Evaluation and improvement of CROPGRO-soybean model for a cool environment in Galicia, northwest Spain

Abstract

This paper describes the evaluation and improvement of the CROPGRO-soybean model for a cool environment in northwest Spain. The model was evaluated with a soybean ( Glycine max L.) 1994 field data set with three cultivars and three planting dates. The original model proved to underestimate biomass and seed yield, so modifications were made to the temperature functions affecting N 2 fixation and photosynthesis, in order to fit better to the experimental data. The modified model was then tested, i.e., validated, with independent experimental data collected in 1995 at the same site with the same cultivars and four planting dates. Comparing observed with simulated data in 1994, the modified model decreased the root mean square error (RMSE) for biomass at harvest and seed yield from 1714 to 466 and from 940 to 333, respectively. For 1995, the validation year, RMSE for biomass decreased from 1366 to 352, although the yield was now overestimated with no significant change in RMSE. The average simulated harvest index (HI) for 1995 was greater than for 1994, the reverse of the measured values. Nevertheless, the modified model was more reliable in predicting crop performances under these cool conditions than the original model. Moreover, the predictions of the modified model for a warm climate (Gainesville, FL) were quite acceptable. We conclude that the modified model can be used successfully over a wider range of climates than the original version.