Regional comparison of CMIP3 and CMIP5 precipitation characteristics

Abstract

The improvements in the regional precipitation characteristics over the land regions of globe from Coupled Model Intercomparison Project-3 (CMIP3) and Coupled Model Intercomparison Project-5 (CMIP5) model group are examined. This research employed 11 General Circulation Model (GCMs) from both generation of models and combined them from both Multi Model Averaging (MMA) and Bayesian Multi Model Averaging (BMA) to ascertain the upgradation in projections for both seasons. BMA method offer better alternative to MMA as higher weights are assigned to better performing models for more skillful regional projections. Our results elucidate the fact that both combinations of models and methods inform spatially similar non-uniform changes in projections of precipitation across these models. The models do not express much progress in mean simulations for both 20th and 21st century projections. Thus lower uncertainty for CMIP5 group than CMIP3 is observed. Therefore, the findings highlight that mere comparisons based on mean statistics alone may not be sufficient to understand their progress. However, advanced methods of analysis provide better insights in their model development.