洞庭湖流域陆地水储量变化长时序重构及水文干旱特征  

Reconstruction of Long-term Terrestrial Water Storage Anomalies and Associated Hydrological Drought Characteristics in the Dongting Lake Basin

作　　者：黄志勇 古天豪 隆院男[1,2] 陈绪慧 杨享运 刘轩宇 黄铭昊 胡庆麟 周思婷 HUANG Zhiyong;GU Tianhao;LONG Yuannan;CHEN Xuhui;YANG Xiangyun;LIU Xuanyu;HUANG Minghao;HU Qinglin;ZHOU Siting(School of Hydraulic and Ocean Engineering,Changsha University of Science&Technology,Changsha 410114,China;Key Laboratory of Dongting Lake Aquatic Eco-Environmental Control and Restoration of Hunan Province,Changsha 410114,China;Satellite Application Center for Ecology and Environment,Beijing 100094,China)

机构地区：[1]长沙理工大学水利与海洋工程学院,湖南长沙410114 [2]洞庭湖水环境治理与生态修复湖南省重点实验室,湖南长沙410114 [3]生态环境部卫星环境应用中心,北京100094

出　　处：《地球科学进展》2024年第12期1285-1298,共14页Advances in Earth Science

基　　金：国家自然科学基金项目(编号:42301404);湖南省自然科学基金项目(编号:2022JJ40480);湖南省大学生创新训练项目(编号:S202310536102)资助。

摘　　要：陆地水储量变化长时序可用于定量表征水文干旱特征。利用重力卫星、全球水文模型(WGHM)和再分析模型(MERRA-2)的陆地水储量变化,以及实测气温和降水量数据,对比广义回归神经网络和长短期记忆神经网络方法在重构洞庭湖流域陆地水储量变化长时序的适宜性,并基于此时序对水文干旱特征进行定量分析。结果表明:广义回归神经网络重构结果优于长短期记忆神经网络,以MERRA-2为输入数据的重构结果优于WGHM。1980年以来洞庭湖流域水储量呈上升趋势,水库蓄水量增加是重要因素。扣除水库蓄水量增加趋势后的水储量亏损指数能更好地揭示历史水文干旱特征。2022年7~12月,洞庭湖流域经历极端干旱,陆地水储量总亏损强度达-790 mm,其中因扣除水库蓄水量增加趋势导致的水储量亏损值达-140 mm。研究突出了水库蓄水量趋势对水文干旱精准评估的重要影响,可为洞庭湖流域乃至全球其他流域或地区水资源动态监测、旱涝灾害监测和评估提供理论和方法指导,为更科学、合理地管理流域水资源提供参考依据。Long-term Terrestrial Water Storage Anomalies(TWSA)can be used to quantitatively characterize hydrological drought features.TWSA data in the Dongting Lake Basin(DTLB)were retrieved from multiple datasets,including satellite gravimetry(Gravity Recovery and Climate Experiment satellites and the succeeding Follow-On mission),global hydrology models(WaterGAP Global Hydrology Model,v2.2e),and a reanalysis model(Modern-Era Retrospective Analysis for Research and Applications,version 2).These TWSA datasets,along with observed temperature and precipitation data,were utilized to compare the suitability of two methods—the General Regression Neural Network and Long Short-Term Memory Neural Network—for reconstructing long-term TWSA in the DTLB.Quantitative analyses of hydrological drought characteristics were conducted using the long-term TWSA.The results indicate that in the DTLB,the reconstructed TWSA using General Regression Neural Network is superior to that obtained using Long Short-Term Memory Neural Network,and reconstructions using Modern-Era Retrospective Analysis for Research and Applications-2 as input data outperform those using WaterGAP Global Hydrology Model.Since 1980,terrestrial water storage in the DTLB has shown an increasing trend,partly influenced by the rising water storage capacity of reservoirs.The water storage deficit index,estimated by removing the increasing trend in reservoir storage,provided a more accurate representation of historical hydrological drought characteristics.From July to December 2022,the DTLB experienced an extreme drought,with a total terrestrial water storage deficit intensity reaching-790 mm,of which-140 mm was attributed to the water storage deficit after adjusting for the increasing trend in reservoir storage.This study①emphasized the significant influence of reservoir water storage trend on accurate assessment of hydrological drought;②provided a theoretical and methodological guidance on dynamic monitoring of water resources,drought and flood monitoring and assessment

关 键 词：重力卫星 水文干旱 长时间序列重构 神经网络 水库蓄水量 

分 类 号：P228.9[天文地球—大地测量学与测量工程]