DOI: https://doi.org/10.36347/sjavs.2026.v13i04.002

Pages: 68-79

Accurate identification of perennial fruit orchards is essential for strengthening horticultural statistics, precision management and policy planning in eastern India, where mango (Mangifera indica L.) and litchi (Litchi chinensis Sonn.) constitute major fruit-based livelihoods. In Bihar, orchard inventories remain largely field-based, and quantitative evaluation of species-level spectral separability using medium-resolution satellite data is limited. This study assesses the spectral distinctness of mango and litchi canopies using single-date Sentinel-2 Level-2A surface reflectance imagery acquired on 14 May 2025 over Gopalganj district, Bihar. Eight bands (B2, B3, B4, B5, B6, B7, B8 and B11) were analysed, with 20 m bands resampled to 10 m, and reflectance statistics were derived from 40 digitised training polygons per species. Multivariate Bhattacharyya distance and its Jeffries–Matusita (JM) transformation were used to quantify separability. The resulting JM value of 1.9588 (0–2 scale) indicates excellent separability and very low expected classification confusion. Band-wise decomposition shows that visible wavelengths dominate separability, with B4 (red) contributing 27.29%, B2 (blue) 20.30%, B3 (green) 18.51% and B5 (red-edge) 15.90%, while SWIR1 (B11) contributes 10.32% and near-infrared bands contribute comparatively less. These patterns are consistent with the spectral reflectance curves, where litchi exhibits higher reflectance in the visible region and mango shows stronger near-infrared scattering. The findings demonstrate that Sentinel-2 surface reflectance, particularly the visible and early red-edge bands, can effectively distinguish mango and litchi orchards in smallholder landscapes, providing a basis for operational fruit-crop mapping, orchard inventory and targeted management in subtropical agro-ecosystems.