Abstract
The western Himalayan region (WHR) was subject to a significant negative trend in the annual and monsoon rainfall during 1902–2005. Annual and seasonal rainfall change over the WHR of India was estimated using 22 rain gauge station rainfall data from the India Meteorological Department. The performance of 13 global climate models (GCMs) from phase 3 of the Coupled Model Intercomparison Project (CMIP3) and 42 GCMs from CMIP5 was evaluated through multiple analysis: the evaluation of the mean annual cycle, annual cycles of interannual variability, spatial patterns, trends, and signal-to-noise ratio. In general, CMIP5 GCMs were more skillful in terms of simulating the annual cycle of interannual variability compared to CMIP3 GCMs. The CMIP3 GCMs failed to reproduce the observed trend, whereas approximately 50% of the CMIP5 GCMs reproduced the statistical distribution of short-term (30 yr) trend estimates than for the longer-term (99 yr) trends from CMIP5 GCMs. GCMs from both CMIP3 and CMIP5 were able to simulate the spatial distribution of observed rainfall in premonsoon and winter months. Based on performance, each model of CMIP3 and CMIP5 was given an overall rank, which puts the high-resolution version of the MIROC3.2 model [MIROC3.2 (hires)] and MIROC5 at the top in CMIP3 and CMIP5, respectively. Robustness of the ranking was judged through a sensitivity analysis, which indicated that ranks were independent during the process of adding or removing any individual method. It also revealed that trend analysis was not a robust method of judging performances of the models as compared to other methods.
A method for estimating and assessing modes of interannual variability in coupled climate models