USING OPTIMIZATION TO ESTIMATE SOIL INPUTS OF CROP MODELS FOR USE IN SITE-SPECIFIC MANAGEMENT

Abstract

Predicting the spatial variability of grain yield is of crucial importance for site-specific management (SSM) becauseit allows for testing of management prescriptions and for correct assessment of agronomic and economic outcomes.One common limitation of crop simulation model use in SSM is the need for accurate values of many inputs from numeroussites in a field. Optimization can be of great help in the estimation of parameters using more easily measured variables suchas yield. We have used simulated annealing and compared parameter estimates and yield predictions resulting from the useof two distinct objective function variables: grain yield and soil-water content. Estimating site-specific soil parameters fromgrain yield measurements led to acceptable errors in grain yield estimates. However, soil-water was not accurately predicted,which made the strategy unreliable. The errors in soil-water were particularly high in the bottom soil layer. In addition, mostof the soil-water holding limits were not valid, especially for the lower limit and saturation. Estimating site-specific soil parametersfrom soil-water content measurements led to acceptable errors in grain yield and soil-water estimates. The estimatedsoil-water holding limits were valid with an exception of saturation for the intermediate soil layers.