First Flush: Fact or Fiction?

For the last several years, the term “first flush” has become more or less a given in stormwater quality issues, despite the fact that studies vary on its prevalence or regularity. In conjunction with a required monitoring program, Salt Lake County, UT, has spent the last several years collecting data for a pollutograph to find out if a first flush does indeed exist in the region and, if so, how that might impact considerations for potential stormwater treatment systems.

A pollutograph provides stormwater quality data throughout a storm, allowing for an evaluation of pollutant concentrations, including the presence or absence of a first flush. If the data in Salt Lake County’s pollutographs showed a strong first-flush phenomenon in the region, the county could use this information, for example, to concentrate treatment on the first part of a storm rather than throughout the entire storm. However, the data collected to date do not indicate the presence of a strong first flush in Salt Lake County, which will certainly impact the county’s future decisions regarding stormwater treatment options.

Salt Lake County is located in an arid area, receiving an average of 16.5 inches of rain per year. The county is surrounded by mountains reaching up to 11,000 feet in elevation, which can get up to 600 inches of snow annually (Figure 1). The countywide watershed is approximately 800 square miles, with all surface waters and its 20 major streams draining to the Great Salt Lake. The county is also a major metropolitan area, home to approximately one million people, with an engineering division responsible for flood control and approximately 64 flood control facilities (Figure 2).

In the 1980s, Salt Lake County had the privilege of being one of 28 municipalities asked to participate in the USEPA Nationwide Urban Runoff Program, commonly referred to as “NURP” and the precursor to the EPA’s stormwater program. In 1992, the county was issued a Phase I municipal separate storm sewer system (MS4) stormwater discharge permit, which requires regular monitoring of representative storm events. As part of a study for that permit, the county’s average storm characteristics showed that a representative storm event has an average rainfall volume of 0.62 inches over an average 6.4-hour period, accumulating at an average 0.1 inch per hour. The minimum allowable rainfall for monitoring a representative storm event was determined from average storm data and observed storm patterns. The specific storm criteria used for the county’s sampling program requires a minimum of 0.20 inch of precipitation that produces runoff. The sampled storm event must be preceded by at least 72 hours of dry weather.

Sampling Program
Continuing the sampling efforts after the NURP study and in accordance with MS4 permit requirements, Salt Lake County conducts outfall sampling twice a year, during the spring and fall. Six stations representing different land uses are sampled. Land uses throughout the county are presented in Figure 3; land uses represented by sample stations are presented in Figure 4 and include commercial, industrial, mixed, transportation, and residential (two stations).

Grab samples are taken from each station approximately 30 minutes before a storm is forecasted to occur and during the rising limb of the runoff event. In addition, flow-weighted composite samples are collected throughout the storm. Samples are analyzed for approximately 22 constituents, including solids, nutrients, oxygen, metals, bacteria, and oil and grease. Event mean concentrations (EMCs) are calculated based on pollutant loads, precipitation, runoff coefficients, and serviced basin area.

Beginning in 2005, Salt Lake County initiated what is referred to as “pollutograph sampling,” which it has since conducted during the fall of 2006 and the spring and fall of 2008. Pollutograph sampling consists of collecting multiple discrete samples throughout a representative storm event in order to generate time versus concentration curves, thus providing an evaluation of pollutant concentration throughout the storm, and the presence of first-flush concentrations. Samples are collected at 10- to 15-minute intervals, for four to six hours, depending on the duration of the storm event.

Using these pollutographs, the first-flush phenomenon was evaluated by a dimensionless plot of the normalized cumulative pollutant mass versus the normalized cumulative runoff volume. The graphs for the 2008 data are presented in Figure 5; the complete set of graphs is available in Salt Lake County’s Stormwater Quality Data Technical Report (Salt Lake County 2009). A 45-degree line (1:1) plotted on each load graph indicates constant pollutant concentration throughout the storm event. A first-flush phenomenon is indicated when the storm line is above the 45-degree line at the earlier stages of the storm event. Based on these results, a slight first flush of total suspended solids (TSS) and total zinc was identified at station LIT-06. The other stations and parameters did not exhibit a first flush.

