Satellite-derived spectral signature and spatial mapping of makhana using Sentinel-2 for aquatic crop monitoring

Timely and accurate crop mapping is important for agricultural monitoring and resource management. In this study, multi-temporal Sentinel-2 imagery acquired during 2024 and 2025 was analysed using an unsupervised classification approach to generate 10 m spatial resolution distribution maps of makhana cultivation in selected districts of North Bihar, India. Temporal spectral reflectance patterns of makhana were examined across its growth period using Sentinel-2 multispectral observations. Vegetation indices showed progressive temporal variation, with NDVI, EVI, and NDRE increasing from 0.11 to 0.53, 0.10 to 0.77, and 0.03 to 0.32, respectively, between April and July, consistent with increasing vegetation cover. In contrast, NDWI, MNDWI, and NDBI decreased from −0.11 to −0.41, −0.10 to −0.25, and −0.01 to −0.19, respectively, indicating reduced open-water spectral contribution during canopy development. District-level mapping estimated makhana cultivation areas of 31,325 ha in 2024 and 37,878.7 ha in 2025 across the study region. The classification achieved an overall accuracy of 78.4%, demonstrating the potential of multi-temporal Sentinel-2 data for regional-scale identification of makhana cultivation. The derived spectral reflectance profiles and index trajectories provide observational insights into seasonal crop behaviour and contribute to the development of remote sensing approaches for monitoring floating aquatic cropping systems.