Multi-model ensembles for regional and national wheat yield forecasts in Argentina

Abstract

While multi-model ensembles (MMEs) of seasonal climate models (SCMs) have been used forcrop yield forecasting, there has not been a systematic attempt to select the most skillful SCMs tooptimize the performance of a MME and improve in-season yield forecasts. Here, we propose astatistical model to forecast regional and national wheat yield variability from 1993–2016 over themain wheat production area in Argentina. Monthly mean temperature and precipitation from thefour months (August–November) before harvest were used as features. The model was validatedfor end-of-season estimation in December using reanalysis data (ERA) from the European Centrefor Medium-Range Weather Forecasts (ECMWF) as well as for in-season forecasts from June toNovember using a MME of three SCMs from 10 SCMs analyzed. A benchmark model forend-of-season yield estimation using ERA data achieved a R2 of 0.33, a root-mean-square error(RMSE) of 9.8% and a receiver operating characteristic (ROC) score of 0.8 on national level. Onregional level, the model demonstrated the best estimation accuracy in the northern sub-humidPampas with a R2 of 0.5, a RMSE of 12.6% and a ROC score of 0.9. Across all months ofinitialization, SCMs from the National Centers for Environmental Prediction, the National Centerfor Atmospheric Research and the Geophysical Fluid Dynamics Laboratory had the highest meanabsolute error of forecasted features compared to ERA data. The most skillful in-season wheat yieldforecasts were possible with a 3-member-MME, combining data from the SCMs of the ECMWF,the National Aeronautics and Space Administration and the French national meteorologicalservice. This MME forecasted wheat yield on national level at the beginning of November, onemonth before harvest, with a R2 of 0.32, a RMSE of 9.9% and a ROC score of 0.7. This approachcan be applied to other crops and regions.