基于DHSVM的遥感土壤水分数据同化对水文过程关键要素的影响——以湘江流域为例  

Impact of DHSVM-based multi-source remote sensing soil moisture data assimilation on key elements of hydrological processes-A case study of the Xiangjiang River Basin

作　　者：闵馨童 朱轲 朱仟 魏庆 MIN Xintong;ZHU Ke;ZHU Qian;WEI Qing(School of Civil Engineering,Southeast University,Nanjing 211189,China)

机构地区：[1]东南大学土木工程学院,江苏南京211189

出　　处：《中山大学学报（自然科学版）（中英文）》2024年第2期35-45,共11页Acta Scientiarum Naturalium Universitatis Sunyatseni

基　　金：国家自然科学基金(52009020)。

摘　　要：本文研究基于集合卡尔曼滤波(EnKF)的多源遥感土壤水分数据同化,对分布式水文模型DHSVM模拟过程中关键要素的影响。以湘江流域为例,选择SMAP和ASCAT遥感土壤水分数据,利用EnKF算法对DHSVM进行土壤水分模块的遥感数据同化。通过对比分析无同化方案、ASCAT同化方案和SMAP同化方案得到的径流和土壤水分结果,评估多源遥感土壤水分数据同化对水文关键变量模拟的影响。结果显示,湘江流域内,ASCAT同化无论是从径流模拟还是土壤水分模拟方面都要优于SMAP同化。径流模拟方面,ASCAT同化方案的NSE(NSE=0.677)相比无同化方案(NSE=0.662)有所提升,BIAS值减小了1.7个百分点。土壤水分模拟方面,相比无同化方案,ASCAT同化NSE提升了10%,BIAS减小了4.7个百分点,RMSE减小了12.5%。相对而言,SMAP同化方案整体模拟效果的改进并不显著。研究结果突出了遥感土壤水分数据同化的有效性,对水文变量模拟的改进具有重要意义。This study aims to investigate the effects of EnKF-based multi-source remote sensing soil moisture data assimilation on key elements of hydrological processes in the context of the Xiangjiang River Basin.SMAP and ASCAT remote sensing soil moisture data were selected for assimilation into the Distributed Hydrology Soil Vegetation Model(DHSVM)using the Ensemble Kalman Filter(EnKF)algorithm.By comparing and analyzing the simulated runoff and soil moisture results of three simulation scenarios:the non-assimilated model,ASCAT-DHSVM,and SMAP-DHSVM,the impact of multi-source remote sensing soil moisture data assimilation on hydrological variable simulation was evaluated.The results indicate that within the Xiangjiang River Basin,the ASCAT assimilation scheme outperforms the SMAP assimilation scheme in both streamflow simulation and soil moisture simulation.In terms of streamflow simulation,the ASCAT assimilation scheme exhibits an overall improvement in Nash-Sutcliffe Efficiency(NSE)(NSE=0.677)compared to the non-assimilation scheme(NSE=0.662),along with a decrease of 1.7 percentage points in BIAS.In terms of soil moisture simulation,compared to the non-assimilation scheme,the ASCAT assimilation scheme shows an overall improvement of 10%in NSE value,a decrease of 4.7 percentage points in BIAS,and a decrease of 12.5%in RMSE.In contrast,the improvement in the overall simulation performance of the SMAP assimilation scheme is statistically insignificant.The results underscore the efficacy of assimilating remote sensing soil moisture data,particularly through the ASCAT assimilation scheme,in enhancing hydrological variable simulation.These findings hold significant implications for water resource management in the Xiangjiang River Basin.

关 键 词：DHSVM ENKF 多源遥感数据 土壤水分 数据同化 

分 类 号：S152.71[农业科学—土壤学]