By Gregory C. Delzer, John S. Zogorski, Thomas J. Lopes, and Robin L. Bosshart
The U.S. Geological Survey (USGS) sampled storm water in 16 cities and metropolitan areas that are required to obtain permits to discharge stormwater from their municipal storm-sewer system into surface water. Concentrations of 62 VOCs, including MTBE and BTEX compounds, were measured in 592 stormwater samples collected in these cities and metropolitan areas from 1991 through 1995. Con centration data for MTBE and BTEX compounds in storm water were compiled and analyzed, and the findings are summarized in this report. This effort was part of an inter agency assessment of the scientific basis and effectiveness of the Nation's oxygenated fuel program and was coordi nated by the Office of Science and Technology Policy, Exec utive Office of the President.
MTBE was the seventh most frequently detected VOC in urban stormwater, following toluene, total xylene, chloro form, total trimethylbenzene, tetrachloroethene, and naph thalene. MTBE was detected in 6.9 percent (41 of 592) of stormwater samples collected. When detected, concentra tions of MTBE ranged from 0.2 to 8.7 micrograms per liter (µg/L), with a median of 1.5 µg/L. All detections of MTBE were less than the lower limit of the U.S. Environmental Protection Agency (EPA) draft lifetime health advisory (20 µg/L) for drinking water. Eighty-three percent of all detections of MTBE in stormwater were in samples col lected during the October through March season of each year (1991-95), which corresponds with the expected sea sonal use of oxygenated gasoline in areas where carbon monoxide exceeds established air-quality standards. The median concentration of MTBE and benzene for all samples was statistically different and higher in samples collected during the October through March season than samples col lected during the April through September season. Sixty-six percent of all MTBE detections occurred with BTEX com pounds, and a proportionate increase in concentrations was found when these compounds occurred together. The pro portionate increase could indicate a common source of MTBE and BTEX for those samples. Toluene and total xylene were the most frequently detected BTEX compounds and the most frequently detected VOCs in these investiga tions. Detected concentrations of toluene and total xylene ranged from 0.2 to 6.6 µg/L and 0.2 to 15 µg/L with median concentrations of 0.3 and 0.4 µg/L, respectively.
Oxygenated gasoline and reformulated gasoline are two classes of gasoline that contain different amounts of oxy genated compounds. Oxygenated gasoline contains no less than 2.7 percent oxygen by weight, or 14.8 percent MTBE by volume. Reformulated gasoline contains no less than 2.0 percent oxygen by weight, or 11 percent MTBE by vol ume. Officials have voluntarily chosen to use oxygenated gasoline to improve air quality in many areas of the United States that are in compliance with air-quality standards. MTBE and other oxygenates also have been used to enhance octane levels of gasoline since the late 1970's; however, the amount of oxygenate used is less than the amount stipulated by the Clean Air Act.
Of 60 VOCs measured in 1993 and 1994 as part of the USGS's National Water-Quality Assessment Program, MTBE was the second most frequently detected VOC in ground water sampled from 210 shallow wells and springs in 8 urban areas (Squillace and others, 1996). MTBE con centrations exceeded 20 µg/L in 3 percent of the samples. The EPA has issued a draft lifetime health advisory for MTBE in drinking water that ranges from 20 to 200 µg/L (U.S. Environmental Protection Agency, 1996). The EPA is continuing to assess MTBE health data. Human-health complaints related to MTBE air exposure by some individu als have been reported in Fairbanks and Anchorage, Alaska; Missoula, Montana; Milwaukee, Wisconsin; and New Jer sey. Symptoms of MTBE exposure included headaches, dizziness, irritated eyes, coughing, disorientation, and nau sea.
Possible nonpoint sources of MTBE in shallow ground water include infiltration of stormwater runoff and precipi tation. Possible point sources include spills on land surfaces that may enter surface water and releases from aboveground and underground gasoline storage tanks. MTBE and other oxygenates released to the environment are expected to move with water in the hydrologic cycle (fig. 1), primarily because they are highly soluble in water. Precipitation can contribute MTBE to surface water by falling directly on a body of water, by overland runoff, and from stormwater dis charges. MTBE is less biodegradable than common gaso line hydrocarbons such as BTEX compounds, and although volatile, MTBE is not expected to rapidly volatilize from deep, slow-moving rivers and some streams (Zogorski and others, 1996). More data are needed to assess the extent of MTBE's movement in the hydrologic cycle.
Figure 1. (25K)
Because of human-health, engine-performance, and water-quality concerns, the Office of Science and Technol ogy Policy, Executive Office of the President, coordinated an interagency assessment of the scientific basis and effec tiveness of the Nation's winter oxygenated gasoline pro gram. Findings from this assessment (Zogorski and others, 1996) indicate that MTBE occurs in urban air, disperses rapidly in water, and could occur in precipitation in propor tion to its concentration in the atmosphere. For a given atmospheric concentration, cooler temperatures result in higher concentrations of MTBE in precipitation than do warmer temperatures.
