Linking transect-based field observations with Sentinel-2 indices to characterize post-fire conditions in a Mediterranean Pinus brutia forest

Abstract

The field-level ecological counterparts of post-fire spectral indices remain poorly understood, particularly when surface-cover components are evaluated separately. This study evaluated the relationship between Sentinel-2-based dNBR, RdNBR, and NDVI indices and transect-based ground cover components in a Pinus brutia–dominated Mediterranean forest following the 2025 Köyceğiz-Akköprü fire. Surface conditions were recorded at 1-m intervals along two orthogonal 10-m transects in each of 73 plots, yielding 1601 point observations assigned to seven mutually exclusive surface-cover categories. Regression models showed that each index was associated with different field or environmental predictors. The NDVI model had the highest explanatory power and was positively associated with live cover (adj. R² = 0.439; β = +0.369). The dNBR model was associated primarily with pre-fire tree cover and slope. These results indicate that dNBR corresponds mainly to the pre-fire structural and topographic context, whereas NDVI corresponds to post-fire live cover and early recovery signals. The negative association between RdNBR and needle litter suggests that relative fire severity indices may be sensitive to residual organic-cover components. Spatial robustness analyses accounting for residual spatial dependence retained these principal associations. Consequently, dNBR, RdNBR, and NDVI are not interchangeable indicators representing the same aspect of post-fire surface conditions, but rather indices that complementarily reflect different ecological components. This study demonstrates that transect-based land cover measurements in Sentinel-2-based post-fire assessments provide a practical field-reference approach that improves the ecological interpretation of spectral indices.