GR4J与新安江模型在湿润地区比较研究  

Comparative Study of GR4J and Xinanjiang Models in Humid Area

作　　者：王何予 郭生练[1] 王俊 崔震 

机构地区：[1]武汉大学水资源与水电工程科学国家重点实验室,湖北 武汉

出　　处：《水资源研究》2021年第1期11-20,共10页Journal of Water Resources Research

摘　　要：为了比较不同水文模型在湿润地区的预报能力和效果,选用新安江流域上游屯溪水文站控制流域作为研究区域,分别利用GR4J和新安江模型对流域进行水文模拟,并开展洪水概率预报研究。结果表明:两个模型在湿润流域的模拟效果相当且都较好,GR4J和新安江模型模拟率定期和检验期的纳什效率系数(NSE)均达到90%以上,且相对误差(RE)较小;两模型在湿润流域均能进行可靠的概率预报,GR4J率定期和检验期的连续概率排位分数(CRPS)为18.73 m3/s、12.57 m3/s,其概率预报评价指标均相对于新安江模型更优,在概率预报方面性能稍好,精度较高;简单和复杂的模型在非特定条件下,可以取得同样效果,湿润地区的径流模拟、预报洪水可选择简单快捷的模型。In order to compare the forecasting abilities and effects of different hydrological models in humid areas, GR4J and Xinanjiang models were used to simulate runoff and probability flood forecasts in the Tunxi basin. The results were summarized as follows: Nash-Sutcliffe efficiency coefficient (NSE) are both over 90% in calibration and validation period, the Relative error (RE) values are also small. Both models can carry out reliable probabilistic forecasts in humid basins. The Continuous ranked probability score (CRPS) values of GR4J in calibration and validation period are 18.73 m3/s and 12.57 m3/s respectively;moreover, most of its probabilistic forecast evaluation indicators are better than Xinanjiang model. GR4J model performs with higher accuracy in probabilistic forecasting. Simple and complex models can achieve the same effect under non-specific conditions, GR4J model is preferred for runoff simulation and flood forecasting in humid areas.

关 键 词：水文模型 比较分析 径流模拟 概率预报 

分 类 号：P33[天文地球—水文科学]