Answers are grounded in this paper and cite their sources.
Wildfires have increased in frequency and severity across the Mediterranean ecosystems of central Chile, emphasizing the need for accurate vegetation fuel mapping to support wildfire risk assessment and management. This study evaluated the potential of Sentinel-1 Synthetic Aperture Radar (SAR) imagery to discriminate vegetation fuel classes defined by the Chilean KITRAL fuel classification system. A total of 3,645 reference polygons were delineated through detailed photo-interpretation and subsequently validated by field surveys to classify four fuel classes: native forest, forest plantation, shrubland, and grassland. Three Sentinel-1 backscatter variables (VV, VH, and the VV/VH ratio) and Haralick texture metrics were evaluated under three predictor scenarios: (A) texture metrics only, (B) backscatter variables only, and (C) the combination of backscatter variables and texture metrics. Classification performance was evaluated for four machine learning algorithms: Random Forest (RF), Support Vector Machine (SVM), Generalized Linear Model (GLM), and Artificial Neural Network (ANN). Backscatter variables consistently outperformed texture metrics when used independently, demonstrating that radar backscatter captures the structural characteristics most relevant for vegetation fuel discrimination. However, integrating backscatter variables and texture metrics yielded the highest classification performance across all algorithms, demonstrating the complementary nature of these predictors sets. Among the evaluated models, SVM achieved the best overall performance (Accuracy = 0.560, Kappa = 0.414, Macro-F1 = 0.552). These results demonstrate that combining Sentinel-1 backscatter variables with texture metrics enhances the discrimination of structurally similar vegetation fuel classes and provides a robust, operational workflows for vegetation fuel mapping in Mediterranean ecosystems of central Chile.
No takes yet. Share an insight, caveat, or question.
Vildo et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: