Corrected GCM data through CMFD data to analysis future runoff changes in the source region of the Yangtze River, China.

The source region of the Yangtze River (SRYR) is located in the hinterland of the Tibetan Plateau (TP). Due to the harsh climate environment, there are few meteorological stations in its territory, and the measured meteorological data are also less. To overcome the impact of lack of measured meteorological data, the China Meteorological Forcing Dataset (CMFD) was used to couple with the meteorological data, to make the meteorological data more representative of the average climate state of the SRYR. Furthermore, we used the Soil and Water Assessment Tool (SWAT) to verify interpolation effect of meteorological data. Then the six global climate models (GCMs) were integrated through deep learning (DL) algorithm to make it closer to the changes in weather data after correction. The future water resource changes of SRYR were obtained by coupling the SWAT model with the integrated meteorological data. The results show that the CMFD dataset has a high precision in the SRYR, and that the average relative error (R_E), coefficient of determination (R²), Nash–Sutcliffe efficiency (E_ns) with the measured temperature and precipitation are 10.66%, 0.99, and 0.93, respectively. The CMFD dataset can be used for meteorological data integration. After the meteorological data integration, the E_ns between simulated runoff and observed runoff increased from 0.64 to 0.70, and the R_E reduced from 13.4 to 10.3%. Under the future climate scenario, the runoff in the SRYR shows a decreasing trend, the average annual runoff decreased by 11.89%, and the distribution of runoff during the year changes is even more dramatic. This shows that the SRYR faces the risk of water resources reduction, which will bring challenges to water resources management in the SRYR. This study provides a new idea for meteorological data correction. [ABSTRACT FROM AUTHOR]