Applicability of Using Soil Concentrations to Predict Groundwater Concentrations during Assessments at Petroleum Contaminated Sites in Florida

Various empirical equations have been offered to facilitate the prediction of the concentration of pollutants in different phases in the environment. Using one such equation, Brown and Flagg (1981), the predicted separation between the saturated soil and the groundwater of benzene, toluene, ethylbenzene and xylenes were compared to the actual laboratory samples from 18 petroleum contaminated sites across Florida. In this study, 50 underground soil and groundwater samples were used to see if the actual soil data could reliably predict the associated groundwater concentration; if the equation is appropriate to use to predict the groundwater concentration of the sum of the four volatile organic aromatics (benzene, toluene, ethylbenzene and xylenes or BTEX); how five variable involved in the sample collection affected the results: sample type; lithology; horizontal distance between the soil and groundwater sample; vertical distance between the soil and groundwater sample; and magnitude of the soil concentration; and how the calculated foc compared to the default foc used in the development of FDEP's CTLs. The soil concentration had a positive correlation with the groundwater concentration and a positive correlation with the Equation. While at first it appeared that the samples collected via direct push technology had a better correlation than those collected via monitoring wells, after further analysis, the groundwater samples collected via monitoring wells and with a vertical distance between the soil sample and the top of the water table of between 1-2 feet, had the best overall correlation. The poorly graded sands, with little to no fines, (SP) had good correlations for ethyl benzene and xylenes concentrations, but not for benzene and toluene concentrations. Meanwhile, the silty sands (SM) and the clayey sand/sandy clays (SC) were the reverse with good correlations with benzene and toluene, but not for the ethyl benzene and xylenes concentrations. As could be reasonably expected, the horizontal distance of less than or equal to five feet between the soil and groundwater sample had a good correlation (with the same location being the best), and greater than 5 feet away was a poor correlation. The vertical distance of one to two feet had the best correlation with the same location not as good, and greater than two feet worse. The soil concentration magnitude range results were not great for any individual constituents but the correlation for BTEX was pretty good from 0.1 to 100 mg/kg. What was interesting was how the individual constituent's concentrations varied over the soil concentration magnitude ranges. Of the 50 data points, the majority of the benzene and toluene concentrations were in the lower ranges ( The average fractional organic carbon, foc, as calculated from the soil and groundwater concentrations, was the highest in the clayey sands/sandy clays, then silty sands, and finally poorly graded sands. More than 80% of the soil samples had a calculated average foc of greater than 0.006 (0.002 and 0.006 are the default foc used to develop FDEP's Cleanup Target Levels). For soils other than poorly graded sands, the fractional organic carbon should be measured during the site assessments to calculate site-specific CTLs to reduce unnecessary remedial efforts. Trying to use the Brown and Flagg (1981) empirical equation to predict the groundwater concentration from the soil concentration at petroleum contaminated sites in Florida, may lead to erroneous data without analyzing the soil for fractional organic carbon. Better data correlation may be obtained if the corresponding organic content of the soil is analyzed along with the soil contaminant concentration. Further research into this subject could be refined by the following methods: 1) ensure the soil and groundwater data are collected from the same location but at various depths (within 0 to 3 ft); 2) analyze the soil for carbon content instead of assuming a carbon content; 3) obtain a better distribution of lithological and soil concentration magnitude samples; 4) collect groundwater samples at the same location via dpt and MWs; and 5) test aging, lithology and surface covering through a controlled field experiment with multiple cells.