This section describes the oil spill risk analysis conducted for the lease sale and, in particular, addresses the probabilities of accidental oil spills occurring and contacting various resources, as well as the number of spills that could occur if the proposed action is implemented.

A great deal of uncertainty exists about the number and size of spills that might occur as well as the environmental factors (especially winds and currents) that would be prevalent at the time of their occurrence. Some uncertainties are a reflection of incomplete and/or imperfect data; however, uncertainty is simply inherent in the problem of describing future events over which complete control cannot be exercised. In view of this, it is important to consider the range of possible effects that could accompany oil and gas exploration and production on the Mid-Atlantic Outer Continental Shelf.

It is equally important to associate these potential effects with quantitative estimates of the probability of their occurrence and destination. For this, the U.S. Geological Survey developed the Oil Spill Risk Analysis Model (OSRAM). For the model's description and user's guide, the reader is referred to Lanfear et al., 1979, and Lanfear and Samuels, 1981.

The analysis for the proposed lease sale was conducted in two sets, with each set consisting of a model run. The model run for the first set included 30 targets or vulnerable resources plus the target "land." For the second set, 31 new targets were used, with target "land" remaining the same as in the first run. Each set, or run, consisted of three parts: first, the probabilities of oil spill occurrence were developed (see Table IV.B.3-1); second, the trajectories followed by oil spills from hypothetical launch points (assumed spill locations, see Figure IV.B.3-1) to vulnerable targets (see Table IV.B.3-2) and/or land segments (see Figure IV.B.3-2) were also developed; and third, combining these results yielded the probabilities representing the overall risk associated with the proposed action.

Probabilities of Oil Spill Occurrence

For the model run, statistical distributions for estimating probabilities of oil spill occurrence were taken from Devanny and Stewart (1974) and Stewart (1975). Data sources were the USGS accident files and production records and a worldwide survey of tanker accidents. The data included all types of spills related to OCS leasing: well blowouts, accidents on platforms, transportation of the oil to shore and, in some cases, further transportation from an intermediate terminus to refineries. More recently, the Futures group completed a data base of historic oil spills and performed a preliminary analysis of spill rates. The data base showed a sharp drop in oil spill occurrences since 1974. Based on it, new revised rates (Lanfear and Amstutz, 1983) are now being used for projecting possible impacts from OCS leasing.

The basic assumptions when using the distributions were: (1) realistic estimates of future spill frequencies can be based on past OCS experience, (2) spills occur independently of each other, and (3) spill rate is dependent on volume of oil produced and handled.

Two estimates (mean and high) of hydrocarbon resources (oil) for the proposed
lease and for the alternatives were supplied by MMS and used for the model's
runs. Estimates for existing leases in the Atlantic also were supplied by
MMS and used in the runs. It should be emphasized that the oil spill risks
discussed in this analysis are based on conditional resources estimates
which are developed assuming that sufficient commercial accumulations of
oil are present in the lease area. If oil is not present in sufficient
quantity, then no oil spill risk exists. This analysis is designed to try
to determine what the risks are if oil is found.

Table IV.B.3-2.

Vulnerable resources and their periods of potential
vulnerability. Source: OSRAM, results for Sale No. 111.

Table 1 in Appendix C shows the probabilities of one or more spills and the expected number of spills from different sources and activities associated with the proposed action, the existing leases, and with the existing imports of hydrocarbons and OCS recoverable resources.

According to this table, the total expected number of spills greater than 1,000 bbl for the proposed mean case is 0.46, for the proposed high case is 0.99, for the existing leases is 1.72 (assumes all blocks in proposed Sale 90 are leased), for imports of crude oil is 4.68, and for imports of crude and refined products is 26.98. (When the total expected number of spills is less than 1, it is assumed 1 spill will occur.) Of these, depending on the case, 40 to 50 percent are projected to be greater than 10,000 barrels in size. In general, the mean number is greater for the transportation phase than for the production phase (platforms). The probability for at least one spill greater than 1,000 barrels in size is approximately 60 percent if only the existing leases are considered. The probabilities of such a spill associated with the present volume of imports are greater than 90 percent. The low probabilities (23 percent) for the same size spill associated with the proposed action is a reflection of the small volume of hydrocarbons estimated to be recoverable from this source.

