陇西祖厉河流域降水插值方法的对比分析

Modelling the Temporal and Spatial Variabilities of Precipitation in Zulihe River Basin of the Western Loess Plateau

降水作为五水转化和水量平衡的一个组成部分，在土壤一植被大气连续体物质与能量转化中起着重要的作用，是分布式水文模型中的重要参量。本研究选择了陇西黄土高原祖厉河流域，采用5种空间插值方法，对该区域降水的时空分布规律进行研究。经分析比较，空间插值法中趋势面方法最佳。趋势面方法经统计模型调整，使得降水模拟精度有所提高。从模拟结果来看：该流域降水呈现明显的纬度地带性分布，自南向北逐渐减少；降水受地形影响较大，海拔高度每降低100m，降水量平均减少20mm；流域绝大地区多年平均降水量分布在280～430mm范围内。该高时空分辨率的降水模拟模型对区域水资源分析及生态环境治理有着重要的意义。

Zulihe River, located in the west-central Chinese Loess Plateau, is one of major tributaries draining to Yellow River. Zulihe River basin is characterized by continental climate. Therefore, water is scarce resource. Precipitation is considered as the only source of water （i. e. , input） and varies both temporally and spatially in the study area. Understanding the temporal and spatial variations in precipitation ate＇fundamental to quantitatively assess the possibility and the paths of ecological restoration and its hydrological effects. To accomplish this understanding, we first evaluate interpolation method commonly used to estimate areal precipitation in order to choose the appropriate model. In the study, we evaluated five methods such as multiple regression model, inverse distance weighted interpolation, Kriging interpolation, spline interpolation, and Trend surface interpolation. The results indicated that the trend surface model is the best one to estimate precipitation in the study area. Then the trend model was adjusted by regression model because trend model was neglect of topbgraphic effect on precipitation. We found the adjusted model had higher accuracy than trend model. The temporal and spatial distributions of precipitation were estimated by combing trend surface interpolating model and regression model. Lower precipitation values are registered in the low valleys of Zulihe River. Higher precipitation values appear in the mountain regions. Annual mean precipitation increases 20 mm by increase of 100 m altitudes. Generally the precipitation decreases from the south to the north.