基于随机森林-三维特征空间法的西南山区石漠化演变格局及其驱动因子
作者:
作者单位:

作者简介:

通讯作者:

中图分类号:

基金项目:

国家自然科学基金项目(42471329);山东省高等学校“青创团队计划”团队项目(2022KJ224)


Spatiotemporal patterns and dominant factors of Karst Rocky Desertification in the southwestern mountainous region based on random forest and three-dimensional feature space
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 文章评论
    摘要:

    全球变化背景下,西南山区石漠化演变过程发生深刻变化,亟需明确其演变格局及其主导影响因子。基于MODIS时序数据产品,融合随机森林算法和三维特征空间模型构建了西南山区石漠化最优遥感监测模型,继而利用地理探测器等分析并解释了2000-2023年西南山区石漠化演变格局及其主导影响因子。结论如下:(1) KNDVI、Albedo、BSI、RBI 是西南山区石漠化最优表征参量,基于KNDVI-Albedo-BSI特征空间模型对于监测石漠化具有最好的实用性,总体精度88.39%,Kappa系数为0.8522,尤其在轻度、中度及极度石漠化等级中具有高稳定性和识别能力;(2)2000-2023年,西南山区石漠化以轻度和中度为主,主要分布于贵州省中西部、四川省与重庆市交界地带、云南省西南及东南部等区域;(3)2000-2023年,西南山区石漠化呈现改善趋势,其中2000-2010年西南山区石漠化呈现改善趋势,2010-2023年,则呈现总体改善,局部恶化趋势;(4)2000-2023年石漠化主要驱动因子发生显著变化,由2000年的自然因子协同主导,转变为2010年的人类活动与自然要素耦合效应增强和2023年的降水跃居首要驱动因子。研究成果可为西南山区石漠化综合治理与修改提供数据和决策支撑。

    Abstract:

    In the context of global change, rocky desertification in the southwestern mountainous region has intensified to some extent, making it urgent to clarify its evolution patterns and dominant driving factors. In this study, based on MODIS data products, an optimal remote sensing monitoring model for rocky desertification was constructed by integrating the Random Forest algorithm, three-dimensional feature space model, and geographical detectors. Subsequently, the evolution patterns and dominant driving factors of rocky desertification from 2000 to 2023 were analyzed and interpreted. The main conclusions were as follows: (1) KNDVI, Albedo, BSI, and RBI were identified as the optimal indicators for characterizing rocky desertification in the southwestern mountainous region. Among them, the KNDVI-Albedo-BSI feature space model demonstrated the highest practical applicability for monitoring rocky desertification, with an overall accuracy of 88.39% and a Kappa coefficient of 0.8522, especially exhibiting high stability and discrimination capability in mild, moderate, and extreme rocky desertification classes. (2) From 2000 to 2023, rocky desertification in the southwestern mountainous region was primarily characterized by mild and moderate levels, mainly distributed in the central and western parts of Guizhou Province, the border areas between Sichuan Province and Chongqing Municipality, and the southwestern and southeastern regions of Yunnan Province. (3) From 2000 to 2023, rocky desertification in the southwestern mountainous region showed an overall improving trend. Specifically, from 2000 to 2010, the region experienced a general improvement, while from 2010 to 2023, the trend continued with overall improvement but localized deterioration in certain areas. (4) The dominant driving factors of rocky desertification underwent significant changes between 2000 and 2023, shifting from co-dominance by natural factors in 2000, to a strengthened coupling effect of human activities and natural elements in 2010, and finally to precipitation becoming the primary driver in 2023. The findings of this study provided essential data and decision-making support for the comprehensive management and mitigation of rocky desertification in the southwestern mountainous region.

    参考文献
    相似文献
    引证文献
引用本文

刘盼盼,郭兵.基于随机森林-三维特征空间法的西南山区石漠化演变格局及其驱动因子.生态学报,2026,46(4):1737~1751

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数: