Spatial Evolution and Driving Mechanisms of Vegetation Cover in China’s Greater Khingan Mountains Based on Explainable Geospatial Machine Learning
Remote Sensing · 10.07.2025 · 17 (14) · 2375
DOI: 10.3390/rs17142375 ↗Kurz gefasst
Fachartikel in der Zeitschrift Remote Sensing, online erschienen am 10.07.2025. Verfasst von Zihao Wang und 3 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 62 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Key R&D Program of China, Key R&D Program of Inner Mongolia, China, Inner Mongolia Natural Science Foundation, China.
Abstract (Originalsprache)
As a crucial ecological barrier in China, the Greater Khingan Mountains play a vital role in global ecological security. Investigating the spatiotemporal variations in fractional vegetation cover (FVC) and the driving mechanisms behind its spatial differentiation is essential. This study introduced a KNDVI-XGeoML framework integrating the Kernel NDVI and explainable geospatial machine learning to analyze the FVC dynamics and the mechanisms driving their spatial differentiation in China’s Greater Khingan Mountains, based on which targeted ecological management strategies were proposed. The key findings reveal that (1) from 2001 to 2022, FVC showed an increasing trend, confirming the effectiveness of ecological restoration. (2) The XGeoML model successfully revealed nonlinear relationships and threshold effects between driving factors and FVC. In addition, both single-factor importance and inter-factor interaction analyses consistently showed that landform factors dominated the spatial distribution of FVC. (3) Regional heterogeneity emerged—human activities drove the northern alpine zones, while landform factors governed other areas. (4) The natural-environment-dominated zones and human-activity-dominated zones were established, and management strategies were proposed: restricting tourism in low-altitude zones, optimizing the cold-resistant vegetation at high elevations, and improving the southern soil conditions to support ecological barrier construction. The innovation lies in merging nonlinear vegetation indices with interpretable machine learning, overcoming the traditional limitations in terms of saturation effects and analyses of spatial heterogeneity. This approach enhances our understanding of high-latitude vegetation dynamics, offering a methodological advancement for precision ecological management. The spatial zoning strategy based on dominant drivers provides actionable insights for maintaining this critical ecological barrier, particularly under climate change pressures. The framework demonstrates strong potential for extrapolation to other ecologically sensitive regions requiring data-driven conservation planning.
Förderung
- National Key R&D Program of China
- Key R&D Program of Inner Mongolia, China
- Inner Mongolia Natural Science Foundation, China
- Forestry Discipline Self-Established Research Projects of Inner Mongolia Agricultural University, China
- Research Team Construction Project of Forestry College, Inner Mongolia Agricultural University, China
Zitieren
APA
Wang, Z., Wang, B., Zhang, Q., & Lü, C. (2025). Spatial Evolution and Driving Mechanisms of Vegetation Cover in China’s Greater Khingan Mountains Based on Explainable Geospatial Machine Learning. Remote Sensing, 17(14), 2375. https://doi.org/10.3390/rs17142375
Vancouver
Wang Z, Wang B, Zhang Q, Lü C. Spatial Evolution and Driving Mechanisms of Vegetation Cover in China’s Greater Khingan Mountains Based on Explainable Geospatial Machine Learning. Remote Sensing. 2025;17(14):2375. doi:10.3390/rs17142375
BibTeX
@article{Wang2025,
author = {Wang, Zihao and Wang, Bing and Zhang, Qiuliang and Lü, Changwei},
title = {Spatial Evolution and Driving Mechanisms of Vegetation Cover in China’s Greater Khingan Mountains Based on Explainable Geospatial Machine Learning},
journal = {Remote Sensing},
year = {2025},
volume = {17},
number = {14},
pages = {2375},
doi = {10.3390/rs17142375}
}
EUVIMED