Determination of cultivar coefficients of crop models using a genetic algorithm: A conceptual framework

Most. crop simulation models include parameters and coefficients that defin e responses and interactions with environmental and management conditions. This article presents a conceptual framework for using a genetic algorithm to determine cultivar coefficients of crop simulation models. The model's m ulti-dimensional outputs were transformed into a scaled and dimensionless f itness evaluation function based on the L-1-metric measure of vector distan ces between the simulated and observed output. The cultivar coefficients we re linearly mapped and encoded into genetic algorithm chromosomes. Empirica l evidence comparing the performance of the cultivar coefficients calculate d by the genetic algorithm to the existing cultivar coefficients that was c alculated by other methods was used to evaluate the performance of this app roach. The evidence suggests that the genetic algorithm may be as good as o r better than the traditional cultivar coefficient calculation method.