Fig. 3: Comparison of spatial patterns of predictability scores between observations and CMIP6 models.
From: Climate-driven changes in the predictability of seasonal precipitation
Pattern correlation coefficients of the map of predictability score (\({{{{{\rm{PC}}}}}}{{{{{{\rm{C}}}}}}}_{{{{{{\rm{NSE}}}}}}}\)) between observations and 26 CMIP6 models for each season. In each model, the map of predictability score for each ensemble member is obtained from the linear model that uses the best 2 principal components (PCs) as predictors for each grid point. The bars and numbers on top represent the multi-ensemble mean (\({\overline{{{{{{\rm{PCC}}}}}}}}_{{{{{{\rm{NSE}}}}}}}\)) and the number of ensemble members (\({n}_{e}\)) of each model, respectively. For models that \({n}_{e}\ge 3\), vertical lines indicate one standard deviation of \({{{{{\rm{PC}}}}}}{{{{{{\rm{C}}}}}}}_{{{{{{\rm{NSE}}}}}}}\). Grey shading indicates the best 10 performing models (boldface, marked in boxes) that were selected for analyses of future changes.