As part of the interagency assessment, data on VOCs in stormwater that were collected by the USGS were compiled and analyzed. This report presents a summary of the 10 most frequently detected VOCs and an analysis of the occurrence of MTBE and BTEX compounds.
The USGS stormwater investigations were not specifi cally designed to characterize gasoline oxygenates in storm water. However, the data that were collected do provide some insights on the occurrence, or lack of occurrence, of MTBE and BTEX compounds in stormwater in the cities sampled. The design of each investigation differed depend ing on the requirements of each EPA Region and the extent of monitoring desired by each municipality. Samples were collected from a variety of stormwater conveyances, such as culverts, concrete pipes, lined ditches, and channels. The drainage area of sampled sites typically was small and ranged from 0.005 to 10.7 square miles. Each site had a predominant land use including residential, commercial,
Figure 2. (13K)
industrial, or highway. Samples generally were collected by hand-dipping VOC sample bottles in stormwater, usually within the first 2 hours of storm runoff. One sample was collected during each storm event with the exception of Colorado Springs, Colorado, where two to four samples were collected (von Guerard and Weiss, 1995). For this report, each sample from Colorado Springs was included in the data analyses. All samples were analyzed at the USGS National Water-Quality Laboratory by gas chromatogra phy/mass spectrometry (Rose and Schroeder, 1995).
MTBE detection frequencies were calculated for the entire study period and for the October through March and April through September seasons for each stormwater project. The October through March season corresponds with the expected winter use of oxygenated gasoline in areas where carbon monoxide exceeds established air- quality standards. Oxygenated gasoline is not expected to be used during the April through September season in areas where carbon monoxide exceeds air-quality standards. Three-hundred five (305) of the samples were collected dur ing the April through September season, and 287 samples were collected during the October through March season. MTBE was detected in 41 of the 592 stormwater samples, or 6.9 percent (fig. 3). Concentrations of MTBE in storm water for all samples, including those less than the mini mum reporting level (MRL), ranged from less than the MRL to 8.7 µg/L, with a median concentration less than the MRL. A detectable concentration is one that is greater than or equal to the MRL. All detections of MTBE were less than the lower limit of the EPA draft lifetime health advi sory (20 µg/L) for drinking water. For those samples in which MTBE was detected, the concentrations ranged from 0.2 to 8.7 µg/L, with a median of 1.5 µg/L. Eighty-three percent, or 34 of 41 detections, occurred during the October through March season when MTBE would likely be used in greater amounts in areas where carbon monoxide concentra tions exceeded air-quality standards.
Table 1. Statistical summary of the 10 most frequently detected volatile organic compounds in urban stormwater in the United States, 1991-95
[µg/L, micrograms perliter] ------------------------------------------------------------------------- VOC Minimum Maximum Median Frequency detected detected detected of concentration concentration concentration detection (µg/L) (µg/L) (µg/L) (percent) ------------------------------------------------------------------------- Toluene 0.2 6.6 0.3 23.2 Total xylene 0.2 15 0.4 17.5 Chloroform 0.2 7.0 0.7 13.4 Total trimethyl- 0.2 15 0.3 12.4 benzene Tetrachloro- 0.2 42 0.6 8.0 ethene Naphthalene 0.2 5.1 0.3 7.4 MTBE 0.2 8.7 1.5 6.9 Dichloro- 0.2 13 0.3 5.9 methane Bromo- 0.2 2.8 0.6 5.8 dichloro- methane Ethylbenzene 0.2 2.0 0.3 5.0 -------------------------------------------------------------------------
Figure 3. (13K)
The MRL for the analytical technique used for MTBE was reduced from 1.0 to 0.2 µg/L in April 1994. However, most of the stormwater samples collected in the USGS investigations described here were analyzed prior to this change. The high MRL of 1.0 µg/L for MTBE for most of the samples censors data with low concentrations. A lower MRL for MTBE would likely result in increased frequency of detection and a lower median detectable concentration. The MRL was different for some stormwater samples because they had to be diluted prior to analysis, which resulted in a higher reporting level. Eighty-three percent of the 592 samples were not diluted and were analyzed with a MRL of 1.0 µg/L. Eleven percent were not diluted and were analyzed with a MRL of 0.2 µg/L. Six percent of the samples were diluted and were analyzed with a MRL that ranged from 0.4 to 100 µg/L.
The median concentration of MTBE for all samples was shown to be statistically different and higher during the October through March season in comparison to the April through September season. The probability of the median concentration of MTBE being the same during these two seasons is less than 1 in 10,000. The strong seasonal detec tion pattern may be attributed to the increased volume and use of MTBE in gasoline during the oxygenated gasoline season, longer atmospheric half-lives of MTBE during the winter, and (or) to the higher solubility and decreased vola tility (the tendency of a liquid or solid to change into the gaseous state) of MTBE in stormwater at winter tempera tures.
