Identification and Prioritization of Climate-Adaptive Types for Territorial Ecological Restoration Zone in the Chengdu-Chongqing Urban Agglomeration

Objective Climate change has intensified global social risks and ecological crises. The increasing frequency of climate hazards such as heatwaves and extreme precipitation, combined with the persistent degradation of forests, soils, and other ecosystems, has become a critical constraint on the sustainable development of human societies. In response to climate change, the Intergovernmental Panel on Climate Change (IPCC) has proposed two overarching strategies: climate mitigation and climate adaptation. Climate adaptation refers to reducing climate-related risks by decreasing human vulnerability or safeguarding ecosystem service functions, and ecological restoration is widely regarded as one of the key approaches to climate adaptation. Ecological restoration can effectively enhance ecosystem services, such as increasing carbon sequestration, reducing surface temperature, and improving water conservation capacity—thereby demonstrating substantial potential for strengthening climate adaptability. However, although existing standards and practices of territorial ecological restoration repeatedly emphasize the integration of climate adaptation considerations, the specific types of climate adaptation within territorial ecological restoration, as well as the pathways and methods for identifying restoration areas and determining their priority with climate adaptation as a core objective, remain insufficiently defined. Therefore, clarifying climate adaptation types relevant to territorial ecological restoration and developing robust identification and priority-setting pathways and methods constitute an urgent research agenda. Methods Taking the Chengdu-Chongqing urban agglomeration as the study area, from an exposure−sensitivity−resilience perspective, three types of territorial ecological restoration zones for climate adaption are defined: climate-exposed restoration zone (CERZ), climate-sensitive restoration zone (CSRZ), and climate-resilient restoration zone (CRRZ). The research route of this study consists of three major steps. 1) Based on the concepts of CERZ, CSRZ, and CRRZ and previous research findings, sample datasets and a double-layer indicator system comprising a “core-characteristic factors” were established to construct the machine learning dataset. 2) An ensemble machine learning model integrating three algorithms—maximum entropy (Maxent), random forest (RF), and categorical boosting (CatBoost)—through logistic regression (LR) was developed. The model performance was evaluated, and the spatial extents and restoration priorities of CERZ, CSRZ, and CRRZ were identified. 3) Differentiated response strategies were proposed according to the identification results and characteristics of CERZ, CSRZ, and CRRZ. Results 1) Compared with the single-layer system, the double-layer framework improves the identification performance by approximately 5%, and the ensemble model improves it by around 3% compared with individual algorithms. 2) Both CERZ and CSRZ show more than 20% of areas at medium or higher restoration priority, indicating an urgent need for ecological restoration in the study area. Meanwhile, CRRZ with medium or higher restoration priority accounts for over 50% of the Chengdu-Chongqing urban agglomeration, suggesting that regional resilience enhancement remains a long-term and challenging task. The spatial patterns of restoration priority for the three zones exhibit the following characteristics. For CERZ, the proportions of restoration priority from high to low are 4.98%, 10.56%, 13.38%, 19.77% and 51.31%. Restoration areas above medium priority total 53,020.54 km2, mainly distributed in the karst regions of southern Sichuan and the Three Gorges Reservoir Area in northeastern Chongqing. For CSRZ, the proportions from high to low are 4.63%, 9.49%, 12.74%, 36.31% and 36.83%. Restoration areas above medium priority total 49,252.83 km2, exhibiting more fragmented patches and lower clustering than CERZ. Unlike CERZ, restoration priority increases within medium and large cities, especially in areas characterized by fragmented vegetation and dense river nets. For CRRZ, the proportions from high to low are 11.50%, 26.60%, 26.27%, 18.61%, 17.02%. Restoration areas above medium priority total 118,035.92 km2, showing both concentrated and dispersed patterns across low-altitude plains and hilly regions in the central parts of the Chengdu-Chongqing urban agglomeration. 3) CERZ aims to integrate engineering, ecological, and economic measures to effectively reduce the direct damage caused by extreme climate events. CSRZ aims to take river basins or key ecological function areas as spatial units to promote the recovery of ecological functions and the construction of ecological networks. CRRZ aims to advance the coordinated development of a green economic system and green infrastructure networks. Conclusion By building double-layer indicator system and ensemble machine learning model, this study clearly reveals the differentiated patterns of territorial ecological restoration in the Chengdu-Chongqing urban agglomeration from an exposure−sensitivity−resilience perspective. The findings provide a replicable framework and technical reference for the scientific and rapid identification of the key areas and nodes of territorial ecological restoration, optimizing restoration layouts and resource allocation, and enhancing regional climate adaptability. In the future, our research may incorporate higher-resolution datasets of samples and indicators to further refine the identification and planning of territorial ecological restoration zones.