CFSR气象数据在流域水文模拟中的适用性评价——以灞河流域为例

近年来,全球再分析气象数据已经越来越多地被运用到世界各地的水文建模中,但是其模拟的效果有很大差异。为探讨CFSR再分析数据在流域水文模拟中的适用性问题,本文以灞河流域为研究区,使用两种气象数据（传统气象数据和CFSR气象数据）构建SWAT水文模型,并从年和月尺度分别进行灞河流域2001-2012年的径流模拟,利用回归分析、纳什效率系数NSE和百分比偏差PBIAS等评价方法对两种数据的模拟效果进行对比。最后,提出了CFSR气象数据订正的方法。结果表明：① CFSR气象数据在灞河流域水文模拟中有一定的适用性,模拟结果的拟合优度R²>0.50,NSE>0.33,|PBIAS|<14.8,纳什效率系数NSE偏低。尽管CFSR气象数据质量存在一定问题,但是经过降雨数据订正后能够取得比较满意的模拟效果。② CFSR气象数据模拟流量比实测流量偏高,这主要是由于CFSR逐日降水数据估算的降雨天数较多、雨强较大,一般会导致该数据在水量平衡方面能够模拟出较高的基流和洪峰流量（个别年份除外）。③ 灞河流域CFSR降水数据（x）与实测降水数据（y）之间的关系大致可用幂指数方程表达：y = 1.4789x^0.8875（R² = 0.98,P<0.001）,每个CFSR站点的拟合方程略微不同,此方程为CFSR降水数据的订正提供了理论基础。

Applicability evaluation of CFSR climate data for hydrologic simulation: A case study in the Bahe River Basin

To discuss the applicability of CFSR reanalysis data in the hydrologic simulation of watersheds, taking the Bahe River Basin as a case, this paper uses two types of weather data to establish SWAT hydrologic model, and carries out the runoff simulation from 2001 to 2012 of the Bahe River Basin on yearly and monthly scales, and makes use of such evaluation methods as repression analysis, Nash-Sutcliffe Efficiency (NSE) and Percent Bias (PBIAS) to compare the simulation effect of both data. At last it puts forward CFSR weather data correction method. The research results show: (1) To some extent, CFSR climate data has applicability for hydrologic simulation in the Bahe River Basin, the R² value of the simulated results is above 0.50, the Nash-Sutcliffe Efficiency is above 0.33, and |Percent Bias (PBIAS)| is below 14.8. Although the quality of the CFSR weather data has some problems, it basically achieves a satisfactory result in the hydrologic simulation of the Bahe River Basin after rainfall data correction. (2) The simulated flow of the CFSR weather data is higher than the observed flow, the reason may be that there are more rainy days and the rainfall intensity is much stronger according to the estimation of day-to-day rainfall data by CFSR, which usually leads to the fact that the data will simulate a higher base flow and flood peak discharge in terms of the Water Balance, except for individual years. (3) The relation between the CFSR rainfall data and the observed rainfall data of the Bahe River Basin can be displayed by a power exponent equation: y=1.4789x^{0. 8875} (R² = 0.98, P <0.001); the fitted equation of each CFSR station varies slightly from others. The equation just lays a theoretical basis for correction of the CFSR rainfall data.