Assessing Global Terrestrial Ecological Resilience: Integrating Critical Slowing Down Theory and Convex Model with XGBoost ‐Driven Mechanism Analysis

Ecological resilience is a core indicator of an ecosystem's capacity to resist disturbances and sustain its functions, and it is a scientifically pivotal topic for advancing the Sustainable Development Goals (SDGs). However, existing research on ecological resilience is constrained by static data and measurement methodologies, which hinder the systematic and accurate characterisation of the spatiotemporal dynamics of global terrestrial ecological resilience and its nonlinear driving mechanisms. This study aims to address this research gap by proposing an improved ecological resilience assessment method based on Leaf Area Index (LAI) data derived from satellite remote sensing observations and exploring the spatiotemporal patterns, driving factors, and future changes of global terrestrial ecological resilience, thereby contributing to the theoretical framework of ecological resilience assessment and providing a practical tool for ecological restoration and rehabilitation. Global terrestrial ecosystems (assessed at the raster scale). Historical period: 2001–2021; future scenarios: based on multi‐scenario data from the Coupled Model Intercomparison Project Phase 6 (CMIP6). Not applicable (this study focuses on global terrestrial ecosystems as a whole, with the leaf area index (LAI) used as a state variable to characterise ecosystem status). An ecological resilience assessment method integrating critical slowing down theory and the convex model approach was proposed. LAI was leveraged as a state variable to quantify global terrestrial ecological resilience at the raster scale. The extreme gradient boosting (XGBoost) algorithm was used to explore the nonlinear responses of factors influencing ecological resilience. Future changes in global ecological resilience were simulated based on multi‐scenario data from the Coupled Model Intercomparison Project Phase 6 (CMIP6). A comprehensive analysis revealed notable regional heterogeneity and spatiotemporal dynamics of global ecological resilience. Most areas with high ecological resilience are distributed in low‐latitude humid zones. From 2001 to 2021, there was a distinct overall upward trend in global ecological resilience. The dominant factors shaping ecological resilience are potential evapotranspiration, vegetation cover and temperature—all of which exhibit significant threshold effects. Simulations of future scenarios confirmed the effectiveness of low‐carbon pathways in mitigating the loss of ecological resilience. Specifically, vulnerable areas demonstrated the highest levels of ecological resilience under the SSP1‐2.6 scenario, reaching a maximum value of 3.97. This study contributes to the existing theoretical framework for ecological resilience assessment in three key ways: (1) enabling the accurate identification of ecologically fragile areas (characterised by weak resistance) and ecologically degraded areas (characterised by declining resilience); (2) clarifying the spatial variation patterns of global terrestrial ecological resilience; and (3) providing a practical tool for ecological restoration and rehabilitation. By unveiling the spatial differentiation of global terrestrial ecological resilience, this systematic approach lays a scientific foundation for targeted ecological restoration and adaptive management.