Lack of data; only nine major rivers in the United States had oil and grease data in the U.S. Environmental Protection Agency’s STORET data base, and several of these consisted of very few observations.
Differences in measuring and reporting data; most of the available oil and grease data in STORET was gathered using either the Soxhlet extraction method or the liquid-liquid extraction method. The minimum detection limit (denoted off-scale low in the STORET records) and approach for reporting values measured below the detection limit varied with location and time. For example, the minimum detection limit on the Delaware River was 2 mg L−1 for data reported from 1988-1994, and 5 mg L−1 for data reported after 1994. By comparison, the minimum detection limit on the Mississippi River was 1 mg L−1 for the entire period of record (1973-1996).
Adjustment of off-scale low measurements; these values were set to half their reported value even though the actual value was unknown.
Estimating the proportion of petroleum-related hydrocarbons and PAH in oil and grease measurements
Quantifying the uncertainty in the estimates presented in this analysis was not possible, but a reasonable estimate of the low and high ranges of the calculated oil and grease values was made by assuming that the data available from the 1990s for the Mississippi and Delaware rivers, respectively, represented the low and high bounds of oil and grease unit loading for the rivers for which STORET data were unavailable in the 1990s, and for coastal zones in North America and the world (Table I-12). Based on these assumptions, the range of worldwide loadings of land-based sources of oil and grease to the sea was 4.5 million−33.3 million tonne yr−1, with a best estimate of 9.4 million tonne yr−1. The values shown in Table I-12 also reflect low, best, and high estimates of oil and grease loadings from Gulf coast refineries. Calculations of oil and grease discharges using daily maximum guidelines (6.0 lbs per 1000 barrels of crude produced) were used as a high estimate of these loadings, while calculations using the monthly average guidelines (3.2 lbs per 1000 barrels of crude produced) were used as a low estimate. The average of the two calculations was used as a best estimate of the loadings.
Estimates of total petroleum hydrocarbons in the Mississippi River were based on the December 2000 average PAH data of Michel (2001), the assumption that PAH constitute 0.1%−10% of total petroleum hydrocarbons, and the 1990s’ measured average oil and grease concentration of 0.84 mg L−1. Thus, using the lower bound of PAH fraction in total hydrocarbons, a lower bound for estimated hydrocarbons in oil and grease was 0.15%, while an upper bound of hydrocarbons as 15% of oil and grease was determined assuming PAH constitute 10% of total petroleum hydrocarbons. The final range of estimates of total hydrocarbons were therefore made by assuming that the low estimate corresponded with the low percentage of total hydrocarbons (i.e., 0.15%) in the low estimate of oil and grease loading, the best estimate corresponded with 1.5% of total hydrocarbons in the best estimate of oil and grease loading, and the high estimate corresponded with the high percentage of total hydrocarbons (i.e., 15%) in the high estimate of oil and grease loading (Table I-13). Thus, the range of land-based petroleum hydrocarbon loading to the sea was 6,800−5,000,000 tonne yr−1, with a best estimate of 141,000 tonne yr−1.
The application of the PAH data of Michel (2001) on the Mississippi River involved uncertainties regarding the degree to which that data were representative of distributions of PAH in land-based discharges to the sea via rivers and coastal discharges. Part of this uncertainty arises from the lack of consistent PAH measurements in the water column. A review of STORET and the USGS’ National Water Information Service (NWIS) data revealed less than a dozen measurements of PAH above detection limits on rivers in the United States. Furthermore, reported water column PAH concentrations in the literature were not consistent with respect to the constituents reported, did not use the same measurement methods, and/or did not include particulate and dissolved concentrations of PAH. Nonetheless, literature-reported data and data provided by Baker (2001) on the Susquehanna River indicated that the Michel (2001) data were within a reasonable range for river total PAH concentrations. Thus, the range of the background measurements of total PAH on the Mississippi River by Michel (2001) (i.e., 100 to 156 ng L−1, with an average of 128.3 ng L−1) were compared with the average oil and grease concentration for the Mississippi River of 0.84 mg L−1 to determine the estimated range of PAH in oil and grease as 0.012% to 0.019%, with a best estimate of 0.015%. The low estimate of PAH loading to the sea from land-based sources was therefore estimated as 0.012% of the low estimate of oil and grease loading, and the high PAH loading estimate was calculated as 0.019% of the high estimate of oil and grease loading. The best estimate of PAH loading from land-based sources was calculated using 0.015% of the best estimate of oil and grease loading (Table I-13). The range of PAH loading to the sea from land-based sources was 500−6,300 tonne yr−1, with a best estimate of 1,400 tonne yr−1.
The average oil and grease loading of 2.68 g m−2 yr−1 estimated in this study (see Table I-8) was comparable to oil and grease loadings estimated for urban areas in other studies (Table I-14). The range of estimates presented in the current analysis (1.13−12.22 g m−2 yr−1) encompassed the estimates of the previous studies. Perry and McIntyre’s (1986) estimate was actually an event-based calculation that should be higher than an annual load. In addition, the estimates by