CAJ | 학술논문

为评价降雨产品在流域尺度水文模拟中的适用性，该文首先依据实际降雨数据评价 GLDAS、TMPA 和ERA-Interim 3种降雨产品在研究流域的可替代性。多指标模糊优选评价表明，基于双线性插值法的 TMPA 和ERA-Interim降雨产品数据在日、月尺度上的模拟效果均较好，相对误差（Bias）分别为5.48%和7.46%，均小于10%，相关系数（CC）在0.82~0.98之间，具有站点降雨的可替代性，可作为水文模型的输入。然后用此2种降雨产品分别驱动SWAT模型，进行研究流域的水文模拟精度分析。TMPA产品的日、月尺度水文模拟结果的Bias分别为10.88%和11.03%，纳什系数（NSE）分别为0.46和0.78，而ERA-Interim的日、月尺度水文模拟结果的Bias分别为9.11%和9.27%，NSE分别为0.44和0.70。以Bias、CC和NSE为特征指标的多指标模糊优选模型的评价结果为：TMPA产品日、月尺度水文模拟结果的相对隶属度值分别为0.51和0.53，大于ERA-Interim的相对隶属度值0.49和0.47，表明TMPA产品更适用于流域尺度的水文模拟。该文评价方法及结论可为类似流域尺度的水文模拟提供参考。
위평개강우산품재류역척도수문모의중적괄용성，해문수선의거실제강우수거평개 GLDAS、TMPA 화ERA-Interim 3충강우산품재연구류역적가체대성。다지표모호우선평개표명，기우쌍선성삽치법적 TMPA 화ERA-Interim강우산품수거재일、월척도상적모의효과균교호，상대오차（Bias）분별위5.48%화7.46%，균소우10%，상관계수（CC）재0.82~0.98지간，구유참점강우적가체대성，가작위수문모형적수입。연후용차2충강우산품분별구동SWAT모형，진행연구류역적수문모의정도분석。TMPA산품적일、월척도수문모의결과적Bias분별위10.88%화11.03%，납십계수（NSE）분별위0.46화0.78，이ERA-Interim적일、월척도수문모의결과적Bias분별위9.11%화9.27%，NSE분별위0.44화0.70。이Bias、CC화NSE위특정지표적다지표모호우선모형적평개결과위：TMPA산품일、월척도수문모의결과적상대대속도치분별위0.51화0.53，대우ERA-Interim적상대대속도치0.49화0.47，표명TMPA산품경괄용우류역척도적수문모의。해문평개방법급결론가위유사류역척도적수문모의제공삼고。
Precipitation is the important forcing data for hydrological models. However, the precision of hydrological modeling in many regions of the world is limited by the lack of precipitation data. To overcome such limitation, some available globally gridded high resolution precipitation datasets have been used to simulate runoff. Although these precipitation products have been proved to have certain accuracy and good potential for hydrologic simulation, selecting the most optimal precipitation product for certain regions is still important because the accuracy and hydrological utility of precipitation dataset vary in different regions. In this study, the suitability of three precipitation products: GLDAS, TMPA and ERA-Interim for hydrological modeling are evaluated using the multi-objective fuzzy optimum model and the Soil and Water Assessment Tool (SWAT) model during the period of 2001-2012 in the Huanren reservoir catchment, located in Hunjiang River of China. SWAT is a physical-based hydrological model, which takes precipitation series from gauged stations as input to simulate runoff. First, in order to obtain the rainfall time series for SWAT model, three interpolation methods, including inverse distance-weighted method, bilinear interpolation method and the nearest point method, were used to compute the station data from the three precipitation products. Then, the quantitative accuracy of these time series were assessed by statistical indices of mean error (ME), root mean square error (RMSE), correlation coefficient (CC) and relative bias (Bias). The multi-objective optimum fuzzy model aiming to identify the appropriate interpolation method and justify the precipitation products alternative to the gauged values was established, which used ME, RMSE, CC and Bias as characteristic indices. The results revealed that the precipitation of GLDAS cannot serve as an alternative due to its underestimation for the whole basin, which can increase the risk of flood control operation. Moreover, the bilinear interpolation was more appropriate to interpolate the grid precipitation to rain gauge compared with the inverse distance-weighted method and the nearest point method. In addition the precipitation of TMPA and ERA-Interim obtained with the bilinear interpolation method were proved to have perfect fitting with the gauge values at daily and monthly scale, thus they both were alternative to the gauge values as the forcing data to the hydrological model. The two precipitation series were then used as the input to the SWAT model that has been calibrated with rain gauge inputs to get the simulated stream flow and their Bias. Correlation coefficient and Nash-Sutcliffe efficiency coefficient (NSE) were calculated against the observed stream flow. Based on the criteria of the statistical indices specified in the present study, it can be concluded that the performances of these simulations were acceptable at daily and monthly scale. With Bias, CC and NSE as characteristic indices, the multi-objective optimum fuzzy model was developed to evaluation the suitability of TMPA and ERA-Interim for hydrological simulation. The results revealed that TMPA was more suitable for hydrological runoff modelling at the basin scale. The methodologies and approaches developed in the present study will be a reference for hydrological modelling of similar watersheds.