In order to study the responses of water balance components to different land cover characteristics in long-term average scale in Chaobai River Basin, the data of precipitation, streamflow and evaportranspiration from 1980 to 2013 were adopted to calibrate a coupled water and heat balance model. Moreover, the change of the streamflow was predicated based on the reasonable forecasting of the underlying surface in the next ten years. The results showed that the model was applicative very well in Chaohe and Baihe river basins. On the basis of the existing underlying surface information, 11 situations of the change of the underlying in the future is predicted respectively by use of both data of the forest land area and the underlying surface parameters of grass watershed. The stremflow in the future was predicated based on the 11 situations respectively. The results showed that the streamflow ranges from 26.47 mm to 53.55 mm in Chaohe River Basin and the streamflow ranges from 17.57 mm to 41.53 mm in Baihe River Basin. The innovation of this work is predicting the streamflow based on the predicting of underlying surface in the future. It is of great significance for the study of water resources and formulating the water managements.
Based on the collected climate data regarding daily temperature, precipitation, wind speed, sunshine hour as well as related humidity from 53 meteorological stations, and atmospheric circulation index and sunspot in the study region during 1961-2013, evapotrans-piration (ET0) was estimated by applying Penman-Monteith model. Additionally, Inverse Dis-tance Weighted was applied to comprehensively investigate the temporal-spatial variations of ET0, precipitation and aridity index (AI). The abrupt change and period of ET0, precipitation and AI were characterized using comprehensive time series analysis conducted with moving M-K test and Morlet wavelet. Principal component analysis was employed to analyze the factors that influenced the AI. The results showed that: In recent 53 years, precipitation displayed an increasing trend (8.81 mm/10 a), ET0 and AI presented decreasing trend on the whole at the rates of -28.73 mm/10 a and -0.05/10 a, suggesting that the regional climate trended to be wetter in Xinjiang. As for annual distribution, ET0 and precipitation both exhibited unimodal distributions with peaks appeared in August (137.12 mm) and July (24.58 mm), the maximum in September (0.9), and the minimum in January (0.46). Spatially, the ET0 in southern Xinjiang was greater than that in northern Xinjiang, and that of east was greater than that of west; the precipitation in northern was greater than that in southern. The spatial patterns of AI and precipitation were opposite. Overall, the AI in south was greater than that in north, and that in basin was greater than that in the mountains. The M-K trend of AI was between 0 - -0.02/a, and the decreasing trend of AI in north was more extraordinary than that in south, consistent with the facts that the north of Xinjiang was wetter than the south. The abrupt changes for ET0 and precipitation occurred in 1987 and 1981, respectively. There were two distinct point mutations of AI in 1981 and 1984. Morlet wavelet and its power spectrum analysis showed: Precipitation exhibited periods of 6.49 a, 5.71 a and 4.35 a (p≤0.05), ET0 showed the period of 21.37 a (p≤0.2), and AI had periods of 6.62 a and 3.45 a (p≤0.1), indicating that it was related to atmospheric circumfluence and El Niño events to some extent. Principal component analysis demonstrated: AI was positively correlated with temperature and negatively correlated with precipitation, and the southern part of Xinjiang was more sensitive to precipitation than the northern part. Correlation coefficient between AI and WYMI, and ENSO were 0.46 (p≤0.05) and -0.34 (p≤0.05), respectively.