Monitoring water and wastewater quality is crucial for environmental management and public health, traditionally relying on physicochemical analyses such as biochemical oxygen demand (BOD) and chemical oxygen demand (COD). However, these conventional methods are time-consuming, costly, and generate toxic waste. In this study, ultraviolet-visible (UV-Vis) and excitation-emission matrix (EEM) fluorescence spectroscopy were combined with chemometric approaches to provide a rapid and sustainable alternative for effluent monitoring. Partial least squares (PLS), unfolded PLS (UPLS), and support vector regression (SVR) models were evaluated using treated wastewater (TWW), raw wastewater (RWW), and their combination. Results demonstrated that UPLS with EEM spectra achieved the best performance for BOD prediction ( R 2 p = 0.845), while SVR with 2D fluorescence at 290 nm yielded competitive results for COD ( R 2 p = 0.738). Importantly, combining TWW and RWW samples enhanced model robustness, highlighting the role of dataset variability in improving predictive accuracy. Overall, EEM fluorescence spectroscopy coupled with multivariate regression offers a fast, cost-effective, and environmentally sustainable strategy for estimating BOD and COD in wastewater.
Paulo et al. (Sat,) studied this question.