ABSTRACT

The U.S. Geological Survey (USGS), in cooperation with U.S. Department of Transportation Federal Highway Administration and the Massachusetts Department of Transportation, conducted a field study from September 2005 through September 2007 to characterize the quality of highway runoff for a wide range of constituents. The highways studied had annual average daily traffic (AADT) volumes from about 3,000 to more than 190,000 vehicles per day. Highway-monitoring stations were installed at 12 locations in Massachusetts on 8 highways. The 12 monitoring stations were subdivided into 4 primary, 4 secondary, and 4 test stations. Each site contained a 100-percent impervious drainage area that included two or more catch basins sharing a common outflow pipe. Paired primary and secondary stations were located within a few miles of each other on a limited-access section of the same highway. Most of the data were collected at the primary and secondary stations, which were located on four principal highways (Route 119, Route 2, Interstate 495, and Interstate 95). The secondary stations were operated simultaneously with the primary stations for at least a year. Data from the four test stations (Route 8, Interstate 195, Interstate 190, and Interstate 93) were used to determine the transferability of the data collected from the principal highways to other highways characterized by different construction techniques, land use, and geography.

Automatic-monitoring techniques were used to collect composite samples of highway runoff and make continuous measurements of several physical characteristics. Flowweighted samples of highway runoff were collected automatically during approximately 140 rain and mixed rain, sleet, and snowstorms. These samples were analyzed for physical characteristics and concentrations of 6 dissolved major ions, total nutrients, 8 total-recoverable metals, suspended sediment, and 85 semivolatile organic compounds (SVOCs), which include priority polyaromatic hydrocarbons (PAHs), phthalate esters, and other anthropogenic or naturally occurring organic compounds. The distribution of particle size of suspended sediment also was determined for composite samples of highway runoff. Samples of highway runoff were collected year round and under various dry antecedent conditions throughout the 2-year sampling period. In addition to samples of highway runoff, supplemental samples also were collected of sediment in highway runoff, background soils, berm materials, maintenance sands, deicing compounds, and vegetation matter. These additional samples were collected near or on the highways to support data analysis.

There were few statistically significant differences between populations of constituent concentrations in samples from the primary and secondary stations on the same principal highways (Mann-Whitney test, 95-percent confidence level). Similarly, there were few statistically significant differences between populations of constituent concentrations for the four principal highways (data from the paired primary and secondary stations for each principal highway) and populations for test stations with similar AADT volumes. Exceptions to this include several total-recoverable metals for stations on Route 2 and Interstate 195 (highways with moderate AADT volumes), and for stations on Interstate 95 and Interstate 93 (highways with high AADT volumes). Supplemental data collected during this study indicate that many of these differences may be explained by the quantity, as well as the quality, of the sediment in samples of highway runoff.

Nonparametric statistical methods also were used to test for differences between populations of sample constituent concentrations among the four principal highways that differed mainly in traffic volume. These results indicate that there were few statistically significant differences (Mann-Whitney test, 95-percent confidence level) for populations of concentrations of most total-recoverable metals and organic compounds among the stations on the four principal highways. There were significant differences between most populations of concentrations in samples from Route 119 (the highway with the lowest AADT volume) and those from the three principal highways with higher AADT volumes. Nevertheless, the median concentrations and upper interquartile ranges for populations of many total-recoverable metals collected from the 12 stations increased with increasing AADT volume, indicating a positive correlation between the two variables. The frequency of detection for various organic compounds also increased with traffic volume for the four principal highways. Furthermore, results for Kendall’s Tau correlation coefficient and Spearman rank correlation coefficient tests indicate that correlations between the median concentrations for total nitrogen (N), total phosphorus (P), all total-recoverable metals, benzo[a]anthracene,benzo[b]fluoranthene, luoranthene, and pyrene to the respective AADT volumes of the eight highways were significantly different at a 95-percent confidence level.

Concentrations of nearly all constituents measured in samples of highway runoff in this study increased substantially during January 2006 through April 2006 and March 2007 through April 2007 compared to the concentrations in samples collected during months without snowfall. Average concentrations of total P, total-recoverable metals, and suspended sediment in samples of highway runoff collected during the winter were about 3 to 11 times the average concentrations observed in samples of non-winter runoff. Results for single-tail Mann-Whitney tests of nearly all populations of constituents between winter composite samples and non-winter composite samples collected from each principal highway indicate that the winter concentrations were significantly higher at a 99-percent confidence interval. Deicing compounds, which contain trace amounts of total N and many total-recoverable metals, account for only a small fraction of the difference between winter and non-winter concentrations. Furthermore, results of analyses of 1,381 runoff events indicate that estimated event-mean concentrations (EMCs) of chloride (Cl) in excess of 10,000 mg/L, which would contain relatively marginal amounts of N and many total-recoverable metals, occurs in less than about 2 percent of the runoff events. Many of the same constituents are associated with sand applied to the highways during the winter. Data from this study indicate that highwaymaintenance sand can account for a substantial amount of the difference between winter and non-winter concentrations of P, iron (Fe), and manganese (Mn). For the monitoring stations on Interstate 495, highway maintenance sand was estimated to account for about 94, 38, and 53 percent of the increase in winter EMCs for total P, Fe, and Mn, respectively.

Results of principal component analysis of all data and data subsets for total-recoverable metals and for PAHs from the principal highways were inconclusive. However, the proportions of most highway constituents originating from likely sources were estimated from the ratios of certain PAH compounds in samples of highway runoff to known ratios of the same PAHs in different matrixes and from ratios of the dominant particulate-associated PAHs to selected elements in exhaust emissions. These analyses indicate that the dominant sources of most PAHs are consistent with automobile emissions. The source for P and most total-recoverable metals other than copper (Cu) and zinc (Zn) are largely local soils, berm materials, vegetation matter, and maintenance sands, and sources of Cu, Zn, and many phthalate compounds are likely tire and brake wear. Conservative contributions of P, Fe, Mn, nickel (Ni), and lead (Pb) from soils proximate to the highways were estimated to account for on average about 37 percent of the median P concentration, 53 percent of the median Fe concentration, 54 percent of the median Mn concentration, 15 percent of the median Ni concentration, and 44 percent of the median Pb concentration in samples of highway runoff from the principal highways. Many of the SVOCs that were detected in samples of highway runoff in this study are associated with fuels, lubricants, antifreeze, windshield fluids, and chemical treatments of engine compartments, finish-paint coats, automotive panels, and other vehicle components. However, other SVOCs in the same samples indicate plant materials, tobacco products, wood preservatives, fecal bacteria, and potential leachates from highway litter, including compounds associated with common beverages and personalcare products.

Concentrations of suspended sediment in composite samples of highway runoff were examined as a potential surrogate for concentrations of sediment-affiliated constituents measured in samples of highway runoff. Relations between suspended sediment and total P, and suspended sediment and many total-recoverable metals indicate that it is possible to estimate reasonable planning-level concentrations for many constituents on the basis of the concentration of a constituent affiliated with suspended sediment and from the average gradated concentrations (concentrations that include suspended sediment of specific size ranges) of suspended sediment from the principal highways in this study. For many applications, the average gradated concentrations of suspended sediment from this study may be used in the absence of site-specific data because results of statistical tests of concentrations of suspended sediment among the four principal highways, and between the principal highways and most test highways, were not significantly different, except for Route 119, which had the lowest AADT volume. Relations between gradated concentrations of suspended sediment and concentrations of total P and many totalrecoverable metals also indicate that a disproportional amount of the concentration for most constituents is associated with fine-grained sediments less than 0.063 millimeters in diameter.

First posted November 2010