Quantifying the necessity and efficacy of topographic correction on reflectance-based versus vegetation-index-based forest disturbance algorithms using Landsat time series

Topographic effects pose a significant challenge to accurate monitoring of forest disturbance in mountainous regions using Landsat time series, yet the actual benefits of topographic correction (TC) remain contentious. This study systematically evaluates the effectiveness of four widely used TC methods—Cosine Correction (CC), Sun-Canopy-Sensor + C (SCS+C), Illumination Correction (IC), and Path Length Correction (PLC)—on two categories of forest disturbance monitoring algorithms: reflectance-based (CCDC, COLD) and vegetation index (VI)-based (VCT, LandTrendr, mLandTrendr, BFAST). Based on extensive reference samples across diverse terrain conditions, our analysis reveals four key findings. First, topographic effects intensify with increasing slope steepness and shading. Second, all TC methods improved monitoring accuracy, with IC consistently performing best across algorithms. Third, improvement varied significantly by algorithm type and terrain: reflectance-based algorithms showed greater F1-score gains (e.g., up to 5.50% for CCDC) than VI-based ones, and enhancements were markedly larger on shaded versus sunlit slopes. Fourth, the necessity of TC is context-dependent: on sunlit slopes below 40°, TC offered minimal accuracy gains for most algorithms and may be omitted, whereas on shaded slopes steeper than 20°, TC is essential to maintain satisfactory accuracy. Nevertheless, even with correction, accuracy on steep shaded slopes (>40°) remained suboptimal, highlighting the limitations of current TC methods under extreme terrain. These findings demonstrate that the value of TC is not universal but is contingent on the specific algorithm and the local topographic context. This research delivers crucial, evidence-based guidance for developing best practices in mountain forest disturbance monitoring, advocating for a tailored approach that matches correction strategies with algorithm selection based on slope and aspect conditions.