Monitoring surface water quality is critical for safeguarding public health, supporting sustainable development, and ensuring ecosystem integrity. Traditional water quality assessment methods are often limited by spatial coverage, time constraints, and high operational costs. This study leverages advanced machine learning algorithms and spectral reflectance data to provide a scalable, accurate, and automated solution for real-time assessment of key water quality indicators including chlorophyll-a, cyanobacteria, green algae, diatoms, cryptophyta, and colored dissolved organic matter (CDOM). Multiple regression models including linear regression, support vector regression, and ensemble methods such as Random Forest, Gradient Boosting, Extra Trees, and XGBoost were systematically evaluated using R \(^{2}\) scores to determine their predictive performance. Feature selection based on correlation analysis enhanced both computational efficiency and interpretability. Results demonstrate that ensemble models, particularly Extra Trees and Random Forest, achieved superior accuracy with R \(^{2}\) scores exceeding 0.93 for most parameters, while simpler models underperformed in capturing the nonlinear relationships inherent in spectral data. These findings underscore the transformative potential of integrating machine learning and spectral sensing for proactive, cost-effective surface water quality management, thus supporting global efforts toward the United Nations Sustainable Development Goals.