainable maize production. The results from the soil variability study also indicate the possibility of increasing maize yield by assigning production to high-yielding soils in particular counties in the region.
The simulation of crops such as maize, which is extensively grown in the U.S. and around the world, with the seasonal analysis program of CERES-Maize version 4.5 can predict risks to crop production due to climate variability signals, especially ENSO. This study confirms that crop simulation models such as DSSAT, when properly calibrated and validated, can be used globally to adapt different crop management strategies based on ENSO forecasts. Overall, it can be concluded that the CERES-Maize model can be used to quantify the impacts of climate and spatial soil variability on maize yields and yield variability. This model can be useful in explaining the spatial variability of yield across a region and can be used by various agencies and stakeholders that need precise yield estimates before harvest. As more sources of fine-resolution soil data become available, there needs to be a focus on integrating soil information with studies that explore the impacts of regional climate variability and climate change on agricultural production. It is important to develop region-specific adaptation strategies based on this quantification of the impact of climate and spatial soil variability on maize yield. Another contribution of this study could be to develop the potential of higher-yielding soils in the region to maximize yields and minimize the yield variability attributable to local climate and soil properties.
Andresen, J. A., Alagarswamy, G., Rotz, C. A., Ritchie, J. T., & Lebaron, A. W. (2001). Weather impacts on maize, soybean, and alfalfa production in the Great Lakes region. Agron. J., 93, 1059-1070. https://doi.org/10.2134/agronj2001.9351059x
Anothai, J., Patanothai, A., Jogloy, S., Pannangpetch, K., Boote, K., & Hoogenboom, G. (2008). A sequential approach for determining the cultivar coefficients of peanut lines using end-of-season data of crop performance trials. Field Crop Res., 108(2), 169-178. https://doi.org/10.1016/j.fcr.2008.04.012
Anwar, M. R., O’Leary, G. J., Rab, M. A., Fisher, P. D., & Armstrong, R. D. (2009). Advances in precision agriculture in south-eastern Australia: V. Effect of seasonal conditions on wheat and barley yield response to applied nitrogen across management zones. Crop Pasture Sci., 60(9), 901-911. https://doi.org/10.1071/CP08351
Armstrong, R. D., Fitzpatrick, J., Rab, M. A., Abuzar, M., Fisher, P. D., & O’Leary, G. J. (2009). Advances in precision agriculture in south-eastern Australia: III. Interactions between soil properties and water use help explain spatial variability of crop production in the Victorian Mallee. Crop Pasture Sci., 60(9), 870-884. https://doi.org/10.1071/CP08349
Bannayan, M., Crout, N. M., & Hoogenboom, G. (2003). Application of the CERES-Wheat model for within-season prediction of winter wheat yield in the United Kingdom. Agron. J., 95(1), 114-125. https://doi.org/10.2134/agronj2003.0114
Bao, Y., Hoogenboom, G., McClendon, R., & Vellidis, G. (2017). A comparison of the performance of the CSM-CERES-Maize and EPIC models using maize variety trial data. Agric. Syst., 150, 109-119. https://doi.org/10.1016/j.agsy.2016.10.006
Basso, B., Ritchie, J. T., Pierce, F. J., Braga, R. P., & Jones, J. W. (2001). Spatial validation of crop models for precision agriculture. Agric. Syst., 68(2), 97-112. https://doi.org/10.1016/S0308-521X(00)00063-9
Batchelor, W. D., & Paz, J. O. (1998). Process-oriented crop growth models as a tool to evaluate spatial yield variability. Proc. 1st Intl. Conf. Geospatial Information in Agriculture and Forestry (vol. 1, pp. 198-205). Ann Arbor, MI: ERIM International.
Batjes, N. H. (2002). ISRIC-WISE global data set of derived soil properties on a 0.5 by 0.5 degree grid (ver. 2.0). Wageningen, The Netherlands: International Soil Reference and Information Centre (ISRIC).
