Thursday, November 8, 2007 - 10:15 AM
331-7
An Application of Transfer Functions and Generalized Likelihood Uncertainty Estimation Technique for DSSAT Models.
Crop models require knowledge of cultivar-specific traits, referred to as genetic coefficients , to predict daily crop growth and development as the plant responds to weather, soil characteristics and management practices. For new varieties, these genetic coefficients are often estimated manually by adjusting few of them so that predicted data fit observations, by fitting statistical models or through optimization by minimizing the error sum of squares between observations and predictions. Successful applications of crop models require that coefficients describing cultivars are available for current varieties grown by farmers in a timely manner. A new approach using maximum likelihood estimation was used to calculate probability density function for each of the six-genetic coefficients required by the DSSAT-Maize model. Results from several case studies showed that, even with a few years of yield data set, this technique produces robust parameter estimates while requiring far less time and model runs compared to other methods.
Audio File