To quantify the strength of the first-flush phenomenon, the mass first-flush ratio was calculated for each load graph for all pollutographs (Figure 6). The mass first-flush ratio is the ratio of normalized cumulative pollutant mass to normalized cumulative runoff volume at selected fractions of runoff volume. For this analysis, the mass first-flush ratio was calculated at 10% and 30% runoff volume (0.1 and 0.3 normalized cumulative runoff volume). The methodology utilized herein is similar to that presented by Kayhanian and Stenstrom (2008). A higher ratio represents a greater first-flush phenomenon; ratios above 1.0 represent the presence of a first flush.

As shown in Figure 5, the presence of a first flush is variable with storms and stations. However, overall, the first-flush phenomenon in the storm events sampled was either not present or weak, with the greatest first flush occurring at station LIT-06 for the 2006 storm event.

The absence of a strong first flush in Salt Lake County follows suit with results from studies in other areas of the country and abroad. A study conducted at North Carolina State University (Tucker 2007), for example, found what was termed a “high inconsistency” in the occurrence of first flush, which the author also noted is consistent with other research. Another study, conducted by A. Taebi and R. L. Droste (2004), indicated a relatively weak first flush for some parameters, no correlation for some, and an increase in the first-flush load of TSS when the intensity and duration of a storm event increases. However, a study conducted by the California Department of Transportation (Caltrans 2005) identified several types of first flushes for highway sites, all indicating that there indeed was a discharge of greater concentrations or mass in the early part of a storm, with the exception of a seasonal first flush. (First-flush types analyzed were polycyclic aromatic hydrocarbon (PAH), litter, particle, and seasonal.)

In a rainfall simulation study conducted along roads in Brisbane, Australia (Barry et al. 2004), storm events were simulated to provide an assessment of pollutant loads under representative and quantifiable rainfall intensity. While this study indicated the presence of a first flush, it is noted that the experimental basins and storm events were small. The Australia Department of Environment and Conservation has since stated that first flush is “most readily observed on small catchments or individual premises, particularly if a high proportion of the catchment is impervious (such as paved surfaces and roads)” (2008). This may help to explain the lack of a first flush in the Salt Lake County basins, which ranged from 67 to 800 acres in size.

While the findings of the Brisbane study do not totally confirm or deny the existence of first flush, the approach is still interesting. In the study, researchers attempted to simulate a storm event through the use of a series of pipes and risers. Water was pumped from a tank at a predetermined rate to risers, with nozzles specially designed to spray water in the form of raindrops. The Australia Department of Environment and Conservation has posted more information about this study on its Web site,

The Verdict
The incongruity of these studies points to some of the problems associated with stormwater sampling. For one, variables are inherent in this type of sampling, including the amount of precipitation and intensity of the storm events, basin size, antecedent dry periods, and possibly specific parameters (e.g., oil and grease do not become entrained in stormwater as quickly as other pollutants). Given the inherent variables in stormwater sampling and the small data set, it is difficult to identify significant trends. The collection of more data in the future will serve to augment the existing data and identify stormwater pollutant trends.

In Salt Lake County, it does not appear that a strong first flush exists, which, for now, means ruling out an exclusively first-flush-treatment strategy. However, the county does plan to continue collecting data and researching new first-flush experiments. At the very least, this research will lead to a better understanding of stormwater quality in the area, including the behavior of stormwater pollutants during a storm event, and, at best, may help the county make more informed and more effective decisions on how to manage stormwater runoff.


The authors would like to give special thanks to Nicholas von Stackelberg of Stantec in Salt Lake City for his work on the statistical analyses for this study. 

About the Author

Julie Howe and Steve Burgon

Julie Howe is an environmental scientist in the Salt Lake City, UT, office of Stantec. Steve Burgon is a stormwater quality manager for the Salt Lake County Engineering Division.

Photo 39297166 © Mike2focus |
Photo 140820417 © Susanne Fritzsche |
Microplastics that were fragmented from larger plastics are called secondary microplastics; they are known as primary microplastics if they originate from small size produced industrial beads, care products or textile fibers.
Photo 43114609 © Joshua Gagnon |
Dreamstime Xxl 43114609