Answers are grounded in this paper and cite their sources.
Tropical and subtropical forests harbor over half of the world’s terrestrial biodiversity and account for a large fraction of the global terrestrial carbon sink. However, under accelerating climate change and human pressure, the dynamics of these forests remain incompletely characterized, particularly in rapidly developing regions where multiple change processes—fire, clear-cutting, afforestation, and deforestation—operate simultaneously. Using 30 m Landsat time-series imagery spanning 1993–2021, fused with China’s annual land-cover dataset (CLCD) and the continuous change-detection algorithm (CCDC), this study constructed a unified, multi-type forest-change dataset for tropical and subtropical China. The spatiotemporal patterns of four dominant change types were quantified through grid-based intensity analysis and the Mann–Kendall trend test, and associations with annual intensity classes were examined using an eXtreme Gradient Boosting (XGBoost) ensemble learner combined with SHapley Additive exPlanations (SHAP) model interpretation. Our results reveal the following: First, spatially, fire and clear-cutting both exhibit a southeast–southwest dual-core aggregation pattern, while afforestation and deforestation show highly congruent, largely dispersed distributions dominated by the moderate period-summary intensity class. Second, temporally, fire shows an overall declining trend, with 52.1% of grid cells exhibiting significant or non-significant decreases; clear-cutting shows significant increases in 28.4% of grid cells. Afforestation and deforestation are both dominated by significant and non-significant increases: 48.7% and 62.8% of grid cells, respectively. For deforestation, 44.8% of the grids included in the peak-year analysis reached their maximum annual intensity during 2011–2015, although this concentration may partly reflect temporal inconsistencies in the underlying CLCD data. Third, the annual intensity-class models identify potential evapotranspiration as a leading predictor for fire and mean temperature as an important environmental covariate for clear-cutting. Slope and population are important predictors for both afforestation and deforestation. In the spatial summaries of SHAP contributions to the annual high-intensity class output, slope is a prominent negative contributor, particularly for afforestation and deforestation. This study provides a unified, multi-type, multi-decade analysis of forest change in tropical and subtropical China, characterizes nonlinear associations with annual intensity-class outputs, and maps their dominant positive and negative SHAP contributors to support precision forest management and climate-adaptive ecological policy design.
No takes yet. Share an insight, caveat, or question.
Wang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: