生态环境学报 ›› 2026, Vol. 35 ›› Issue (5): 793-804.DOI: 10.16258/j.cnki.1674-5906.2026.05.012

• 研究论文【环境科学】 • 上一篇    下一篇

基于InVEST模型的三江源黄河流域水源涵养时空演变与驱动机制

石洪飞1(), 侯光良1,2,*(), 曹明珠1, 关佳萌1, 何家豪1, 唐中华1, 马曙光1   

  1. 1 青海师范大学地理科学学院青海 西宁 810008
    2 青海师范大学/青海省自然地理与环境过程重点实验室青海 西宁 810008
  • 收稿日期:2025-09-12 修回日期:2026-01-21 接受日期:2026-03-06 出版日期:2026-05-18 发布日期:2026-05-08
  • 通讯作者: *E-mail: Hgl20@163.com
  • 作者简介:石洪飞(2001年生),男,硕士研究生,研究方向为全球变化与人类适应。E-mail: 2058057578@qq.com
  • 基金资助:
    青海省科技厅基础研究计划项目(2025-ZJ-943M)

Water Conservation in the Three-River Headwaters Region of the Yellow River: Spatiotemporal Changes and Driving Forces Using the InVEST Model

SHI Hongfei1(), HOU Guangliang1,2,*(), CAO Mingzhu1, GUAN Jiameng1, HE Jiahao1, TANG Zhonghua1, MA Shuguang1   

  1. 1 College of Geographical Sciences, Qinghai Normal University, Xining 810008, P. R. China
    2 Qinghai Provincial Key Laboratory of Physical Geography and Environmental Processes/Qinghai Normal University, Xining 810008, P. R. China
  • Received:2025-09-12 Revised:2026-01-21 Accepted:2026-03-06 Online:2026-05-18 Published:2026-05-08

摘要:

为揭示三江源黄河流域水源涵养的时空演变特征及驱动机制,同时为该区域生态保护与水源涵养功能提升提供科学依据。基于InVEST模型(3.15.0版)评估了2000-2020年三江源黄河流域水源涵养时空变化及驱动机制。结果显示:1)时序上,水源涵养能力与涵养量呈“上升-下降-上升”波动,年均涵养深度为31.9-54.3 mm,涵养总量为11.2×108-19.6×108 m3。2005年达峰值,2015年为谷值,Mann-Kendall趋势检验未通过显著性水平(p=1.000)。空间上呈现“东南高、西北低”的稳定格局。2)土地利用中,草地贡献了73.7%-77.2%的涵养总量,是主要贡献者;Kruskal-Wallis检验表明不同土地利用类型间涵养能力存在极显著差异(p<0.001,ȵ2=0.961)。揭示了未利用地(以高寒裸地为主)因表层结皮与冻土隔水作用表现出较强持水能力,而水域在模型中表征值最低,提示生态管理需兼顾“功能质量”与“空间数量”。3)驱动机制上,降水量是主导因子(q值:0.548-0.653),但核心发现为降水量与土地利用类型的交互作用具有最强解释力(q值:0.712-0.794),且存在非线性增强效应。地形因子(如坡度)单独解释力较弱,但与降水交互后影响力显著上升,揭示其通过调控降水再分配间接影响涵养格局的次级驱动机制。综上,三江源黄河流域水源涵养演变是由水热条件主导、多因子非线性协同作用的结果,尤其在冻土环境下,下垫面属性对降水的响应机制是驱动过程的核心。

关键词: 水源涵养能力, InVEST模型, 地理探测器, 驱动因素分析, 土地利用

Abstract:

Understanding the spatiotemporal dynamics of water retention and its drivers is essential for the ecological management of the Three-River-Source Region. This study employed the InVEST model (v3.15.0) to quantify water retention changes and identify their driving mechanisms within the Yellow River source area from 2000 to 2020. Our analysis reveals three key findings. First, water retention exhibited a non-significant fluctuating trend over the study period, peaking in 2005 (19.6×108 m3) and reaching a low in 2015 (11.2×108 m3), with no statistically significant monotonic trend detected by the Mann-Kendall test (p=1.000). Spatially, a consistent pattern of higher retention in the southeast and lower in the northwest was maintained. Second, land use played a decisive role. Grassland contributed 73.7%-77.2% of the total retention volume, while highly significant differences in retention capacity were found among land use types (Kruskal-Wallis test, p<0.001). Notably, unused land (primarily alpine bare soil) showed unexpectedly high water retention, attributed to surface crusts and underlying permafrost, whereas open water bodies returned the lowest model values—a finding that underscores the need to manage both the functional quality and spatial coverage of ecosystems. Third, while precipitation was the primary individual driver (q-statistic: 0.548-0.653), the interaction between precipitation and land use type was the most influential factor (q-statistic: 0.712-0.794), demonstrating a nonlinear enhancement effect. Topographic factors like slope acted primarily as secondary drivers by interacting with precipitation to influence its spatial redistribution. We conclude that water retention in this alpine permafrost region is governed by a complex, nonlinear synergy where the land surface’s response to hydroclimatic forcing, rather than any single factor, is central to the process.

Key words: water conservation capacity, InVEST Model, geodetector, driving factors analysis, land use

中图分类号: