Abstract
In stormwater monitoring, a composite sample is often obtained from a series of aliquots collected during a storm event, to create a single sample that is representative of the entire storm event runoff. The constituent concentrations in such composite samples are referred to as event mean concentrations (EMCs). The aliquots comprising the composite sample are usually collected using a flow-weighted sampling scheme, in which each aliquot is taken after a predetermined volume of flow has passed. Stormwater flow is not easy to measure accurately and consistently, and it has been recognized that errors in flow measurement when collecting aliquots will affect the resulting EMCs.
In this study, the effect of flow measurement error on the flow-weighted EMC of Total Suspended Solids (TSS) was investigated, using hydrograph and pollutograph data from a first flush characterization study conducted by others at California highway drainage facilities in Los Angeles, California. Systematic errors in measuring flow depth in three different kinds of flumes (Parshall, Palmer-Bowlus, and H-type) were applied to flow calculations for 18 storms at three sites. These errors distorted the measured hydrographs. Using a constant aliquot pacing volume for each storm event, the altered hydrographs were used to determine when aliquots would have been collected in the field. The TSS concentrations in these aliquots were obtained from the associated pollutographs. The aliquot concentrations were then averaged to obtain an EMC for each storm event, as if an actual flow-weighted sampling scheme had occurred. Systematic stage errors ranging from -25% to +25% of the maximum depths of the flumes were used to modify flow data.
Negative stage measurement errors caused fewer aliquots to be collected, often below the minimum number needed to properly monitor an event. Stage measurement errors from -10% to +25% of the maximum depth of the flume produced data sets with at least 17 simulated storms in which six or more aliquots were recorded. For these data sets, mean EMC errors varied from -5% to +2% and the median values varied from -2% to +2%. As stage measurement error departed from zero in both the positive and negative directions, the range or variance in EMC error data sets increased, as did the probability that the EMC error for an individual storm will be above or below a specified value. The results of this study will allow stormwater practitioners to better determine the magnitude of flow measurement error that can be tolerated for a desired level of accuracy in resulting EMCs.