Beven, K. J., & Binley, A. M. (1992). The future of distributed models: Model calibration and uncertainity prediction. Hydrol. Proc., 6(3), 279-298. https://doi.org/10.1002/hyp.3360060305
Bouman, B. A., Van Keulen, H., Van Laar, H. H., & Rabbinge, R. (1996). The “School of deWit” crop growth simulation models: A pedigree and historical overview. Agric. Syst., 52(2), 171-198. https://doi.org/10.1016/0308-521X(96)00011-X
Carbone, G. J., Mearns, L. O., Mavromatis, T., Sadler, E. J., & Stoksbury, D. (2003). Evaluating CROPGRO-Soybean performance for use in climate impact studies. Agron. J., 95(3), 537-544. https://doi.org/10.2134/agronj2003.0537
Challinor, A. J., Watson, J., Lobell, D. B., Howden, S. M., Smith, D. R., & Chetri, N. (2014). A meta-analysis of crop yield under climate change and adaptation. Nature Climate Change, 4, 287-291. https://doi.org/10.1038/nclimate2153
COAPS. (2012). Data. Tallahassee, FL: Center for Ocean-Atmospheric Prediction Studies. Retrieved from https://coaps.fsu.edu/data
deWit, A., Baruth, B., Boogaard, H., Diepen, K. V., Kraalingen, D. V., Micale, F., ... Wijngaart, R. V. (2010). Using ERA-INTERIM for regional crop yield forecasting in Europe. Climate Res., 44(1), 41-53. https://doi.org/10.3354/cr00872
ESRI. (2012). ArcGIS Desktop: Release 10.1. Redlands, CA: Environmental Systems Research Institute (ESRI).
Faivre, R., Leenhardt, D., Voltz, M., Benoit, M., Papy, F., Dedieu, G., & Wallach, D. (2004). Spatialising crop models. Agronomie, 24(4), 205-217. https://doi.org/10.1051/agro:2004016
FAO. (2012). Harmonized world soil database. Ver. 1.2. Rome, Italy: FAO. Retrieved from http://www.fao.org/soils-portal/soil-survey/soil-maps-and-databases/harmonized-world-soil-database-v12/en/
Garcia y Garcia, A., Guerra, L. C., & Hoogenboom, G. (2008). Impact of generated solar radiation on simulated crop growth and yield. Ecol. Model., 210(3), 312-326. https://doi.org/10.1016/j.ecolmodel.2007.08.003
Garcia y Garcia, A., Hoogenboom, G., Guerra, L. C., Paz, J. O., & Fraisse, C. W. (2006). Analysis of the inter-annual variation of peanut yield in Georgia using a dynamic crop simulation model. Trans. ASABE, 49(6), 2005-2015. https://doi.org/10.13031/2013.22281
Gijsman, A. J., Jagtap, S. S., & Jones, J. W. (2002). Wading through a swamp of complete confusion: How to choose a method for estimating soil water retention parameters for crop models. European J. Agron., 18(1-2), 75-105. https://doi.org/10.1016/S1161-0301(02)00098-9
Gijsman, A. J., Thorton, P. K., & Hoogenboom, G. (2007). Using the WISE database to parameterize soil inputs for crop simulation models. Comput. Electron. Agric., 56(2), 85-100. https://doi.org/10.1016/j.compag.2007.01.001
Glass, K. M., & van Santen, E. (2011). Evaluation of corn hybrids in Alabama. Auburn, AL: Auburn University Cooperative Extension Service.
Glotter, M. J., Moyer, E. J., Ruane, A. C., & Elliott, J. W. (2016). Evaluating the sensitivity of agricultural model performance to different climate inputs. J. Appl. Meteorol. Climatol., 55, 579-594. https://doi.org/10.1175/JAMC-D-15-0120.1
Hansen, J. W., & Jones, J. W. (2000). Scaling-up crop models for climate variability applications. Agric. Syst., 65(1), 43-72. https://doi.org/10.1016/S0308-521X(00)00025-1
Hansen, J. W., Hodges, A. W., & Jones, J. W. (1998). ENSO influences on agriculture in the southeastern U.S. J. Climate, 11, 404-411. https://doi.org/10.1175/1520-0442(1998)011<0404:EIOAIT>2.0.CO;2
Hartkamp, A. D., White, J. W., & Hoogenboom, H. (2009). Interfacing geographic information systems with agronomic modeling: A review. Agron. J., 91(5), 762-772. https://doi.org/10.2134/agronj1999.915761x
Hodges, T., French, V., & LeDuck, S. K. (1985). Estimating solar radiation for plant simulation models. AgRISTARS Tech. Report YM-15-00403, JSC-20239. Columbia, MO: National Oceanic and Atmospheric Administration.
