军事知识服务平台
基于多尺度分解与深度学习融合的河道水温变化归因分析
Attribution analysis of river water temperature changes based on the integration of multi-scale decomposition and deep learning:A case study of the cascade reservoir system in the lower Jinsha River
摘要 梯级水库运行对于河道水温具有显著影响.本文基于最大重叠离散小波变换(MODWT)与多分辨率分析(MRA),结合ERA5-Land再分析数据与Bi-LSTM模型,构建了气候变化与水库运行对河道水温影响的量化归因框架,并以金沙江下游向家坝站为例,探究了梯级水库群运行对下游河道水温的多尺度影响.研究结果表明:(1)梯级水库联合运行显著削弱了中高频尺度(日、周、月尺度)的水温波动;(2)在低频尺度(季节及年际尺度)上,水库运行导致水温变化趋于平缓并出现相位延迟,且该现象随水库数量增加而增强,在四库运行期水温相位相较于气温相位滞后50d;(3)归因分析结果表明:在季节尺度,水库运行是水温改变的主要影响因素,随着水库数量增加呈现非线性变化;在年尺度,两库运行时期年均水温相较于自然时期下降0.22℃,水库运行为主要驱动因素造成降幅达0.33℃,进入四库运行时期,水温相较于自然时期上升0.41 ℃,此时完全由气候变化驱动.研究揭示了梯级水库对水温影响的多时间尺度特征,提出了新的归因分析方法,为水库生态调度与流域水温管理提供了科学依据.
Abstract:
The operation of cascade reservoirs significantly alters river thermal regimes.This study develops a quantitative attribu-tion framework integrating maximal overlap discrete wavelet transform(MODWT),multiresolution analysis(MRA),ERA5-Land reanalysis data,and a Bi-LSTM model to distinguish the effects of climate change and reservoir operation on river water tempera-ture.Using the Xiangjiaba Station on the lower Jinsha River as a case study,the multi-scale impacts of cascade reservoir operation on water temperature were examined.The results show that:(1)Cascade reservoir joint operation substantially attenuates water temperature fluctuations at medium-to-high frequency scales(daily,weekly,and monthly).(2)At low-frequency scales(seasonal and interannual),reservoir operation leads to smoother water temperature variations with phase delays,an effect that intensifies as the number of reservoirs increases.During the four-reservoir operation period,the water temperature phase lagged behind the air temperature phase by 50 days.(3)Attribution analysis indicates that at the seasonal scale,reservoir operation is the dominant driver of water temperature change,exhibiting a nonlinear increase in influence with the growing number of reservoirs.At the annu-al scale,during the two-reservoir operation period,the annual mean water temperature decreased by 0.22℃relative to the natural period,primarily due to reservoir operations,which contributed a decrease of 0.33 ℃.In the four-reservoir operation period,wa-ter temperature increased by 0.41℃ compared to the natural period,a change entirely attributable to climate change.This study reveals the multi-temporal scale characteristics of cascade reservoir impacts on water temperature,proposes a novel attribution methodology,and provides a scientific basis for ecological reservoir operation and watershed thermal management.
刊 名 湖泊科学 ISTICEIPKU
年,卷(期): 2026, 38(5)
机标分类号: TP391.41 R195.3 TN911.73
在线出版日期: 2026年9月28日
基金项目: 52409096:国家自然科学基金52009066:国家自然科学基金GJCZX-JJ-202418:水资源高效利用与工程安全国家工程研究中心开放研究基金项目,