BLB_TVI: A hyperspectral triangular index for satellite-scale characterization of rice bacterial leaf blight severity

Bacterial leaf blight (BLB) is among the most destructive rice diseases in Asia. Although hyperspectral remote sensing has shown strong potential for crop disease detection, most existing studies focus on leaf-, canopy-, or UAV-scale disease assessment, and a BLB-specific hyperspectral index for satellite-scale severity characterization remains lacking. To address this gap, we developed a BLB-specific hyperspectral triangular vegetation index (BLB_TVI) from Zhuhai-1 imagery within a cross-scale framework integrating ground survey data and UAV observations. A satellite-pixel-consistent severity reference, termed the Rice Bacterial Leaf Blight Index (RBLBI), was constructed within this framework to bridge the mismatch between field observations and 10 m satellite pixels. The NDVI-based UAV classification achieved high accuracy, and the derived RBLBI was strongly correlated with the ground-based disease index (DI) (R 2 = 0.76). Among the tested indices, the optimal BLB_TVI, defined by the 686, 806, and 880 nm bands, showed the strongest relationship with RBLBI (R 2 = 0.77) and remained strongly associated with DI (R 2 = 0.73). In an independent cross-year and cross-site dataset collected in 2025, BLB_TVI also maintained a strong association with DI (R 2 = 0.70), indicating promising transferability. These results indicate that BLB_TVI provides an interpretable hyperspectral approach for satellite-scale BLB severity characterization.