基于XGBoost-SHAP的长江流域骤旱影响因素分析
Analysis of Factors Influencing Flash Droughts in the Changjiang River Basin Based on XGBoost-SHAP
DOI: 10.12677/jwrr.2026.152014, PDF,    科研立项经费支持
作者: 陈喾郡, 董前进*, 杨子浩:武汉大学水资源工程与调度全国重点实验室,湖北 武汉
关键词: 骤旱影响因素XGBoost-SHAP方法长江流域Flash Drought Influencing Factors XGBoost-SHAP Method Changjiang River Basin
摘要: 骤旱是21世纪初提出的新型干旱概念,与传统干旱相比,其具有快速发生、强度高、影响大等特点,导致预测困难、应急响应时间紧迫,对生态、农业与社会经济构成严峻挑战。当前骤旱研究在关键影响因子识别与驱动机制解析方面仍存在不足。长江流域骤旱形成机制复杂,不同区域影响因子差异显著,目前尚缺乏全流域尺度下各子区域响应机理的系统认识。本研究基于根区土壤水定义识别骤旱,并采用XGBoost-SHAP方法分析长江流域骤旱响应机制,进一步引入时空变量揭示骤旱的时空演变规律。结果表明:长江流域各子流域骤旱影响因子存在显著差异;引入时空变量后,各子流域模型拟合效果(R2)均有提升,且骤旱强度与骤旱强化率的主要影响因子多转为时空变量,说明骤旱具有明显的时空依赖性;不同子流域对时空变量的响应各异,反映骤旱影响机制具有空间异质性。本研究可为构建骤旱多因子协同分析框架、发展面向不同子流域的骤旱预测模型、提升骤旱早期预警能力及制定分区适应策略提供科学依据。
Abstract: Flash drought is a recently proposed drought concept characterized by rapid onset, high intensity, and severe impacts, making it more challenging to predict and leaving limited time for emergency response. It poses serious threats to ecosystems, agriculture, and socio-economic systems. However, current research on flash droughts still faces challenges, particularly in identifying key influencing factors and systematically understanding their driving mechanisms. Taking the Changjiang River Basin as an example, the formation mechanisms of flash droughts are complex, with significant regional differences in influencing factors. A comprehensive understanding of the response mechanisms across sub-basins at the whole-basin scale is still lacking. This study identifies flash drought events based on root-zone soil moisture and analyzes their driving mechanisms in the Changjiang River Basin using the XGBoost-SHAP approach. Furthermore, spatio-temporal variables are incorporated to investigate the evolution patterns of flash droughts. The results show that the influencing factors of flash droughts vary significantly among sub-basins. After introducing spatio-temporal variables, the R2 values of the models for all sub-basins improve obviously, and the primary influencing factors for both flash drought intensity and intensification rate shift to spatio-temporal variables, indicating strong spatio-temporal dependency. Different sub-basins exhibit distinct responses to spatio-temporal variables, reflecting the spatial heterogeneity of flash drought impact mechanisms. Different sub-basins exhibit distinct responses to spatio-temporal variables, reflecting the spatial heterogeneity of flash drought impact mechanisms. This study can support the development of a multi-factor synergistic analysis framework for flash droughts, facilitate the establishment of sub-basin-specific prediction models, and enhance early warning capabilities and regional adaptation strategies.
