This research introduces VIMALA, a hybrid deep learning architecture named AquaSpatioTemporalNet, aimed at monitoring and forecasting changes in water bodies across Gujarat using satellite data from Sentinel-2 and S-2 Harmonized. The model combines convolutional neural networks (CNNs) for spatial feature extraction, long short-term memory (LSTM) networks for capturing temporal variations, and transformers to handle long-range dependencies in time-series data. The preprocessing pipeline includes cloud masking, NDWI calculation, clipping to the Gujarat region, band selection (B3 and B8), and atmospheric correction to ensure high-quality inputs. Key indices like the Normalized Difference Water Index (NDWI) and Normalized Difference Vegetation Index (NDVI) are employed to detect and quantify water bodies and differentiate them from surrounding vegetation. The AquaSpatioTemporalNet architecture demonstrates improved performance over traditional models by achieving higher accuracy in predicting water body dynamics, as evidenced by precision, recall, and reduced mean squared error (MSE). The system provides valuable insights into the temporal evolution of water resources, supporting more effective decision-making for water resource management. The proposed methodology offers a robust solution for large-scale environmental monitoring and can be applied to other geographic regions facing similar challenges.