Because in actuality the number of spills is a discrete variable, that is, happening in whole numbers and not in fractions, based on Table IV.B.3-1 and for purposes of the analysis found in the EIS, the following assumptions are made regarding the number of spills associated with the proposed action (mean case). Over a 25-year period, one spill greater than 1,000 barrels is assumed to result, either from a platform accident or from transportation by tankers. The present rate of imports of crude and refined products might result in 27 spills over the 25-year period; and, in the combined case, the proposal plus the present volume of imports would result in a total of 27 spills greater than 1,000 barrels in size, which assumes that some imports will be backed out by production from the proposed sale. Trajectories Followed by Oil Spills--the Conditional Probabilities

For the second part of each set, 2,000 spills (500 per season) are assumed to have occurred at each of the locations shown in Figures 2 and 3 in Appendix C and IV.B.3-1.c, and along segments of the transportation routes in Figure 4 of the same Appendix. One-point trajectories then represented the path followed by spills moving under the influence of the surface winds and currents characteristic to each cell of the model's grid. percentage of these spills that contacted targets--vulnerable resources or coastal land-segments (equal length segments and county segments--see Appendix C, Figures 5 and 6--are the conditional probabilities representing the risk of any one target being contacted by an oil spill. The term "conditional" reflects the fact that "a spill is assumed to occur." Thus, these probabilities are "conditioned" by that assumption and do not reflect the probabilities that a spill will occur. They allow identification of the relative risks of contact by an oil spill of the different targets should a spill occur within discrete areas of the proposal, or the existing leases, or along transportation routes.

These transportation routes are highly idealized paths assumed being followed by cargo vessels (including tankers) within the study area. For these idealized paths, the assumptions were: ships would follow the shortest route to port of destination (on a globe, a great circle); northbound vessels would follow-where applicable--the Gulf Stream; oil imports from Indian Ocean sources would move partly through the Suez Canal and partly around the southern tip of Africa, the Cape of Good Hope; imports from North Sea fields would approach the study area through routes in the vicinity of Georges Bank, off Massachusetts; and crude oil from the proposed leases would be transported to shore facilities through those same routes followed by tankers carrying import products. These probabilities are found in Tables 3 to 11 in Appendix C of this EIS, tabulated under three time periods: 3, 10, and 30 days. Since the model does not account for weathering or cleanup of the spilled oil, these time periods allow for some consideration of the level of toxicity (highest within 3 days), possibility of cleanup operations (usually attempted and achieved within 10 days), and a period beyond which oil spills are considered not detectable (after 30 days). When using these probabilities, various approaches are used by the different analysts, according to their needs and/or characteristics of the vulnerable resources at issue. In some cases, the analyst--after comparing the relative risks to the resources --will discuss and assess the impacts to targets located in areas of highest vulnerability to a spill greater than 1,000 barrels in size that is assumed to have occurred and traveled under the influence of winds and currents for a period of 30 days. Also considered are relevant characteristics (population status, seasonality, etc.). Then, a conclusion is reached about the projected level of impact (major, moderate, minor, or negligible) to each individual resource in the high risk areas. Next, the risks to remaining resources in the region are compared to those in the area with highest contact probabilities, and conclusions about the relative level of impacts are then reached--based on the high risk levels. In other cases, the analyst does not reach conclusions based on these probabilities, but considers them in addition to the "joint" probabilities (see Overall Risks-the Joint Probabilities).

It should be emphasized that if only these probabilities are considered, the various levels of impacts projected are based on the assumptions that a spill greater than 1,000 barrels in size has occurred and traveled under the influence of winds and currents and do not take into consideration the probabilities that no oil exists in the sale area or that--if oil is found--the probabilities that no oil spill may occur. In other words, the premises in this part of the analysis are: oil exists in commercial quantities; exploration, development, production, and transportation activities take place; and, as a result of these, spills associated with the proposed action, existing leases, and imports of hydrocarbons have occurred in the region.

Overall Risk--the Joint Probabilities

There is a critical difference between the conditional probabilities discussed above and the overall, or joint, probabilities discussed below. Conditional probabilities depend only on the surface winds and geostrophic currents in the