Evaluation of a multisite weather generator on precipitation simulation in the Yangtze river basin.

This paper presents the evaluation of a multisite statistical weather generator (MulGETS: Multisite weather Generator of École de Technologie Supérieure) based on its simulation effect of precipitation in the Yangtze River Basin. MulGETS effectively generates spatially correlated sequences of precipitation simultaneously, while maintaining their spatial and temporal distribution characteristics. On the spatial scales, the accuracy of the model varies from station to station, and in general, the errors are lower at stations in the middle and lower reaches of the Yangtze River Basin than in the upper reaches. This difference is likely to exist because of the lower amount of rainfall and more complex topography than those of the upper river basins. On the temporal scales, the simulated values are more precise on the annual scale than on the seasonal scale. Large relative errors occur more frequently in winter, ranging from -35% to 25%. MulGETS can consistently produce precipitation by considering the intensity, magnitude, and duration indices with sub-basin varied observations. However, the precipitation maxima were much lower than the observations. This work shows the general reasonability of the model in downscaling precipitation in the Yangtze River Basin. [ABSTRACT FROM AUTHOR]