基于ERA5-Land数据集的玛纳斯河径流模拟研究

气象数据是水文过程研究的关键要素,再分析数据的发展为资料缺乏地区的径流模拟提供了新的解决方案。为研究ERA5-Land再分析数据集在径流模拟中的适用性,本文以玛纳斯河流域肯斯瓦特水文站以上流域为研究区,选取多个评价指标对ERA5-Land降水和温度进行准确性评价,并采用经验模态分解(EOF)分析其在研究区内的分布特点。在准确性方面,ERA5-Land与实测数据具有较好相关性,降水探测率为0.96,能反映大多数的降水事件,但与实测数据相比总体偏高21.81%,气温准确性好于降水,总体拟合效果较好,最优范围为-520 ℃,在极值部分不确定性有所增加。EOF决定性模态表明研究区内降水、气温变化趋势基本一致,即易受大尺度天气系统影响。利用该数据集驱动SWAT模型在月、日尺度上对玛纳斯河流域进行径流模拟,在验证期纳什系数(NSE)分别为0.88和0.82,具有较好的模拟效果。ERA5-Land再分析数据集可为西北缺乏实测气象资料地区径流模拟提供参考。

Simulation of Manas River runoff based on ERA5-Land dataset

Meteorological data is a key factor in hydrological process research, and the development of reanalysis data provides a new solution for runoff simulation in data-scarce regions. In order to study the applicability of ERA5-Land reanalysis dataset in local runoff simulation, several evaluation indexes were selected to evaluate the accuracy of ERA5-Land precipitation and temperature in the Manas River basin above the Kanswat hydrological station, and the distribution characteristics of the two ERA5-Land components in the study area were analyzed by empirical orthogonal function (EOF). In terms of accuracy, ERA5-Land has a good correlation with measurement data. The precipitation detection rate was 0.96, which can reflect most precipitation events, but was 21.81% higher than the measured rate. The accuracy of temperature was better comparably, with better overall fitting accuracy, yielding an optimal range of -5-20 ℃, with increasing uncertainties near the extremum. The EOF deterministic mode showed that the precipitation and temperature flutuations in the study area followed basically the same trend, that is, they were susceptible to the influence of large-scale weather system. In the validation period, the Nash coefficients (NSE) were 0.88 and 0.82 for the final SWAT runoff models at daily and monthly scales, respectively. The ERA5-Land reanalysis dataset can provide a reference for runoff simulation in northwest China where measured meteorological data are lacking.