文章引用:陈喾郡, 董前进, 杨子浩. 基于XGBoost-SHAP的长江流域骤旱影响因素分析[J]. 水资源研究, 2026, 15(2): 107-118. https://doi.org/10.12677/jwrr.2026.152014

参考文献

[1] HUO, J., LIU, C., YU, X., JIA, G. and CHEN, L. Effects of watershed char and climate variables on annual runoff in different climatic zones in China. Science of the Total Environment, 2021, 754: 142157.[CrossRef] [PubMed]
[2] CAO, S., ZHANG, L., HE, Y., ZHANG, Y., CHEN, Y., YAO, S., YANG, W. and SUN, Q. Effects and contributions of meteorological drought on agricultural drought under different climatic zones and vegetation types in Northwest China. Science of the Total Environment, 2022, 821: 153270.[CrossRef] [PubMed]
[3] MISHRA, A. K., SINGH, V. P. A review of drought concepts. Journal of Hydrology, 2010, 391(1): 202-216. [Google Scholar] [CrossRef
[4] YANG, X., WU, F., YUAN, S., REN, L., SHEFFIELD, J., FANG, X., et al. Quantifying the impact of human activities on hydrological drought and drought propagation in China using the PCR-GLOBWB v2.0 model. Water Resources Research, 2024, 60: e2023WR035443.[CrossRef
[5] YUAN, X., WANG, Y., JI, P., WU, P., SHEFFIELD, J. and OTKIN, J. A. A global transition to flash droughts under climate change. Science, 2023, 380: 187-191.[CrossRef] [PubMed]
[6] ANDERSON, M. C., HAIN, C., OTKIN, J. A., ZHAN, X., Mo, K., SVOBODA, M., WARDLOW, B. and PIMSTEIN, A. An intercomparison of drought indicators based on thermal remote sensing and NLDAS simulations. Journal of Hydrometeorology, 2013, 14: 1035-1056.[CrossRef
[7] OTKIN, J. A., HAIGH, T., MUCIA, A., ANDERSON, M. C. and HAIN, C. Comparison of agricultural stakeholder survey results and drought monitoring datasets during the 2016 U.S. Northern Plains flash drought. Weather, Climate, and Society, 2018, 10(4): 867-883. [Google Scholar] [CrossRef
[8] MUKHERJEE, S. and MISHRA, A. K. A multivariate flash drought indicator for identifying global hotspots and associated climate controls. Geophysical Research Letters, 2022, 49(2): e2021GL096804. [Google Scholar] [CrossRef
[9] FORD, T. W. and QUIRING, S. M. Comparison of contemporary in situ, model, and satellite remote sensing soil moisture with a focus on drought monitoring. Water Resources Research, 2019, 55(2): 1565-1582. [Google Scholar] [CrossRef
[10] CHRISTIAN, J. I., HOBBINS, M., HOELL, A., OTKIN, J. A., FORD, T. W., CRAVENS, A. E., POWLEN, K. A., WANG, H. and MISHRA, V. Flash drought: A state of the science review. WIREs Water, 2024, 11(3): e1714. [Google Scholar] [CrossRef
[11] MAHTO, S. S. and MISHRA, V. Increasing risk of simultaneous occurrence of flash drought in major global croplands. Environmental Research Letters, 2023, 18(4): 044044. [Google Scholar] [CrossRef
[12] YUAN, X., WANG, L., WU, P., JI, P., SHEFFIELD, J. and ZHANG, M. Anthropogenic shift towards higher risk of flash drought over China. Nature Communications, 2019, 10(1): 4661. [Google Scholar] [CrossRef] [PubMed]
[13] CHRISTIAN, J. I., BASARA, J. B., OTKIN, J. A., HUNT, E. D., WAKEFIELD, R. A., FLANAGAN, P. X. and XIAO, X. A methodology for flash drought identification: Application of flash drought frequency across the United States. Journal of Hydrometeorology, 2019, 20(5): 833-846. [Google Scholar] [CrossRef
[14] KANG, H., SRIDHAR, V. and ALI, S. A. Climate change impacts on conventional and flash droughts in the Mekong River basin. Science of the Total Environment, 2022, 838(2): 155845. [Google Scholar] [CrossRef] [PubMed]
[15] LI, B., BEAUDOING, H., RODELL, M., NASA/GSFC/HSL. GLDAS catchment land surface model L4 daily 0.25 × 0.25 degree GRACE-DA1 V2.2. Greenbelt, Maryland, USA, Goddard Earth Sciences Data and Information Services Center (GESDISC), 2020.
[16] TOZER, B, SANDWELL, D. T., SMITH, W. H. F., OLSON, C., BEALE, J. R. and WESSEL, P. Global bathymetry and topography at 15 arc sec: SRTM15+. Earth and Space Science, 2019, 6: 1847-1864.[CrossRef
[17] FORD, T. and LABOSIER, C. Meteorological conditions associated with the onset of flash drought in the eastern United States. Agricultural and Forest Meteorology, 2017, 247: 414-423.[CrossRef
[18] CHEN, T. and GUESTRIN, C. XGBoost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining. New York: Association for Computing Machinery, 2016: 785-794.[CrossRef
[19] LUNDBERG, S. M. and LEE, S. A unified approach to interpreting model predictions. In Proceedings of the 31st international conference on neural information processing systems. New York: Association for Computing Machinery, 2017: 4768-4777.