Features Extraction (FE) plays a crucial role in image classification by reducing the dimensionality of the raw hyperspectral Remote Sensing (RS) data while retaining discriminative information. This technique helps to simplify complex hyperspectral data, which contain hundreds of spectral bands, to make them more manageable by identifying the most important information for classification. Reducing the number of dimensions, this helps to overcome the problem of the "curse of dimensionality", improves classification accuracy, and speeds up data processing. This study proposed an innovative approach to improve the accuracy of water body extraction from hyperspectral RS data by combining FE and Convolutional Extreme Learning Machine (CELM) with evolutionary algorithms. This method integrates several advanced techniques to optimize water surface extraction. FE allows us to select the most relevant information from hyperspectral data, reducing complexity while preserving essential details. The addition of evolutionary algorithms allows us to automatically optimize the model parameters, improving its performance. CELM is trained in a supervised manner directly on raw data to learn discriminative features for classification steps. Then, these extracted features are used for the final classification using the CELM with the hybridization of evolutionary algorithms (EAs) such as Genetic Algorithms (GAs). This hybrid approach aims to overcome the challenges related to the spectral variability of water bodies and the presence of mixed pixels, thus offering a more robust and accurate solution for water resource mapping from hyperspectral images. In order to validate the effectiveness of our approach, we conducted experiments on hyperspectral data acquired by the RISMA (PRecursore IperSpettrale della Missione Applicativa) satellite. The obtained results were then compared with the existing methods documented in the scientific literature using recognized evaluation metrics such as precision, accuracy, recall, Intersection Over Union (IOU), and F1 score.