Calibration of GREENLAB Model for Maize with Sparse Experimental Data

Abstract

Simplification of field measurement to reduce the time-consuming data collection for calibration is important to facilitate the application of the GREENLAB model. The effect of such simplifications on the accuracy of parameter values should be quantified in order to define to what extent simplifications are valid. This study introduced a new method for model parameter optimization with sparse data of maize using a multi-fitting technique, evaluated the effect of such simplifications on the parameter values, and validated the calibrated model with four independent field data sets. The results showed that coefficients of variance (CV) among different simplifications were below 15% for most parameter values. The parameter values of the beta function varied more compared with those of relative sink strength for different simplifications. Organ biomass under four different climate regimes was simulated based on parameter values optimized with a sparse dataset. Significant (P<0.05) deviations of simulation vs. observation correlations from the 1:1 relationship were only observed for internodes of second experiment in 2003. Thus, multi-fitting with sparse data can provide reasonable accuracy of parameter values.