Multi-Model Forecast Quality Assessment of CMIP6 Decadal Predictions

Abstract Decadal climate predictions are a relatively new source of information for interannual to decadal time scales, which is increasing interest users. Forecast quality assessment essential identify windows opportunity (e.g., variables, regions, and forecast periods) with skill that can be used develop services inform users in several sectors define benchmarks improvements systems. This work evaluates the multi-model forecasts near-surface air temperature, precipitation, Atlantic multidecadal variability index (AMV), global temperature (GSAT) anomalies generated from all available retrospective contributing phase 6 Coupled Model Intercomparison Project (CMIP6). The generally show high predicting AMV, GSAT, while more limited precipitation. Different approaches generating compared, finding small differences between them. ensemble also compared individual best system usually provides highest skill. However, reasonable choice not having select each particular variable, period, region. Furthermore, historical simulations estimate impact initialization. An added value found ocean land regions it reduced Moreover, full subensemble measure size. Finally, implications these results context, requires issued near–real time, discussed.