基于WEP模型和TRMM_PR的唐家山堰塞湖入湖径流预报

提高堰塞湖入湖径流量预报的准确性对其下游洪水灾害的预警与预防具有重要意义，但由于堰塞湖往往形成于缺乏观测资料的山区，给预报工作带来了很大困难。为解决上述技术难题，本文采用将具有物理机制的WEP分布式水文模型与热带降雨测量卫星的降雨数据（TRMM_PR）相结合的方法，以2008年5月12日四川汶川大地震形成的唐家山塞湖为研究对象，对其入湖径流量进行了逐日模拟预报。首先，由1:250000的DEM数据生成研究区域（12029.25km2）内的数字水系，并收集了相关的三期土地利用（1985、2000、2005年）、土壤等下垫面数据；然后，采用研究时段内4个气象站逐日降水、气温、日照时数、风速、相对湿度数据作为输入来驱动模型，并选择区内4个水文站1976-1980年的月均径流观测数据对模型进行验证；最后，为提高模拟预报精度，下载了热带降雨测量卫星的TRMM_PR数据，对模型的降水输入进行了改进。结果显示：利用地面气象站点观测降水数据对TRMM_PR数据进行修正后，可在一定程度上改善模型的降水输入，有利于提高模型的径流预报精度。

Predicting inflows to Tangjiashan barrier lake based on WEP model and TRMM precipitation data

Inflow prediction to barrier lake is of importance for flood disaster warning and prevention to the downstream regions, but it is difficult because barrier lakes are usually formed in mountainous regions and lack of gauge data. To solve this technical problem, this study tries to predict inflows to the Tangjiashan barrier lake formed during the Wenchuan earthquake on 12 May 2008 by combining the WEP distributed hydrological Model and the TRMM-based precipitation estimates. Firstly, the DEM data with a scale of 1:250000 were utilized to generate the digital river system in the study area (12029.25km2), and related underlying surface data (land use, soil etc.) were collected. Then, daily precipitation, temperature, sunshine hours, wind speed, relative humidity data of 4 meteorological stations were used as inputs to drive the model, and five-year daily data from 1976 to 1980 of 4 flow gauge stations were applied to validate the WEP model. Finally, the TRMM_PR three-hourly data from May 12 to June 9 were downloaded to modify the precipitation input and run the model for the prediction of inflow to the barrier lake. The application results showed an acceptable accuracy, illustrating that the modulation of the TRMM_PR data by the gauge data could improve the precipitation input for the model, which is also beneficial to improve the results of runoff prediction in ungauged basin.