Several research groups have developed crop models to evaluate various aspects of the crop system. This chapter will be a brief overview of the various applications of those models.

MANAGEMENT AND CROPPING STRATEGIES

Alocilja and Ritchie (1988) developed a software package for agricultural extension workers, researchers and government policy makers to use in the design and management of the rice-production system. Using the CERES-Rice model in conjunction with the simulation multicriteria optimization technique (SMOT), these individuals can be assisted in decision making for agrotechnology transfer.

Using data from South Africa, deVos and Mallet (198-) evaluated CERES-Maize and CORNF for assessing cropping strategies. They found that the models agree well with observed values of total above-ground plant dry mass, leaf area index, grain yield and soil water content. They also found that CORNF tended to underpredict leaf area index, and CERES-Maize provided realistic estimations of soil water content.

PREDICTING YIELD

"Tom Hodges (University of Missouri), S.K. Leduc (NOAA/NESDIS) and A. Eddy (Oklahom Climate Survey) used the CERES-Maize model to forecast yields during the 1983 drought. The researchers used actual weather conditions up to July 1, but simulated the weather for the remainder of the growing season based on historical data. The model was run at 51 first order stations in 14 midwestern states.

Hodges et al. (1987) tested the ability of CERES-Maize to predict annual fluctuations in maize (zea mays L.) production in the U.S. cornbelt for a four-year period, 1982-85. Results indicate the model may be used for large area yield and production estimation in the U.S. with minimal regional calibration. It also has potential for large area yield estimation in other parts of the world where daily maximum and minimum temperature, precipitation and solar radiation data are available.

Botner et al. (1986) also used CERES-Maize in projecting corn yield in the U.S. cornbelt. Test results indicate that using simulation models in an operational context for large-scale crop yield estimation is feasible and can be accomplished within reasonable cost if historic data and calibrated inputs to the model parameters have been determined.

Fei, Qing-Pei and Ripley (1985) used the CERES-Wheat model in southern Saskatchewan as an operational tool to predict yields. They initially tested the model with 25 years of historical data from the Saskatoon crop district. After correcting for a "technology trend" of 32 Kg/ha per year, they found that the CERES model was capable of simulating yields for 1960-84 with a correlation coefficient of 0.70. They also found that the main weakness in the model is over estimation in high yield years and under-estimation in low-yield years. This problem appears to be caused by the model's inability to simulate adequate root growth.

Larsen (1981) worked with CERES-Wheat and TAMW to determine whether a plant growth simulation approach could be successfully used to make large-area yield forecasts for winter wheat. Results showed that many plant responses are simulated well but improvement is needed in several areas before accurate yield forecasts can be realized.

"A. Van Dyk and Tom Hodges (University of Missouri) used CERES-Maize in a study that linked satellite information with crop simulation models. The study utilized satellite data to provide inputs on leaf area index, solar radiation, and crop emergence. Actual yields were compared to yields produced by CERES-Maize.

"W.R. Berti, D.L. Karlen and J.E. Pasons (all USDA-ARS, Florence, SC) used CERES-Maize to compare measured and predicted corn yields as affected by soil series. Results indicate that predicted values on corn grain yields ranged from 95% to 19% greater than measured values for six soil types included in this study. They recommend that site-specific data on initial soil water and depth of rooting could be used to improve predictions.

"E.L. Piper and A. Weiss (both University of Nebraska) are evaluating the response of a structured version of CERES-Maize to reduction in plant population or defoliation at various life stages. CERES-Maize projections were as much as 38% less than actual results when the population was reduced during vegetative growth. The model over-predicted kernel number and under-predicted kernel weight, except at 100% defoliation during vegetative growth. The authors suggest that the relationship between carbon redistribution during vegetative growth after defoliation, and the prediction of kernel number should be investigated.

"William Iwig and Benjamin Klugh, Jr. (both USDA Statistical Reporting Service, Washington, D.C.) have evaluated two feedback versions of the CERES-model. The first version forces the modeled values of leaf number and vegetative biomass to statistically match measurements of the observed corn crop made on any day prior to tasseling. The second version performs additional adjustments to the modeled yield components of kernels per plant and kernel weight based on feedback data. Analyses using six test data sets indicated that only the second feedback version produced significantly improved estimates of yield and kernel weight, and that neither version produced improved estimates of kernels per plant.

"Keating, et. al, have been adapting the CERES-Maize model for use in eastern Kenya where rainfall is low and unreliable. CERES-Maize was used to examine the effects of plant population on the long-term returns and risks of maize production at two sites in Kenya with different levels of soil fertility. Results indicate that high plant population would tend to increase long-term average yields in areas with non-limiting soil fertility. However, where nitrogen is strongly limiting, high plant populations were expected to reduce long-term average yields and increase the risks of crop failure. The model provided an accurate description of grain yield, but simulation of above ground biomass was less accurate."

"Carberry, et al. have been validating CERES-Maize in semi-arid, tropical environments in Australia, and have been adapting it for use in such areas. The model initially did not predict yields accurately in the Northern Territory region of australia. Functions of the model describing phenology, leaf growth and senescence, and grain growth were later revised and validated. Subsequent calibration reduced the root mean square deviation for observed grain yields from 3480 to 2015 kg/ha."

"Mulliken has reported successful results in using the CERES-Maize model to predict yields. Mulliken reported that the model predicted 116 bushels per acre (bu/a) for a dryland corn crop and the actual yield was 113 bu/a. In another instance where the corn was irrigated, CERES-Maize predicted yields of 198 bu/a and actual yields were 194 bu/a.

