Uncertainty quantification and reliability analysis of CMIP5 projections for the Indian summer monsoon

摘要

A 'reliability ensemble averaging (REA)' technique is proposed to provide a quantitative estimate of associated uncertainty range and reliability of future climate change projections for Indian summer monsoon (June-September), simulated by the state-of-the-art Coupled General Circulation Models (CGCMs) under Coupled Model Intercomparison Project 5 (CMIP5). An evaluation of historical as well as future (RCP4.5 scenario) simulations of ten CGCMs in the REA technique projects a mean monsoon warming of 1.215 degrees C with an associated uncertainty range (+/- delta(Delta T)) of 0.22 degrees C, and an all-India precipitation increase by 7.109 mm/month with an associated uncertainty (+/- delta(Delta P)) of 2.592 mm/month for 2021-2050. REA technique also reflects a reduction in uncertainty range compared to simpler ensemble average approach and is characterized by consistently high reliability index in a comparative study with individual CGCMs. These results suggest the viability of REA methodology in providing realistic future Indian monsoon projections by incorporating model performance and model convergence criteria.