Vegetation cover change and classification-based zonation of Fanjing Mountain using google earth engine

Abstract

This study utilizes the Google Earth Engine (GEE) platform, integrating Landsat and Sentinel-2 imagery from 1987 to 2022 alongside terrain data to develop a technical system for “vegetation cover inversion–classification–vertical belt spectrum analysis.” The Fractional Vegetation Cover (FVC) is estimated using a pixel dichotomy model, while spatiotemporal changes are evaluated through Theil-Sen trend analysis and Mann-Kendall tests. A multi-feature combination scheme has been designed, employing random forest models to enhance forest classification and analyze altitude zoning characteristics. The findings indicate that: (1) Over nearly 36 years, FVC in Fanjing Mountain has demonstrated a pattern of “fluctuating decline–adjustment–jump increase,” with 2007 serving as a pivotal turning point; notably, 41.65% of the area exhibited significant improvement, particularly in the southern region; (2) The classification model that integrates terrain features, red-edge index, and texture attributes (T9 scheme) achieves an overall accuracy of 77.88%, with terrain factors being the most influential in significantly enhancing the identification of shrub meadows and evergreen broadleaf forests; (3) Distinct vertical zoning of vegetation is observed: evergreen broadleaf forests dominate lower to mid-elevations (750–1550 m), deciduous broadleaf forests and coniferous forests are found at mid-to-high elevations, while shrub meadows are concentrated above 2400 m, forming a gradient succession sequence. This study provides methodological support and scientific evidence for ecological conservation efforts in subtropical mountain regions.