Predicting crop yields under climate change conditions from monthly GCM weather projections

Estimation of changes in crop yields is currently based on projections of atmospheric General Circulation Models (GCM) and the use of crop simulators. Crop simulators require daily input of environmental variables. GCMs produce monthly projections of climatic variables. Our objective was to explore the possibility of using monthly weather projections in yield estimates. We considered atmospheric CO2 level, total solar radiation, average maximum and minimum temperature, and rainfall for five months of the growing season. The group method of data handling (GMDH) was applied to relate crop yields to these variables. Projections of GCMs were downscaled to provide daily weather variables for the Mississippi Delta, and weather patterns were obtained in 50 replications for each GCM. The soybean crop simulator GLYCIM was used to generate crop yields on sandy loam, loam and silt loam soils. The equations built with GMDH explained 81–85% of yield variability, and included solar radiation in July and August, CO2 level, minimum temperature in June and August, and rainfall in August.