GeoSFM水文模型在西藏拉萨河流域的应用

Application of Geo SFM Model in the Lhasa River Basin,Tibet

基于物理过程的半分布式水文模型USGS Geo SFM(Geospatial Stream Flow Model)对拉萨河2005年1—12月日径流量进行了模拟,并与同期的观测资料进行了对比。结果表明,Geo SFM模型在拉萨河流域的模拟效果较好,Nash-Sutcliffe效率系数为0.72,模拟和观测值之间的线性相关系数达0.89。由于Geo SFM模型输入参数较少,很多可以从应用广泛的全球或大尺度数据中获取,尤其是卫星遥感降水估算产品直接可以作为模型的降水驱动参数,所以在没有或缺少水文气象观测资料的地区应用前景广阔。

In this study,the daily river flow of Lhasa River basin,Tibet,China is simulated using USGS Geo SFM model and the validation is made using corresponding daily gauge data from Lhasa hydrological station. The results show that the general performance of Geo SFM model in Lhasa River basin is good and reasonable. Quantitatively,the simulated discharge can explain 80% of the variability in the observed discharge. The Nash- Sutcliffe efficiency coefficient（ NSCE） and correlation between the observed and simulated runoff is 0. 72 and 0. 89,respectively.Qualitatively,the general trend of the simulated and observed daily runoff is same,which presents that the both simulated and observed river flow in Lhasa River basin is low with very small inter-daily variations before Jun. After onset of raining season,the runoff of Lhasa River is increasing fast until the river discharge reaches to the highest level in later August. During this period,the variation of daily discharge is high due to daily rainfall fluctuation.Most of discharge peaks in the basin are captured by Geo SFM model,but the simulated value is less than the observed value. The highest peak of discharge in later August is also smaller than the observed value. From the end of August,along with monsoon offset and rapid decease in rainfall,the simulated and observed discharge is deceasing fast. Until early October the river discharge reaches to a relatively stable stage. However,the average simulated value is smaller than the observed value and during the raining season the inter-daily variations of simulated runoff are higher than the observed runoff. Geo SFM is a physically-based semi-distributed hydrologic model. It have few parameters and variable input data requirements to simulate the dynamics of runoff processes by using remotely sensed data such as satellite-derived rainfall products and widely available continental or global data sets.Therefore,Geo SFM is very useful in sparsely hydro-meteorological station distributed regions.