Hoogenboom, G. (2000). Contribution of agrometeorology to the simulation of crop production and its applications. Agric. Forest Meteorol., 103(1-2), 137-157.
Hoogenboom, G., Jones, J. W., Traore, P. C., & Boote, K. J. (2012). Chapter 2: Experiments and data for model evaluation and application. In Improving soil fertility recommendations in Africa using the Decision Support System for Agrotechnology Transfer (DSSAT) (pp. 9-18). Dordrecht, The Netherlands: Springer. https://doi.org/10.1007/978-94-007-2960-5_2
Hoogenboom, G., Jones, J. W., Wilkens, P. W., Porter, C. H., Boote, K. J., Hunt, L. A., ... Tsuji, G. Y. (2010). Decision Support System for Agrotechnology Transfer (DSSAT) Ver. 4.5. Honolulu, HI: University of Hawaii. Retrieved from http://dssat.net
Jones, C. A., Kiniry, J. R., & Dyke, P. T. (1986). CERES-Maize: A simulation model of maize growth and development (1st. ed.). College Station, TX: Texas A&M University Press.
Jones, J. W., Antle, J. M., Basso, B. O., Boote, K. J., Conant, R. T., Foster, I., ... Wheeler, T. R. (2017). Brief history of agricultural systems modeling. Agric. Syst., 155, 240-254. https://doi.org/10.1016/j.agsy.2016.05.014
Jones, J. W., Hoogenboom, G., Porter, C. H., Boote, K. J., Batchelor, W. D., Hunt, L. A., ... Ritchie, J. T. (2003). DSSAT cropping system model. European J. Agron., 18(3-4), 235-265. https://doi.org/10.1016/S1161-0301(02)00107-7
Kreznor, W. R., Olson, K. R., Banwart, W. L., & Johnson, D. L. (1989). Soil, landscape, and erosion relationships in a northwest Illinois watershed. SSSA J., 53(6), 1763-1771. https://doi.org/10.2136/sssaj1989.03615995005300060026x
Legler, D. M., Bryant, K. J., & O’Brien, J. J. (1999). Impact of ENSO-related climate anomalies on crop yields in the U.S. Climatic Change, 42(2), 351-375. https://doi.org/10.1023/A:1005401101129
Liu, H. L., Yang, J. Y., Tan, C. S., Drury, C. F., Reynolds, W. D., Zhang, T. Q., ... Googenboom, G. (2011). Simulating water content, crop yield, and nitrate-N loss under free and controlled tile drainage with subsurface irrigation using the DSSAT model. Agric. Water Mgmt., 98(6), 1105-1111. https://doi.org/10.1016/j.agwat.2011.01.017
McNider, R. T., Handyside, C., Doty, K., Ellenburg, W. L., Cruise, J. F., Christy, J. R., ... Hoogenboom, G. (2015). An integrated crop and hydrologic modeling system to estimate hydrologic impacts of crop irrigation demands. Environ. Model. Software, 72, 341-355. https://doi.org/10.1016/j.envsoft.2014.10.009
Mearns, L. O., Easterling, W., Hays, C., & Marx, D. (2001). Comparison of agricultural impacts of climate change calculated from high and low resolution climate change scenarios: Part I. The uncertainty due to spatial scale. Climatic Change, 51(2), 131-172. https://doi.org/10.1023/A:1012297314857
Mera, R. J., Niyogi, D., Buol, G. S., Wilkerson, G., & Semazzi, F. H. (2006). Potential individual versus simultaneous climate change effects on soybean (C3) and maize (C4) crops: An agrotechnology model based study. Global Planet. Change, 54(1-2), 163-182. https://doi.org/10.1016/j.gloplacha.2005.11.003
Moore, I. D., Gessler, P. E., Nielsen, G. A., & Peterson, G. A. (1993). Soil attributes prediction using terrain analysis. SSSA J., 57(2), 443-452. https://doi.org/10.2136/sssaj1993.03615995005700020026x
Mourtzinis, S., Ortiz, B. V., & Damianidis, D. (2016). Climate change and ENSO effects on southeastern U.S. climate patterns and maize yield. Sci. Reports, 6, article 29777. https://doi.org/10.1038/srep29777
NASS. (2015). Crop production: 2014 summary. Washington, DC: USDA National Agricultural Statistics Service. https://www.usda.gov/nass/PUBS/TODAYRPT/cropan16.pdf
NCDC. (2012). National Climate Data Center. Asheville, NC: National Oceanic and Atmospheric Administration. Retrieved from http://www.ncdc.noaa.gov/oa/ncdc.html
NCSS. (2013). NCSS soil characterization database. Lincoln, NE: National Cooperative Soil Survey. Retrieved from http://ncsslabdatamart.sc.egov.usda.gov.
