Abstract
Future changes in the annual cycle of the precipitation in central Asia (CA) were estimated based on the historical and representative concentration pathway 8.5 (RCP8.5) experiments from 25 models of the Coupled Model Intercomparison Project phase 5. Compared with the Global Precipitation Climatology Project observations, the historical (1979–1999) experiments showed that most models can capture the migration of rainfall centres, but remarkable discrepancies exist in the location and intensity of rainfall centres between simulations and observations. Considering the skill scores of precipitation and pattern correlations of circulations, which are closely related to the precipitation for each month, for the 25 models, the four best models (e.g., CanESM2, CMCC‐CMS, MIROC5 and MPI‐ESM‐LR) with relatively good performance were selected. The four models' ensemble mean indicated that the migration and location of the precipitation centres were better reproduced, except the intensity of the centres was overestimated, compared with the result that only took the precipitation into consideration. Based on the four best models' ensemble mean under RCP8.5 scenarios, precipitation was projected to increase dramatically over most of the CA region in the boreal cold seasons (November, December, January, February, March and April) and May, with the maximum in December at the end of 21st century (2079–2099), and several positive centres were located in the Pamirs Plateau, the Tianshan Mountains and the northern Himalayas. The precipitation changes were weak in the boreal warm seasons (June, July, August, September and October), with a dry centre located in the Pamirs Plateau and a wet centre in the northern Himalayas. These projected changes can be directly attributed to the overlay of anomalous moisture convergence and evaporation, and other possible mechanisms of dynamic and thermodynamic factors need to be further elucidated.