Performance of the CMIP5 models in the simulation of the Himalaya-Tibetan Plateau monsoon

In this paper, the performance of 28 CMIP5 models in simulating the climate of the Himalaya-Tibetan Plateau (HTP) for the recent past (1975–2005) is evaluated using the observations from the Asian Precipitation Highly-Resolved Observational Data Integration Towards Evaluation (APHRODITE). Many models realistically simulate the spatial distribution of surface air temperature (T_as) and precipitation with pattern correlation as high as 0.8; however, they possess severe biases in their magnitude. The biases in T_as appear to be associated with the biases in the surface elevation. All the models capture the observed phase of the annual cycle of the T_as but underestimate the amplitude. For precipitation, the phase is captured by most models (except few), but the amplitude is overestimated in all models. In the mid-intensity precipitation range (10–80 mm day⁻¹), most of the models overestimate the probability of occurrence and show large intermodel differences. Most of the models fail to simulate the spatial distribution of the trend in T_as and precipitation. As compared to many individual models, the biases are noted to reduce when using multimodel means (MMMs); however, the MMMs also failed to capture the observed trends in both T_as and precipitation. Many models still struggle to capture the large-scale phenomena, such as the location and intensity of upper-level Asian anticyclone and middle troposphere temperature maximum over the HTP, which have large implications on the HTP as well as the Indian summer monsoon. The results show that none of the models capture all features of the HTP monsoon, and hence, further improvement in the parameterization schemes and resolution is required to gain more confidence in the projection of HTP climate using these models.