Forest disturbance plays a major negative role in maintaining the integrity and stability of ecosystem functions. Existing researches have largely focused on disturbance events causing abrupt spectral changes, such as fires or logging, while paying insufficient attention to gradual forest dynamics caused by chronic stressors like drought or overgrazing. Although the long-term observational technologies have made it possible to detect such gradual change events, quantifying the complex, non-linear driving mechanisms and the degree differences in forest greening and browning remains challenging currently. In response, this study leveraged the Google Earth Engine platform to analyze Landsat time-series data spanning from 1987 to 2022 in Dali Prefecture, China. This work integrated the Continuous Change Detection and Classification (CCDC) model to characterize both abrupt and gradual forest dynamics. Subsequently, a persistent forest mask, generated from the Vegetation Change Tracker algorithm, was applied to separate gradual greening and browning trends, followed by a quantitative characterization of forest greening and browning. On this basis, LightGBM and SHapley Additive exPlanations (SHAP) were employed to investigate the underlying driving mechanisms of those gradual changes. The results indicated that CCDC achieved high accuracy in detecting abrupt changes (Overall Accuracy = 0.93), while the identified gradual changes aligned well with local forest patrol records. The Enhanced Vegetation Index was proved to be the optimal index for monitoring gradual changes, and the region exhibited an overall forest greening trend. Attribution analysis revealed that forest greening was primarily driven by temperature, elevation and slope, whereas forest browning was mainly influenced by elevation, soil bulk density and soil organic carbon content. The maps and analytical insights generated in this study provide critical data support for formulating targeted forest management strategies. • Precise quantification of gradual forest change using harmonic analysis. • Dali Prefecture's persistent forest shows a greening trend, with greening areas far exceeding browning. • Key factors affecting forest greening (TMP, elevation and slope) and browning (elevation, SBD and SOCC).
Yin et al. (Tue,) studied this question.