Accurate water body extraction from satellite imagery is important for applications such as hydrological modeling, disaster management, and environmental monitoring. This paper introduces a novel method combining clustering and classification algorithms simultaneously for efficient and precise water body detection. The proposed approach integrates an advanced Iterative Self-Organizing Data Analysis Technique (ISODATA) clustering algorithm with spectral water indices, namely the Automated Water Extraction Index ( \(AWEI_{nsh}\) ) and the Modified Normalized Difference Water Index (MNDWI), and a lazy k-Nearest Neighbors (k-NN) classifier. By utilizing a reference image generated from \(AWEI_{nsh}\) , the method iteratively refines water and non-water clusters through neighborhood-based classification. The results demonstrate that the proposed algorithm outperforms traditional spectral indices and approaches like U-Net in terms of execution time and accuracy, achieving up to 96% accuracy on diverse datasets, including the Barak River and Sunderban delta regions. Also, Simultaneous Clustering Classification (SCC) achieves an F1-score of up to 95.06% compared to 87.83% for U-Net, and an IoU of 90.41% compared to 78.30%, reflecting gains of approximately 7.23% in \(F_{1}\) score and 12.11% in IoU. Unlike deep learning models, this method requires no pre-training, making it computationally efficient and suitable for real-time applications. The study highlights the robustness of SCC in overcoming limitations of conventional and deep-learning-based techniques for water body extraction.
