Utilizing In Situ Ultraviolet-Visual Spectroscopy to Measure Nutrients and Sediment Concentrations in Stormwater Runoff

The capacity to collect meaningful data to estimate stormwater runoff water quality and subsequent system removal performance is key to selecting the appropriate solutions to protect water resources. Historically, grab sampling and automated composite sampling approaches have been used with training and comprehensive quality assurance protocols to produce defensible data. Innovative approaches use real-time ultraviolet-visual spectrometry (UV-Vis) can provide a powerful tool for understanding pollutant loading regionally. This study found that real-time UV-Vis sensing is a potential new tool for understanding stormwater effluent pollutant dynamics. Researchers compared data from real-time sensing using UV-Vis spectrometers to develop calibration curves for predicting pollutant concentrations in stormwater flows. Results from paired laboratory data and raw spectral data established calibrations for the stormwater runoff composition and support further investigations of the use of this technology to predict in situ concentrations of total Kjeldahl nitrogen (TKN), dissolved organic carbon (DOC), total phosphorus (TP), total suspended solids (TSS), total nitrogen (TN), and nitrate as nitrogen (${\mathrm{NO}}_3-\mathrm{N}$) resulted in robust models with $R^2$ values in the range 0.99–0.93. Using partial least squares regression (PLSR) methods, the study demonstrated a strong correlation between concentrations generated by the raw absorbance data across the full available spectrum (220 to 730 nm). These results indicate the potential for developing specific stormwater calibration curves for pollutants of interest representative of stormwater runoff. Collectively, these results indicate that real-time UV-Vis spectrometers can redefine stormwater control monitoring by potentially delivering more accurate, more repeatable laboratory quality data instantaneously, with greater efficiency.

Results from single grab samples of stormwater events are insufficient to develop estimates of storm event mean concentrations (EMCs). Commonly grab samples are taken during the rising limb of a hydrograph and used to characterize first flush phenomena. It is now known that first flush is a simple way to think about pollutant build-up and wash-off dynamics; however, it is a more theoretical concept because actual runoff concentrations for different pollutants vary due to many watershed and pollutant characteristics. Analyzing individual samples may be helpful to identify trends in loading rates of various constituents over a storm event and their changing intensities, but this significantly adds to laboratory costs and personnel hours to manage and process each event. Composite samples, on the other hand, offer affordable analytics and management but represent average pollutant concentrations that cannot be further discretized. These methods coupled with the difficulty of predicting dynamic storm variations in real time necessary to appropriately sample the storm leads to masked representation of what is happening in real time. The results of this paper introduce a real-time analytical method for stormwater chemistry assessment that bridges many of the contemporary sampling pitfalls. Regressions were within typical acceptable ranges allowed in stormwater chemistry analytical results from certified labs. These results encourage advancing the use of these technologies with greater efficiency and at a lower overall cost than conventionally available methodologies.