Topographic heterogeneity and long-term vegetation dynamics in a humid mountain landscape: Evidence from a Shannon entropy-based framework

Topographic heterogeneity can mediate radiation, moisture redistribution, and microclimate, yet it is commonly represented by single terrain variables. We evaluated long-term vegetation dynamics across low, moderate, and high topographic heterogeneity (TH) classes in Trabzon, northeastern Türkiye, using a 5 m-grid Shannon entropy-based Topographic Heterogeneity Index (THI), Landsat NDVI/EVI (1985–2025), CORINE land cover, elevation strata, and TerraClimate PDSI (1985–2024). High TH maintained higher mean productivity than low TH, with a larger contrast for EVI (32.3%) than NDVI (9.8%); differences were strongest below 2000 m and converged at higher elevations. All classes greened significantly, although the moderate class showed the largest trend magnitude, whereas high TH showed lower relative temporal variability. After controlling class-specific temporal trends and clustering inference by year, current-year PDSI effects were not supported. Previous-year PDSI slopes differed among TH classes; the only individually significant slope was a negative EVI association in low TH (β = −0.00524, p = 0.0167), while NDVI class-specific slopes were not significant. Trend-adjusted drought metrics showed event-specific responses rather than a uniform drought-year decline; recovery and resilience ratios remained above 1 but were based on four events. A raster-scale sensitivity analysis retained 77.5% agreement with the 5 m classification at 30 m (κ = 0.586), declining to 72.0% at 100 m (κ = 0.468). The results support a robust association of terrain complexity with vegetation productivity and relative stability, while emphasizing limits related to spatial scale and regional drought indices.