赣江流域多种数据同化方案的径流模拟比较

Comparison of data assimilation based approach for daily streamflow simulation under multiple scenarios in Ganjiang River Basin

为了探讨水文模型在不同水文数据同化方案下的径流模拟差异,本文采用集合卡尔曼滤波算法,以遥感蒸散发产品、实测径流为观测数据,构建了基于新安江模型的数据同化框架。基于此框架设计了4种不同同化方案(DA-ET、DA-ET(K)、DA-ET-Q、DA-ET-Q(K))以及1种对照方案OL,以赣江流域开展实例研究,评估了水文数据同化中遥感蒸散发产品的时间分辨率、模型蒸散发相关参数时变与否以及多源数据同化对径流模拟的影响。结果表明:在DA-ET方案下,同化两种不同时间分辨率的蒸散发产品均能提高模型整体的径流模拟精度,且时间分辨率更高的产品的同化效果更好;在DA-ET方案的基础上,考虑加入实测径流进行同化能够提升模型径流模拟精度,且DA-ET(K)与DA-ET-Q(K)方案所得径流相对误差的减幅均超过了20%,说明在蒸散发同化过程中同时考虑蒸散发参数动态变化的结果更优;相较于OL方案,4种同化方案均能不同程度地提高模型对径流高水部分的模拟能力,但DA-ET-Q(K)方案表现最差,而其余方案差异并不显著。本研究有助于进一步了解不同数据同化方案在径流模拟中的差异,从而为水资源高效利用与科学管理提供科学依据。

In order to investigate the impact of hydrological data assimilation scenarios on streamflow simulations of hydrological models,a data assimilation framework was proposed based on the Xin anjiang model and the ensemble Kalman filter,in which remotely sensed evapotranspiration products and measured streamflow were used as observation data.Based on this framework,four assimilation scenarios including DA-ET,DA-ET(K),DA-ET-Q,DA-ET-Q(K)and a comparative scenario(i.e.,OL)were designed and tested in Ganjiang River Basin,aiming to evaluate the impact of the time resolution of remote sensing evapotranspiration products,whether the time-parameters related to model evapotranspiration and multi-source data assimilation on runoff simulation after hydrological data assimilation.The results showed that assimilation of two evapotranspiration products with different time resolutions could improve the accuracy of model s streamflow simulations in the DA-ET scenario,and model performed better by assimilating the products with higher temporal resolutions.On the basis of DA-ET scenario,adding measured streamflow into assimilation could improve the accuracy of model s streamflow simulations,and the reduction of the relative error obtained by DA-ET(K)and DA-ET-Q(K)scenarios was more than 20%.The results indicated that the accuracy of streamflow simulations was improved through assimilation of ET products,especially when the model evapotranspiration parameter was treated to be time-varying.Compared to the OL scenario,four assimilation scenarios could improve the accuracy of model s the 10%highest streamflow simulations in different degrees,but the other scenarios with no significant difference was better than DA-ET-Q(K)scenarios.Our results could further understand the differences of streamflow simulation in different data assimilation scenarios and provided scientific basis for the efficient utilization and management of water resources.