Submitted to: Geospatial Information in Agriculture and Forestry International Conference
Publication Type: Proceedings
Publication Acceptance Date: June 1, 1998
Publication Date: N/A
Interpretive Summary: Site-specific farming is growing rapidly in almost every arena: research, new product development, user adoption, and education/extension. As more users request information, research and extension are pressed to deliver site-specific management recommendations for fertilizer, pesticide, and irrigation applications. Predicting outcomes for all possible weather, soil, and management options would appear to be a task ideally suited to computer simulation models of crop growth and yield. Several models currently appear to have the combination of distribution, support, testing, and ease of use to be suitable for site-specific applications. For this work, the sensitivity of the DSSAT/CERES-Maize model was tested for six years and several soil types typical of the SE Coastal Plain. Variables examined were selected because they appear to be important to explain yield variation (soil type, depth to clay layer, soil water, and crop temperature) or have been used for variable-rate management (nitrogen rate and seeding rate). The model did not match the observed soil type yields, nor the effect of depth to clay. It did respond to water content and crop temperature, but the inputs to the model may need modification because these variables are not currently handled in a way suited to site-specific farming. The model was not as sensitive as expected to nitrogen or population. The results suggest directions for model development that would improve applicability for site-specific agriculture.
Technical Abstract: When site-specific farming became technologically feasible, existing crop models made computer simulation a natural choice for predicting yield under various combinations of soil, weather, and management. However, modeling for site-specific farming may require both greater accuracy and sensitivity to more parameters than in current models. Results demonstrate that the corn model (DSSAT v3) appears less sensitive than expected to soil type, depth to clay, nitrogen, and plant population, suggesting areas where enhancements could be made. It appears appropriately sensitive to rainfall, indicating sensitivity to soil water content is generally correct. However, less sensitivity is shown by the curve number procedure to calculate runoff. The model also responds to maximum air temperature, but local variation in crop temperature exceeds variation in air temperature. Routines may be needed to accommodate within-field redistribution of runoff and to calculate crop temperature from water stress. Model accuracy issues aside, to accommodate spatial inputs and model runs will require enhanced interfaces. All these issues suggest enhancements for crop growth models that would improve applicability to site-specific agriculture.