In order to verify and assess GRAPES more comprehensively, this research attempts to simulate the weather processes of Tarim Basin, southern Xinjiang, China in January, April, July and October 2013 by days, and compares those with actual data from monitoring stations in order to analyze the prediction capability of GRAPES The results show that GRAPES is able to predict 2 m temperatures, wind and rainfall in desert areas, but there exist errors GRAPES can predict the daily trend of 2 m temperatures in Tarim Basin but the errors of the extremum values are obvious As for the prediction of wind field, GPAPES performs better in January, April and October when wind speed is higher, which means it basically is able to predict wind direction in desert areas However, in July, when the wind speed is low, GRAPES performs unsatisfactorily The problems in precipitation prediction mainly lie in greater false prediction rate, and inaccuracy also occurs in the prediction of precipitation centers