NOAA. (2015). El Niño Southern Oscillation: ENSO index. College Park, MD: National Oceanic and Atmospheric Administration, Climate Prediction Center. Retrieved from http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ensoyears.shtml
Palosuo, T., Kersebaum, K. C., Angulo, C., Hlavinka., P., Moriondo, M., Olesen, J. E., ... Rotter, R. (2011). Simulation of winter wheat yield and its variability in different climates of Europe: A comparison of eight crop growth models. European J. Agron., 35(3), 103-114. https://doi.org/10.1016/j.eja.2011.05.001
Paola, A. D., Valentini, R., & Santini, M. (2016). An overview of available crop growth and yield models for studies and assessments in agriculture. J. Sci. Food Agric., 96(3), 709-714. https://doi.org/10.1002/jsfa.7359
Paz, J. O., Batchelor, W. D., Babcock, B. A., Colvin, T. S., Logsdon, S. D., Kaspar, T. C., & Karlen, D. L. (1999). Model-based technique to determine variable-rate nitrogen for corn. Agric. Syst., 61(1), 69-75. https://doi.org/10.1016/S0308-521X(99)00035-9
Persson, T., Garcia y Garcia, A., Paz, J., Jones, J. W., & Hoogenboom, G. (2009). Maize ethanol feedstock production and net energy value as affected by climate variability and crop management practices. Agric. Syst., 100(1-3), 11-21. https://doi.org/10.1016/j.agsy.2008.11.004
Quirring, S. M., & Legates, D. R. (2008). Application of CERES-Maize for within-season prediction of rainfed corn yields in Delaware, USA. Agric. Forest Meteorol., 148(6-7), 964-975. https://doi.org/10.1016/j.agrformet.2008.01.009
Resop, J. P., Fleisher, D. H., Wang, Q., Timlin, D. J., & Reddy, V. R. (2012). Combining explanatory crop models with geospatial data for regional analyses of crop yield using field-scale modeling units. Comput. Electron. Agric., 89, 51-61. https://doi.org/10.1016/j.compag.2012.08.001
Ritchie, J. T., Singh, U., Godwin, D. C., & Bowen, W. T. (1998). Cereal growth, development, and yield. In G. Y. Tsuji, G. Hogenboom, & P. K. Thorton (Eds.), Understanding options for agricultural production (vol. 7, pp. 79-98). Dordrecht, The Netherlands: Springer. https://doi.org/10.1007/978-94-017-3624-4_5
Romero, C. C., Hoogenboom, G., Baigorria, G. A., Koo, J., Gijsman, A. J., & Wood, S. (2012). Reanalysis of a global soil database for crop and environmental modeling. Environ. Model. Software, 35(1), 163-170. https://doi.org/10.1016/j.envsoft.2012.02.018
SAS. (2010). SAS for Windows 9.1. Cary, NC: SAS Institute, Inc.
Schlenker, W., & Roberts, M. J. (2009). Nonlinear temperature effects indicate severe damages to U.S. crop yields under climate change. Proc. Natl. Acad. Sci., 106(37), 15594-15598. https://doi.org/10.1073/pnas.0906865106
Shahandeh, H., Wright, A. L., & Hons, F. M. (2011). Use of soil nitrogen parameters and texture for spatially variable nitrogen fertilization. Prec. Agric., 12(1), 146-163. https://doi.org/10.1007/s11119-010-9163-8
Sharda, V., Hoogenboom, G., Handyside, C., McNider, R. T., & Ortiz, B. (2013). Use of DSSAT crop model for assessment of spatial soil variability impact on maize yield at a regional scale. Pullman, WA: Washington State University Academic Showcase.
