Sentinel-2 Enhanced Crop Classification: A Time-Weighted Dynamic Time Warping Methodology with Limited Samples

Crop classification serves as a pivotal component in agricultural management, offering crucial insights into crop vitality, distribution, and yield estimation. The adoption of remote sensing methodologies has emerged as a pivotal approach for expansive and timely crop delineation, effectively circumventing the constraints encountered by conventional methods. This study introduces an innovative strategy for crop categorization in Krishna district, India, employing the Time-Weighted Dynamic Time Warping (TWDTW) and Dynamic Time Warping (DTW) technique. Through the Sentinel-2 satellite data, alongside supplementary datasets like DEM and water masks, the methodology adeptly addresses challenges linked to cloud cover and land surface disparities. The proposed technique yields promising outcomes in identifying six major crop classes, encompassing rice, sugarcane, maize, cotton, and other crops, alongside delineating water bodies. The amalgamation of spectral and temporal attributes highlights the potential for precise and dependable crop mapping, thereby contributing significantly to precision agriculture and land management strategies.