ENMAP HYPERSPECTRAL IMAGERY-BASED CROP TYPE CLASSIFICATION USING MACHINE LEARNING ALGORITHMS IN AN IRRIGATED AGRICULTURAL LANDSCAPE OF BUKHARA REGION, UZBEKISTAN

Accurate and timely crop type mapping is essential for agricultural monitoring, yield forecasting, and resource management in irrigated landscapes. This study evaluates the performance of six machine learning algorithms for classifying cotton and wheat crops using EnMAP hyperspectral satellite imagery (224 spectral bands, 400–2500 nm, 30 m spatial resolution) in the Tashkent region of Uzbekistan. Preprocessing included radiometric and atmospheric correction, Minimum Noise Fraction (MNF) and Principal Component Analysis (PCA) transformations, and spectral band selection. A spectral library of crop-specific signatures was constructed from field observations collected with the Field Maps mobile application. Classifiers evaluated include Random Forest (RF), Support Vector Machine (SVM), Gradient Boosting (GB), Decision Tree (DT), k-Nearest Neighbors (kNN), and Naïve Bayes (NB), assessed using 10-fold cross-validation on a 70/30 training–testing split. Random Forest achieved the highest overall accuracy (OA = 95%) and Kappa coefficient (K = 0.91), followed by SVM (OA = 94%) and Gradient Boosting (OA = 92%). The results confirm that hyperspectral data significantly enhance crop discrimination accuracy compared to conventional multispectral imagery.