军事知识服务平台
未来气候情景及土地利用变化下湟水河流域水源涵养量时空演变特征分析
Predictive analysis of spatiotemporal evolution characteristics of water conservation in the Huangshui River Basin under future climate and land-use scenarios
摘要 为评估未来气候与土地利用变化对流域水源涵养的影响,本文以湟水河流域为研究区,针对全球逐日降尺度气候预估数据集的降水数据在复杂地形区存在较大偏差的问题,引入中国区域地面气象要素驱动数据集的观测数据来训练卷积神经网络-长短期记忆网络模型,对降水序列进行校正并生成高精度驱动数据.在此基础上耦合InVEST模型,模拟不同共享社会经济路径的气候情景与土地利用情景下,2020-2050年湟水河流域水源涵养量的时空演变,并采用双因素因子分离法分解驱动贡献.结果表明:通过卷积神经网络-长短期记忆网络模型校正显著降低了降水偏差,对强降水特征的再现能力明显改善,从而提高了水源涵养模拟的输入可信度.未来湟水河流域水源涵养量总体呈波动上升,2020-2030年增长较快、2030-2050年趋于平缓,其变化主要受降水增加与蒸散增强的共同制约,但总体上降水增量的补偿作用更强;空间格局表现为"西部上游高、东部下游低",且高值区范围逐步扩展.归因结果显示,气候变化为主导因素(贡献率>76%),土地利用变化次之(约19%),二者交互作用呈增强趋势,SSP5-8.5与生态保护叠加情景下水源涵养量最高.研究揭示了暖湿化背景下湟水河流域水源涵养对"水热同增"的响应特征,可为流域生态修复与极端气候风险适应提供依据.
Abstract:
To predict the impacts of future climate change and land-use change on water conservation in the Huangshui River Basin,this study addresses the precipitation biases in the NASA Earth Exchange Global Daily Downscaled Projections(NEX-GDDP-CMIP6)dataset over complex terrain by training a convolutional neural network-long short-term memory(CNN-LSTM)model with observations from the China Meteorological Forcing Data-set(CMFD).The precipitation series was bias-corrected to produce improved climate forcing data.The corrected forcing was then coupled with the InVEST model to simulate the spatiotemporal evolution of water conservation from 2020 to 2050 under different Shared Socioeconomic Pathway(SSP)climate scenarios and land-use scenarios,and a two-factor separation method was applied to quantify the contributions of the driving factors.Results show that the CNN-LSTM correction can substantially reduce the precipitation biases and markedly improve the representation of heavy precipitation,thereby enhancing the reliability of meteorological inputs for water conservation simulations.Water conservation is projected to increase overall with interannual fluctuations,growing faster from 2020 to 2030 and gradually leveling off from 2030 to 2050.The variation is primarily constrained by the combined effects of increased precipitation and enhanced evapotranspiration,while the compensating effect of precipitation increase is dominant.Spatially,water conservation is higher in the western upstream area and lower in the eastern downstream area,with the high-value zone expanding gradually.Attribution analysis indicates that climate change is the domi-nant factor(contribution rate>76%),followed by land-use change(about 19%),and the interaction between the two shows an increasing trend.The highest water conservation is projected under the combined SSP5-8.5 and ecologi-cal protection scenario.This study reveals the response characteristics of water conservation in the Huangshui River Basin to the concurrent increases in water supply and atmospheric evaporative demand under a warming-wetting background,providing a basis for ecological restoration and adaptation to extreme climate risks in the basin.
刊 名 水利学报 ISTICEIPKU
年,卷(期): 2026, 57(9)
分类号: TV211.1
机标分类号: P467 S161 TP391.41
在线出版日期: 2026年10月30日
基金项目: 2023YFC3008502:国家重点研发计划2024-ZJ-773:青海省应用基础研究项目SKS-2022002:水利部重大科技项目52309103:国家自然科学基金22JCYBJC01180:天津市自然科学基金,