One-way analyses of variance (anova) without replication were performed 

 to determine whether statistically significant differences were detectable 

 between observed and predicted wind factor and direction. The results are 

 summarized in Tables 2-8 to 2-11. 



The following conclusions can be drawn from the above analysis: 



o For short-term predictions (24 hours) a vectorial addition 

 of wind and tides adequately forecasted movement. 



o For long-terin predictions the winds dominated the movement of 

 the oil. 



o 3.5% of the wind speed is an adequate value for the wind 

 drift factor. 



o Oil spills tend to move downwind. 



2.3.2 Center for Experiment Design and Data Analysis 



Another modeling study was carried out by NOAA personnel at the Center 

 for Experiment Design and Data Analysis (CEDDA) at the request of the OSC on 

 December 28. In this study, a wind drift (3% at 15° to the right) , based on 

 15 years of historical wind records from the Nantucket Light Ship were con- 

 sidered both alone and in combination with an assumed local current estimated 

 from measurements taken at WHOI's location site D (0.25 knots at 270°). 

 Several cases were run using both winter (January-February-March) and spring 

 (April-May-June) wind records. These modeling results were forwarded to 

 Cmdr. Morgan for his use in upgrading the forecasting effort. 



At CEDDA, the centroid of hypothetical spills were tracked to model the 

 probability of impact from an offshore oil spill. Such processes as oil 

 spreading and weathering were not included. Tidal currents were neglected, 

 and surface slick movement was considered to be a linear combination of a 

 seasonal baroclinic current (sea current) and a time-dependent wind-driven 

 current. The spatial field of seasonal baroclinic currents was input to the 

 computer program, and a wind-driven current vector was calculated based on a 

 3% wind factor adjusted to include a 15° Coriolis deflection angle between 

 wind drift and wind vectors. The wind field was taken directly from coastal 

 meteorological station data at Nantucket Light Ship from 1955 to 1970, ob- 

 tained from the National Climatic Center in Asheville, North Carolina. 



A hypothetical path of oil movement was generated by computing trajec- 

 tories from each third hourly wind observation. After each 3-hour ly step, 

 the geographical position of the centroid of the hypothetical slick was 

 compared with the beach location. When this position and the beach location 

 coincided, an impact event was assumed and the procedures terminated. Upon 

 impact, the wind record used for the wind-drift current calculation was 

 advanced 72 hours and modeling of a new spill event was begun. If no beach 

 impact was found within a modeling time of 1200 hours, the oil mass was 

 assumed degraded and the procedure again terminated. Approximately 40 



48 



