Recreation service (RS) in heritage-based tourism is crucial for redefining the value of heritage resources. Yet, existing studies often neglect public perception-driven RS variation, limiting the future functionality of the scenic areas and the heritage resource valorization. Taking scenic areas in Hunan province, China, as a case study, we developed an RS assessment framework based on big data to evaluate RS and the comprehensive recreation service index, followed by K-means clustering for scenic area zoning. Key RS determinants were identified using XGBoost. Key findings include: (1) 56 recreation activities, 8 RSs, and 8 recreation scenic areas were identified; (2) Key determinants include popularity (importance = 0.114), proximity to cities (0.246), and adjacency to renowned scenic areas (0.355). This framework addresses the gap that focuses solely on spatiotemporal patterns while neglecting RS typology differences in scenic areas, and provides evidence-based insights to enhance heritage utilization and sustainable tourism in scenic areas.

Wildfire regimes in mountain forest landscapes are shaped by complex interactions among topography, climate, vegetation, and human activity. However, predicting and interpreting fire occurrence in topographically heterogeneous regions remains challenging because fire–environment relationships vary strongly across elevation gradients and temporal scales. This study developed a hybrid machine-learning framework integrating an Information Value Model (IVM), Random Forest (RF), and Convolutional Neural Network (CNN) to decode elevation-mediated wildfire regimes in western Sichuan, China, a mountainous forest region characterized by strong vertical environmental gradients and high ecological conservation value. Multi-source datasets, including Moderate Resolution Imaging Spectroradiometer (MODIS) burned-area records, topographic variables, monthly meteorological data, vegetation indices, land-cover information, and human-accessibility proxies, were integrated at a 500 m spatial resolution. Environmentally comparable non-fire samples were generated from unburned vegetated pixels, and model training, RF-based feature selection, hyperparameter tuning using Particle Swarm Optimization (PSO), and performance evaluation were conducted within a nested spatial block cross-validation framework. The model produced continuous wildfire occurrence probabilities and showed strong discriminatory performance under the adopted validation protocol, with AUC values exceeding 0.95 across temporal datasets and low probability-error metrics. RF importance and correlation analyses identified mean temperature, elevation, and precipitation as the dominant predictors of wildfire probability. Spatial analyses revealed pronounced elevation-mediated differentiation in wildfire regimes: low-elevation valleys showed higher fire probability and stronger associations with human-accessibility proxies, whereas high-elevation plateau areas exhibited lower and more scattered fire patterns associated with climatic constraints. Seasonal and monthly analyses further showed that winter and spring fires dominated the regional fire regime, with risk intensifying during the pre-monsoon dry period. By combining probabilistic fire-risk mapping, spatial-context learning, and elevation-gradient interpretation, this study provides a transferable framework for understanding wildfire regimes in complex mountain forest landscapes. The findings support adaptive forest fire management, targeted monitoring, and risk zoning in mountainous regions where forest ecosystems, human activities, and conservation values intersect.

Urban agglomerations are complex adaptive social–ecological systems, and understanding the spatiotemporal evolution of landscape ecological resilience is important for their sustainable management. This study extended the previously developed Risk–Potential–Connectivity (RPC) framework for the Changsha–Zhuzhou–Xiangtan Urban Agglomeration (CZXUA) by identifying adaptive cycle categories at the grid scale using Local Moran’s I and explicit classification rules. Previously generated SD–PLUS land use projections were used to evaluate future changes in the RPC dimensions, integrated resilience, and adaptive cycle composition. (1) From 2000 to 2020, mean landscape ecological resilience decreased from 0.3642 to 0.3345, representing a net decline of 8.15%. Global Moran’s I increased from 0.746 to 0.787, indicating stronger spatial clustering. The spatial pattern remained relatively stable, with higher resilience in peripheral ecological areas, lower resilience in urban cores, and a persistent northwest–southeast orientation. (2) Adaptive cycle composition changed only moderately. The conservation (K) category accounted for more than 45% of the study area, while the release (Ω) category accounted for approximately 19% and was concentrated mainly in urban cores. The exploitation (r) category declined after 2010, whereas the smaller reorganization (α) and transitional (t) categories were more sensitive to classification settings. (3) Using previously generated land use projections as scenario inputs, future RPC responses differed clearly among scenarios. SSP126 maintained more intact ecological patches, relatively stable connectivity, and the smallest resilience change, with increases in r and K. SSP245 showed the greatest connectivity decline, expansion of areas with declining resilience, a reduction in r, and increases in Ω and α. SSP585 produced the highest landscape ecological risk, the largest declines in ecological potential and integrated resilience, reductions in r and K, and increases in α and t. These conditional results indicate that limiting construction land expansion, protecting ecological land, and maintaining ecological connectivity may reduce resilience losses under future urban development. The RPC adaptive cycle framework can support the identification of priorities for ecological protection, restoration, connectivity enhancement, and risk reduction.