Dr. Dana Savidge provided some insight into the computer models  (See here.) that project oil from the Gulf spill reaching the Atlantic coast.

The large scale UCAR models get processes 90% right, but, of course, the devil is in the details, and the details (with respect to concentration, form, and approach of oil to the coast) are NOT resolved by these models AT ALL.

If oil gets into the Loop Current, it inevitable that SOME oil in SOME form will pass by our coast.  But these models do not include any processes that remove or change the oil during passage, such as

1) microbial degradation

2) photochemical breakdown

3) phase transitions (dispersed oil into tarballs and vice versa)

4) lateral mixing and dispersion

All of these things will happen.  We do not know how much, how fast — but all of them will reduce the amount of oil arriving off our shores relative to the model projections, because the model includes no loss terms.  In addition, the “alarmist yellow” of the color scale does not convey intuitively the great extent to which oil will diluted during its transit.

Another important point is that these large scale ocean models are designed to simulate open ocean conditions away from coast.  Coastal boundaries — including all of continental shelves — are poorly represented in these models, which means that any results that show oil reaching (or not reaching) the coast should not be considered reliable forecasts. They are more like random noise or blur at the edges of the bigger image.

The UCAR text associated with the model runs on the web goes into this a little bit, but not as much as it should.