PREDICTING CLIMATE IMPACTS ON GROWTH AND YIELD

"A.L. duPisani used CERES-Maize as a drought prediction tool. The model is being used to assess drought impacts on maize at an early stage so that policy makers can have an objective measure to declare drought-stricken areas. To assess the impact of early season weather on crop yields, the model was run with actual weather data up to a given date. Historic data were used for the remainder of the growing season. Excellent correlations were found between yield predictions and actual yields."

DROUGHT INDEXING

Booysen (1987) used wheat growth simulation models (the CERES model, Ritchie 1985; the Utah model; the Kanemasu model, Kanemasu, et al.; and TAMW) to determine their effectiveness in deriving a crop specific drought index (CSDI). Featuring the environment and the crop, i.e. transpiration, root absorption, phenological growth stages, and the climatic and soil moisture variables, the CSDI performed better than the Palmer drought severity index by a small margin. A simpler empirical model may have achieved the same results as the more complex models, but calibration and adaptation to the area would most likely have had to be performed.

IRRIGATION

Algozin (1986) used CERES-Maize to identify irrigation scheduling strategies for efficient use of natural resources and for economic profitability. His results show that economic efficiency was achieved at high application rates while irrigation water productivity was highest at small application rates.

Berrada (1983) used the CERES-Wheat model to study continuous wheat and fallow-wheat cropping systems and then compared no-till and conventional tillage within each system. Higher yields and lower incidence of failure were achieved with no-till than with conventional tillage in continuous wheat. In fallow wheat however, these two tillage methods performed similarly. According to Berrada, the simulation results confirm the hypothesis that with a no-till management system, continuous winter wheat can be a feasible alternative to fallow-wheat in western Nebraska.

Worman et al. used the CERES-Maize model to compare dryland and fallow cropping and various levels of irrigation. The model was validated by simulating yields for field experiments performed over an 8-year period. They found the simulated yields were close to actual yields.

"Boggess and Ritchie used CERES-Maize to link agronomic crop response information with economic and financial data in economic analysis and risk assessment of alternative irrigation strategies. CERES-Maize was used to generate yield predictions and alternative irrigation strategies. Alternate irrigation strategies were ranked on the basis of net returns and risk. Boggess summarized that the optimum management strategy depends on whether the producer desires to maximize yields, profits, or utility."

DRAINAGE

Brink (1985) linked a CERES crop model with a water management model DRAINMOD and created a model that simulates crop growth and yield as well as handles water balance for a water management system. A better description of the effective root zone depth was achieved because the roots responded realistically to soil water conditions existing in drainage. Also, by adding a supplementary method that accounts for the presence of a water table, he obtained an improved evapotranspiration routine.

WATER FLOW AND SOLUTE TRANSPORT

"K.W. Molten, J.C. Parker, T.B. Brumback, J.C. Baker and E.W. Carson (all of Virginia Tech University) have developed an enhanced version of CERES-Maize by adding an adapted version of the RHIZOS portion of the cotton growth model GOSSYM. RHIZOS uses a two-dimensional transport approach to simulate the movement of soil water and nitrate. The enhanced model allows the user to select a water balance or a two-dimensional transport approach to the movement and plant uptake of water and nitrogen."

NITROGEN UPTAKE

"Jones and Kiniry evaluated the performance of the CERES-Maize model from several locations and disciplines over the past six years. Model predictions were compared with measured values for experimental data from many regions. The nitrogen-limiting version of the model produced realistic simulations of the effects of nitrogen on biomass, total nitrogen uptake, grain nitrogen concentrations, total nitrogen in the grain and grain yield. The non-nitrogen version produced estimates of maximum leaf area index, maximum above ground biomass grain numbers, and grain yields which had highly significant correlations with measured values."

FERTILIZER

"C.A. Baanante, D.C. Godwin and J.T. Ritchie used the CERES-Maize and CERES-Wheat models coupled with a stochastic weather generator to simulate responses to various fertilizer strategies in Australia and Benin, West Africa. For each strategy, 50 crops were simulated. Strategies were compared on the basis of yield, total added return, and net benefits of fertilizer. Results indicated substantial differences between strategies in Benin, but only small benefits in Australia. Godwin and Ritchie also tested the sensitivity of the nitrogen component of the CERES-Maize model by two methods. First, 1-year field experiments were simulated and the impact of changing nitrogen management variable (fertilizer rate, timing placement, depth, and source on yield and nitrogen uptake) was examined. Second, climatic data were used to simulate the growth of 50 crops. analyses indicated CERES-Maize exhibited high sensitivity to climatic variables and some soil variables, but less sensitivity to nitrogen transformation rate variables.

ROOT GROWTH

Bland (monograph chapter) and Jones, Bland, Ritchie and Williams (monograph chapter) used models to simulate root system growth. The models flexibility and responsiveness enabled them to simulate root growth in a variety of soils, climates and species, and generated root length density profiles similar to those documented in the literature.

PEST MODELING

Muchen (Ph.D. thesis, 988) used the CERES-Maize model to learn more about the ecology and management of a major nematode pest(Pratylenchus Zeae) of maize in Zimbabwe. The model predicted the population density of P. Zeae in maize roots with a mere error of 7%; it was sensitive to weather data and to different initial population densities of P. Zeae in the soil; it predicted the correct silking date of maize variety R215 and above- and below-ground dry biomass with mean errors of 17.7% and 11.1%. The research showed the model could be incorporated in future predictive P. Zeae maize yield and crop loss assessments.