Water is a fundamental resource for life and the progression of human civilization. The escalating global population and its growing demand for clean water are posing significant challenges to the existing water resources. Regular monitoring of water bodies and large-scale estimation of Surface Water Quality Parameters (SWQPs) is crucial; conversely, remote sensing provides extensive spatial and temporal coverage. This study integrates satellite data, Artificial Neural Networks (ANNs), and ground truth water quality data for modelling SWQPs. Initially, atmospheric correction algorithms, including Fast Line of Sight Atmospheric Analysis of Hypercubes (FLAASH), Quick Atmospheric Correction (QUAC), Dark Object Subtraction (DOS), and Atmospheric Correction (ATCOR), have been employed to produce surface reflectance values of the water region. Then, a python code was developed to determine the coefficient of determination (R²) and Root Mean Squared Error (RMSE) values between each atmospherically corrected algorithm and the Landsat8 reference data. Secondly, this study compiles the most precise atmospherically corrected algorithm FLAASH with ANN to construct a Landsat8-based Backpropagation Neural Network (BPNN) model for modeling Total Organic Carbon (TOC) and Total Dissolved Solids (TDS). Thirdly, the developed BPNN models were trained with varying epochs (800 to 10,000) to assess their impact on the training process. Adjusting the epochs to 1200 guarantees that the models remain cost-effective and efficient while maintaining accuracy, thereby minimizing processing time and computational costs for future initiatives.