For each of the three cities (Phoenix, Arizona, Colorado Springs, Colorado, and Denver, Colorado) with known MTBE use, MTBE was detected only in stormwater sam ples collected during the season when oxygenated gasoline was in use (fig. 3). MTBE was detected in 16 of 40 (40 percent) samples and concentrations ranged from 1.0 to 4.2 µg/L, with a median concentration of 1.5 µg/L.
Detection of MTBE in cities confirmed not to use oxy genated fuel may be attributable to the use of MTBE in gas oline as an octane enhancer. These cities include Atlanta, Georgia, Baton Rouge, Louisiana, Birmingham, Alabama, and Dallas/Fort Worth and San Antonio, Texas.
Seventy-nine percent (467) of the 592 stormwater sam ples were collected from the eight cities that had at least one detectable concentration of MTBE. Thirty-two percent (148) of the 467 samples had detectable concentrations of either MTBE or BTEX (fig. 4). Among the 148 stormwater samples, 18 percent contained both MTBE and BTEX com pounds, 72 percent contained only BTEX compounds, and 10 percent contained only MTBE. Sixty-six percent of the 41 MTBE detections were in samples that also had detect able concentrations of BTEX compounds. A proportionate increase in concentration was found when these compounds were detected together, which could indicate a common source of MTBE and BTEX for these samples. Toluene and total xylene were the most frequently detected BTEX com pounds and were also the most frequently detected VOCs in these investigations. When detected, concentrations of tolu ene and total xylene ranged from 0.2 to 6.6 µg/L and 0.2 to 15 µg/L with median concentrations of 0.3 and 0.4 µg/L, respectively. BTEX compounds were analyzed with a MRL of 0.2 µg/L throughout the study.
Figure 4. (8K)
Detections of toluene, total xylene, and ethylbenzene did not follow a seasonal detection pattern. The median con centrations for these compounds were shown to be statisti cally similar during the October through March season in comparison to the April through September season. The detection of benzene, however, followed a seasonal detec tion pattern similar to that for MTBE. The probability of the median concentrations of benzene being the same dur ing these two seasons is about 1 in 3,300. The strong seasonal detection pattern for benzene may be attributed, in part, to the higher solubility and decreased volatility of ben zene during colder winter temperatures.
MTBE was detected in 6.9 percent (41) of 592 storm water samples collected in 16 cities and metropolitan areas from 1991 through 1995. When detected, concentrations of MTBE ranged from 0.2 to 8.7 µg/L, with a median of 1.5 µg/L. All concentrations of MTBE were less than the lower limit of the EPA draft lifetime advisory (20 µg/L) for drinking water. Toluene and total xylene were the most fre quently detected BTEX compounds and, of 62 VOCs ana lyzed, the most frequently detected VOCs. When detected, concentrations of toluene and total xylene ranged from 0.2 to 6.6 µg/L and 0.2 to 15 µg/L with median concentrations of 0.3 and 0.4 µg/L, respectively. Eighty-three percent of all detected concentrations of MTBE occurred during the October through March season, which corresponds with the expected seasonal use of oxygenated gasoline in areas where carbon monoxide exceeds established air-quality standards. Forty percent of stormwater samples that were collected during the October through March season in areas with confirmed use of oxygenated gasoline contained detectable concentrations of MTBE. The median concen tration of MTBE and benzene for all samples was statisti cally different and higher for samples collected during the October through March season than for those collected dur ing the April through September season. Sixty-six percent of all MTBE detections occurred with BTEX compounds, and the proportionate increase in concentrations when these compounds occurred together could indicate a common source.
These data raise questions that remain to be answered because these stormwater investigations were not designed specifically to characterize the occurrence, sources, and behavior of oxygenated gasoline components in stormwater. These include:
(1) What are the ranges and seasonal distributions of concentrations of MTBE in stormwater, including munici pal separate-storm-sewer systems and combined sewer overflows, in other urban areas of the United States?
(2) What is the persistence of MTBE in streams or rivers that receive stormwater runoff? Are the concentrations in the receiving stream a cause for concern about potential effects on aquatic life? Similarly, what effect, if any, does MTBE have on public water supplies from surface-water sources?
(3) What proportion of MTBE that is detected in urban stormwater is contributed by precipitation compared to that contributed by overland runoff? How much MTBE is con tributed to surface water by precipitation that falls directly on larger bodies of water such as reservoirs and lakes?
(4) Do other oxygenates react similarly to MTBE in the hydrologic cycle and occur in stormwater?
(5) Is land use an important factor in the occurrence of MTBE or BTEX compounds in urban stormwater?
(6) Is stormwater recharge and (or) precipitation that contains VOCs an important source of MTBE to ground water in urban environments?
The U.S. Geological Survey continues to work on these questions in cooperation with city, state, and other Federal agencies.