Effects of Postprocessing Decisions on Flow-Weighted Event Mean Concentrations

Abstract

The sensitivity of an industry-defining stormwater quality metric, the flow-weighted event mean concentration (EMC), to nonstandardized calculation protocols was evaluated. Our objective was to explicate quantitative methods and practical aspects of obtaining EMCs and provide recommendations for minimizing bias where validation is not possible. EMCs may be generated by poststorm compositing of discrete (also known as grab) samples or by collecting flow-weighted composite samples using automated equipment. Three methodological crossroads (flow attribution, volume integration, and flow-sample resolution) were assessed for their relative impact on the flow-weighted EMC. Field monitoring campaigns produced 38 total suspended solids (TSS) pollutographs and hydrographs characterizing untreated and treated runoff that were used to generate EMCs using each available calculation scheme. Untreated runoff was collected from a parking lot and an asphalt street. Treated runoff was collected from a grassed swale system, a proprietary treatment system in series with a constructed wetland, and a permeable pavement. A combinatorial averaging method was utilized to quantify each scheme’s effect; the resultant EMC was compared against an adopted benchmark EMC. EMC outcomes from methodologies that use prior flow attribution underestimate the presumed highest accuracy benchmark EMC by 12.9% on average, regardless of the volume integration method or flow-sample resolution used. No other decision variables demonstrated a meaningful effect size. Composite sampling with autosamplers fits a prior flow-sample attribution scheme, but advanced program options can ameliorate the significant bias associated with prior flow attribution. Outcomes are meaningful for stormwater managers to reduce potential uncertainty when characterizing stormwater control measure (SCM) and collating SCM monitoring data across studies with differing sample collection methods. An open-source web application is offered for any user to conduct flow-weighted EMC calculations using the benchmark method. There are two common methods by which a water quality EMC can be generated from field monitoring data: grab sampling over the hydrograph duration with poststorm compositing, and intermittent programmed autosampling into a flow-weighted composite. The EMC is the weighted average concentration of a given pollutant in the runoff from a given storm event, which may be reported by different data collection or postprocessing methods. This paper investigated and quantified the systematic differences between these two methods based on calculation and procedural assumptions. The analysis suggests that autosamplers using default settings may underestimate the EMC by 13.7% relative to the presumed more accurate grab-sample approach. Fortunately, most autosamplers are equipped with advanced user settings that may reduce the bias compared with the benchmark method.