An Evaluation of Precipitation in Dongting Lake Basin on CMIP5 Models

The rainfall in the Dongting Lake Basin influences tens of millions of people, and its long-term change remains uncertain. In this paper, 15 CMIP5 models with precipitation data for the time period of 2006-2019 for which reliable observations are available under the RCP4.5 scenario were evaluated for their applicability, and the models with better simulation results were selected for predicting the precipitation in the Dongting Lake Basin during the flood season (April-September) in the mid-21st century (2020-2049). The results of the study show that (i) most models behaved reasonably consistent with the observation in the Dongting Lake Basin, and predicted an upward trend for the future precipitation while the multi-model ensemble (MME) showed a relatively slow increasing trend of 0.8 mm/year; (ii) the future precipitation in Dongting Lake Basin presented a variation form of "peak-valley-peak-valley", suggesting strong interannual and interdecadal variations; (iii) the interannual variability showed great agreement with large-scale circulation, implying that the rainfall is controlled by the circulation. The analysis of the wind fields at 200 hPa and 850 hPa in the peak and valley years showed that the characteristics of low-level convergence and high-level divergence were significantly stronger in the peak precipitation years than in the weak years; moreover, the teleconnection pattern of "+ - +" from Europe, the Ural Mountains, and East Asia was clearly manifested in the 500 hPa height field of the Dongting Lake Basin, which can influence the intensity of the trough ridge over East Asia and change the low-level water vapor convergence and divergence, thus affecting the source of water vapor in the Dongting Lake Basin.