To classify water regions in satellite imagery of flood-affected areas, our study compares three machine learning algorithms: Support Vector Machine (SVM), Random Forest, and K-means clustering with our proposed methodology which categorizes water, buildings, and vegetation areas and categorizes as flooded and non-flooded images. Accuracy, precision, recall, and F1 score were used to assess the algorithms, with NDWI (Normalized Difference Water Index) serving as a crucial component. To make feature extraction easier, a sample dataset of flood photos was processed, which included image transformation and resizing. Training data from Kaggle (Flood Area Segmentation section) was randomly sampled and tested to improve computing performance. When it came to recognizing water zones, the Random Forest model performed better than SVM, with an accuracy of 99%. K-Means clustering offered a baseline for comparison at 94%. The findings demonstrate how sophisticated machine learning methods may be used to effectively classify water bodies, enhancing flood monitoring and control.