Sharda, V., Srivastava, P., Ingram, K. T., Chelliah, M., & Kalin, L. (2012). Quantification of El Niño Southern Oscillation impact on precipitation and streamflows for improved management of water resources in Alabama. J. Soil Water Cons., 67(3), 158-172. https://doi.org/10.2489/jswc.67.3.158
Sheshukov, A. Y., Daggupati, P., Douglas-Mankin, K. R., & Lee, M. C. (2011). High spatial resolution soil data for watershed modeling: 1. Development of SSURGO-ArcSWAT utility. J. Nat. Environ. Sci., 2(2), 15-24.
Soler, C. M., Bado, V. B., Traore, K., NcNair Bostic, W., Jones, J. W., & Hoogenboom, G. (2011). Soil organic carbon dynamics and crop yield for different crop rotations in a degraded ferruginous tropical soil in a semi-arid region: A simulation approach. J. Agric. Sci., 149(5), 579-593. https://doi.org/10.1017/S0021859611000050
Soler, C. M., Sentelhas, P. C., & Hoogenboom, G. (2007). Application of the CSM-CERES-Maize model for planting date evaluation and yield forecasting for maize grown off-season in a
subtropical environment. European J. Agron., 27(2-4), 165-177. https://doi.org/10.1016/j.eja.2007.03.002
Stone, R. C., & Meinke, H. (2005). Operational seasonal forecasting of crop performance. Phil. Trans. Royal Soc. B., 360(1463), 2109-2124. https://doi.org/10.1098/rstb.2005.1753
Thorp, K. R., DeJonge, K. C., Kaleita, A. L., Batchelor, W. D., & Paz, J. O. (2008). Methodology for the use of DSSAT models for precision agriculture decision support. Comput. Electron. Agric., 64(2), 276-285. https://doi.org/10.1016/j.compag.2008.05.022
Thorton, P. K., & Hoogenboom, G. (1994). A computer program to analyze single-season crop model outputs. Agron. J., 86(5), 860-868. https://doi.org/10.2134/agronj1994.00021962008600050020x
Tremblay, N., Bouroubi, M. Y., Vigneault, P., & Belec, C. (2011). Guidelines for in-season nitrogen application for maize (Zea mays L.) based on soil and terrain properties. Field Crop Res., 122, 273-283.
Trenberth, K. E., & Hoar, T. J. (1996). The 1990-1995 El Niño Southern Oscillation event: Longest on record. Geophys. Res. Letters, 23, 57-60. https://doi.org/10.1029/95GL03602
Tsuji, G. Y., Hoogenboom, G., & Thorton, P. (1998). Understanding options for agricultural production. Dordrecht, The Netherlands: Springer. https://doi.org/10.1007/978-94-017-3624-4
Tubiello, F. N., Rosenzweig, C., Goldberg, R. A., Jagtap, S., & Jones, J. W. (2002). Effects of climate change on U.S. crop production: Simulation results using two different GCM scenarios: Part I. Wheat, potato, maize, and citrus. Climate Res., 20, 259-270. https://doi.org/10.3354/cr020259
Wassenaar, T., Lagacherie, P., & Legros, J. P. (1999). Modeling wheat yield responses to soil and climate variability at the regional scale. Climate Res., 11, 209-220. https://doi.org/10.3354/cr011209
White, J. W., Hoogenboom, G., Stackhouse Jr., P. W., & Hoell, J. M. (2008). Evaluation of NASA satellite and assimilation model-derived long-term daily temperature data over the continental U.S. Agric. Forest Meteorol., 148(10), 1574-1584. https://doi.org/10.1016/j.agrformet.2008.05.017
Willmott, C. J., Robesonb, S. M., & Matsuuraa, K. (2012). Short communication: A refined index of model performance. Intl. J. Climatol., 32(13), 2088-2094. https://doi.org/10.1002/joc.2419
Wright, R. J., Boyer, D. G., Winant, W. M., & Perry, H. D. (1990). The influence of soil factors on yield differences among landscape positions in an Appalachian cornfield. Soil Sci., 149, 375-382. https://doi.org/10.1097/00010694-199006000-00009
Wu, W., Chen, J., Liu, H. G., & Hoogenboom, G. (2010). Parameterizing soil and weather inputs for crop simulation models using the VEMAP database. Agric. Ecosyst. Environ., 135(1-2), 111-118. https://doi.org/10.1016/j.agee.2009.08.016