Topographic effects amplify forest disturbances detected by yearly wide-time-window Landsat time series

Widespread topographic effects in remote sensing images of mountainous regions with rugged terrain severely hinder satellite-based global land surface monitoring and change detection. However, Landsat time series (LTSs) disturbance detection remain highly debated unquantified. In this study, we proposed a novel postprocessing approach to quantify the impact on long time-series (1989 − 2020) forest detection, taking Hengduan Mountains Region (HDMR) southwest China as an example. This applied pixel-by-pixel simulation based semiempirical sun-canopy-sensor C-correction (SCS+C) model identify remove topography-induced disturbances from LandTrendr-detected disturbances. Filtering detected different patch sizes was conducted, its effectiveness removing quantitatively evaluated. The results showed that 10.43% total area topography-induced. Removing increased traditional area-adjusted overall accuracies (OAs) map by 3.50% 0.62%. Topography-induced occurred mainly northwest-facing (300°−360°) slopes gradient 30°−55° higher latitude between day year (DOY) pairs (such DOYs 90, 330, 360) considerable disparity solar azimuth zenith angle. outperformed spatial noise filtering minimum mapping unit (MMU), which considerably decreased percentage at expense decreasing producer’s accuracy (PA) detecting disturbed forests. study highlighted importance shed new light promising potential correction method replace preprocessing LTSs globally.