雅江流域多源遥感驱动的SRM模型精度影响因子研究

Analysis of factors affecting accuracy of snowmelt runoff model based on multi source remote sensing in the Yarlung Zangbo River Basin

利用多源遥感数据驱动SRM模型模拟雅鲁藏布江4个不同水文站控制流域的融雪过程,分析流域特征(面积、地形地貌因子、植被类型)、模型输入变量(气温、降水、积雪覆盖率)对遥感驱动SRM模型模拟精度的影响.研究表明:遥感驱动的SRM模型在雅鲁藏布江研究区域模拟精度较高,NSE系数达到0.83以上;但其精度受流域面积大小的影响,具有尺度阈,在一定面积尺度上模拟精度最大,流域地形越复杂,其模拟效果越差;植被覆盖类型影响融雪径流过程的模拟,进而影响模拟结果;模型3个主要输入变量气温、降水、积雪覆盖率对遥感驱动的SRM模型的模拟精度同样有很大程度的影响.研究结果对资料匮乏的高海拔地区水资源利用具有重要意义,同时可为遥感驱动的SRM模型的推广应用提供有效支撑.

The snowmelt runoff model（SRM）based on multi-source remote sensing was used to simulate the snowmelt process in four watersheds in the Yarlung Zangbo River Basin.The objective was to analyze the effect of watershed features（including area,topographic factor,vegetation type）and input variables（including temperature,precipitation,area of snow）of the model on the accuracy of SRM based on multi-source remote sensing.It was found that the simulation fit-indexes of Nash-Sutcliffe coefficient（NSE）was more than 0.83,the accuracy of SRM was higher in the Yarlung Zangbo River Basin.The accuracy was influenced by the area of watershed,with a scale threshold,higher of the accuracy value in the specific area and more complex of the watershed topography,worse of the simulation result.Types of vegetation cover affected the snowmelt runoff process and the simulation result.Temperature,precipitation,area of snow,the three major input variables also influenced the accuracy of simulation significantly.This study has important implications for the utilization of water resources in high altitude areas which lack sufficient data.The present study can provide effective support for promotion and application of SRM based on multi-source remote sensing.