Strong regional influence of climatic forcing datasets on global crop model ensembles

We present results from the Agricultural Model Intercomparison and Improvement Project (AgMIP) Global Gridded Crop (GGCMI) Phase I, which aligned 14 global gridded crop models (GGCMs) 11 climatic forcing datasets (CFDs) in order to understand how selection of climate data affects simulated historical productivity maize, wheat, rice soybean. Results show that CFDs demonstrate mean biases differences probability extreme events, with larger uncertainty around precipitation regions where observational for systems are scarce. Countries simulations correlate highly reported FAO national production anomalies tend have high correlations across most CFDs, whose influence we isolate using multi-GGCM ensembles each CFD. Correlations compare favorably signal detected other studies, although many countries is not primarily climate-limited (particularly rice). Bias-adjusted often were among highest model-observation correlations, all produced correlation at least one top-producing country. Analysis multi-CFD-multi-GGCM (up 91 members) shows benefits over use smaller subset some farming systems, bigger always better. Our analysis suggests assessments should prioritize based on multiple as long a top-performing CFD utilized focus region.