Skill of rice yields forecasting over Mainland Southeast Asia using the ECMWF SEAS5 ensemble prediction system and the WOFOST crop model
Ubolya Wanthanaporn, Iwan Supit, Winai Chaowiwat, Ronald Hutjes
Abstract
This study evaluates the potential use of European Centre for Medium-Range Weather Forecast (ECMWF) ensemble prediction system-5 (SEAS5) to force the WOrld FOod Studies crop model (WOFOST) for predicting rice production in Mainland Southeast Asia (MSEA). The assessment covers a 30-year period (1985–2014) by comparing yield using the SEAS5 weather data with benchmark yield simulation based on reference climate data from WATCH Forcing Data ERA-5 (WFDE5). Two cultivation simulations were used: a water- and nutrient-limited (WN-limited) simulation representing cultivation in the rainfed area, and a nutrient-limited (N-limited) simulation representing cultivation in the irrigation area. SEAS5 shows consistent yield prediction skills between the two simulations, suggesting that water availability is not the primary factor influencing yield forecasting performance. Therefore, rainfall forecasting skill is not the main source of yield prediction skill. However, other variables, especially temperature, influence the yield prediction skill. SEAS5 exhibits high performance in predicting rice yield from early planting in the main season, with the ability to capture anomalous rice yields and consistent accuracy with lead times of one to three months . SEAS5 skills are limited when the rice planting times are delayed by one or two months during the main season. Similarly, limited skill is observed in the dry season. SEAS5 demonstrate reliable performance for crop yield prediction at the beginning of the main season, which is potentially valuable for national-level strategies and planning.