ABSTRACT

In 2008, the North Carolina General Assembly passed House Bill 2436 that required the North Carolina Department of Transportation (NCDOT) to study the water-quality effects of bridges on receiving streams. In response, the NCDOT and the U.S. Geological Survey (USGS) collaborated on a study to provide information necessary to address the requirements of the Bill. To better understand the effects of stormwater runoff from bridges on receiving streams, the following tasks were performed: (1) characterize stormwater runoff quality and quantity from a representative selection of bridges in North Carolina; (2) measure stream water quality upstream from selected bridges to compare bridge deck stormwater concentrations and loads to stream constituent concentrations and loads; and (3) determine if the chemistry of bed sediments upstream and downstream from selected bridges differs substantially based on presence or absence of a best management practice for bridge runoff.

The USGS measured bridge deck runoff from 15 bridges, stream water-quality data at 4 bridge deck runoff sites, and streambed sediment chemistry at 30 bridges across North Carolina. The bridges selected for study had differing sizes, differing ecoregions and land-use characteristics, and a range of annual average daily traffic (AADT). Runoff from both concrete and asphalt deck bridges was sampled. Composite samples of bridge deck runoff were collected for 12 to 15 storms at each bridge. Additionally, routine (monthly) samples of base-flow streamwater and at least seven samples of streamwater during storms were collected over a 12-month period at four sites. Samples were analyzed for a wide range of constituents, including dissolved and total recoverable metals and nutrients, major ions, total suspended solids, suspended-sediment concentration, oil and grease, petroleum hydrocarbons, and semivolatile organic compounds (SVOCs). Parameters of concern (POCs) were defined as analytes with at least one exceedance of a water-quality threshold or were otherwise known to have potentially deleterious effects on receiving streams. The 28 POCs included metals, nutrients, pH, suspended solids concentration, polycyclic aromatic hydrocarbons, and other organic compounds. Results and discussion were limited to these POCs for water samples.

Bridge deck runoff concentrations were generally shown to be statistically higher for bridges located in urban areas than those in rural areas. There was no strong relation between concentrations and AADT, which could be explained by the relatively low traffic volumes at the study sites. When sites with larger ranges of AADT have been studied, especially sites with volumes substantially above and below about 30,000 vehicles, runoff concentrations tended to roughly scale with AADT. The selection process for monitoring sites included an analysis of the AADT frequency distributions in North Carolina; only about 1 percent of bridges in North Carolina have AADT volumes in excess of 30,000 vehicles. Because of the small percentage of bridges in North Carolina with AADT volumes in excess of 30,000 and the extremely limited number of those bridges with runoff collection systems, only two bridge sites with an AADT volume greater than 30,000 (Mallard Creek and Mango Creek) were included in the study. Concentrations of most constituents in bridge deck runoff samples were generally statistically higher in winter compared to all other seasons, pointing to reduced volatilization at lower temperatures and higher total suspended solids concentrations in the winter (likely from deicing treatments) as potential explanations. The runoff samples from the Coastal Plain bridges generally had statistically lower concentrations than samples from the bridges in the Blue Ridge and Piedmont ecoregions.

Results of the statistical testing and comparisons of the bridge deck runoff and stream concentrations indicate that the bridge deck runoff concentrations were only statistically higher than the corresponding stream (routine and storm) concentrations for 36 percent of the comparisons. Thus, with the exception of concentrations of dissolved copper and zinc, total recoverable nickel, and polycyclic aromatic hydrocarbons, which were consistently higher in bridge deck runoff, the bridge deck runoff concentrations at all sites were similar to those measured in the receiving streams at the four stream sampling sites. Comparisons of bridge deck and stream loads indicate that all the bridge deck runoff loads were lower (and generally orders of magnitude lower) than the stream loads for all POCs. The inverse was true for total yields (load per unit drainage area) of the POCs. The bridge deck runoff yields were generally higher than the yields from the four stream sites for all of the POCs. The bridge deck runoff yields can be used to estimate loads at other bridges with similar characteristics and to provide planning-level estimates of the contributing total load from all highways in a watershed. The effect of bridge deck runoff loads on receiving waters should also be evaluated in light of the bioassays, which only showed potential ecological effects for one bridge deck runoff sample (collected in the winter), and benthic macroinvertebrate survey results, which revealed no significant difference upstream and downstream from the study bridge sites.

The rate at which bridge deck runoff mixes with, and is diluted by, the receiving stream was determined by using empirical relations and measured flow conditions at the four gaged stream sites for various steady-state hydraulic conditions. The dilution curves indicated that although in a few cases the maximum concentrations of some constituents in the bridge deck runoff plume exceeded water-quality thresholds by up to 4 times the threshold, levels were reduced to the ambient stream concentration rapidly (generally within 50 feet downstream from the injection point), and in some cases, were actually lower than the stream concentration.

The analysis of the bed sediment quality revealed no obvious patterns in downstream increases in inorganic analytes and total organic carbon at the sampled bridge sites. There was no consistent downstream enrichment of bed sediment with SVOCs, even at the bituminous (asphalt) bridges nor were there any obvious patterns related to urban versus rural bridges or with traffic volume. Possible explanations of these bed sediment results are as follows: (1) bridge decks are not contributing measurable quantities of these analytes to bed sediments; (2) these analytes were efficiently transported downstream, or contaminated bed sediments were scoured from the immediate bridge vicinity during high-flow events; (3) the contributing watershed effects on the bed sediment overwhelm any signature that the relatively small bridge deck area contributes; or most likely (4) a combination of all three of the possible explanations.

Although this study did not show bridge deck runoff to consistently be a primary source of pollutants to receiving streams, there is an indication that under certain conditions (that is, runoff following deicing treatments into stream base-flow conditions) bridge deck runoff can be a significant environmental stressor. The data, analysis, and relations associated with this study can be used by the NCDOT to (1) predict the constituent load from a bridge; (2) provide general information regarding the potential effects a bridge may have on its receiving stream or that all highways may have within a watershed; and (3) provide information needed to select the most efficient best management practice at a bridge construction, replacement, or other highway project site.

First posted October 28, 2011