wind transition probability matrix. Seasonally specific first-order transi- 

 tion matrices were derived from historic data covering 5 years of observa- 

 tions at the Georges Bank and Nantucket Shoals weather towers. 



Surface current velocity fields used in the model were the same for both 

 verification and forecasting modes, but changed from December to January. 

 Two wind drift angles, 0° and 20°, were used in the trajectory simulations, 

 and sensitivity of the results to this parameter is discussed below. 



Before the grounding of the Avgo Merohant, model runs conducted as part 

 of the oil spill risk analysis for the Georges Bank area had indicated that 

 by far the most likely trajectory for oil spilled in that area would be to 

 the southeast (Figure 2-11), a result which was later borne out in movements 

 of Avgo Merchant slick. However, the opportunity for more detailed and 

 precise testing of the model arose with the availability of voluminous data 

 on slick location and winds provided by NOAA and USCG personnel on the scene 

 and by National Weather Service offices in New Jersey and Boston. 



Time series of observed winds were input to the model in order to make 

 deterministic simulations of slick trajectory. A series of points were 

 released every 3 hours from the site of the Argo Merchant grounding and 

 transported as described above by currents and observed 3-hour ly winds, 

 beginning with the reported wind at 1600, December 17. A display of all such 

 points after n time steps should resemble the location and general shape of 

 the slick at a time 3n hours from 1600, December 17. Figures 2-12 and 

 Figures VI- 18 to VII-21 show displays for two dates in December, based on 

 observed winds from two locations. Maps of the actual slick were provided 

 by NOAA and USCG personnel involved in overflights of the area. The model 

 predictions do not take into account differential rate of oil released from 

 the ship or oil dispersion over time. In Figures VII-17 to VII-19 the wind 

 input is not modified by any assumed drift angle, and it is clear in Figure 

 VII-19 that, in the case of wind data from the USCGC Vigilant, use of a drift 

 angle to the right would have narrowed the gap between observed and predicted 

 patterns. 



One might be inclined to deduce from the above that wind data from the 

 Vigilant were most representative of conditions along the oil slick and that 

 choice of a positive drift angle was appropriate. However, although the 

 Vigilant was at the scene of the Argo Merchant during this time, the Nantuc- 

 ket Light Ship was anchored near 40°30'N latitude, 69°29'W longitude, actu- 

 ally closer to the southeastern extremity of the slick. Thus, data from 

 neither station were clearly preferable on the basis of location. 



Figures VII-20 and VII-21 show comparisons of predicted and observed 

 slick locations based on the assumption of a 20° drift angle to the right. 

 In summary, no clear conclusions can be drawn concerning the best choice of 

 wind station and drift angle in the absence of more information. 



In the days following the grounding of the Argo Merchant, model runs 

 were regularly made to estimate the probability that oil would ultimately 

 affect the coast. In each run, large numbers of simulated trajectories were 



53 



