setup can act to keep the water elevation raised in the prototype data. Since 

 WIFM cannot model this effect, the computed water level at Hatteras begins 

 to fall after the peak of the stcira. Prototype data for Nags Head are not 

 available (the tide gage disappeared during the storm) but estimates of 

 the surge have been made by SAU from other information. I'his estimate ap- 

 pears with computations for Nags Head in Plate 13 with no datum shift esti- 

 mates removed. The computed and estimated prototype records agree in form, 

 and the roughly 2-ft elevation difference can be accounted for by datum 

 shifts, wave setup, and errors in Grygiel's estimate of the record at Nags 

 Head. 



46. Plates 14-16 compare the computed storm surge of Hurricane Donna 

 v;ith prototype data for the offshore model. Ten sites are shewn, and the 

 datum shift estimates of Table 3 have been removed from the prototype 

 records. Atlantic Beach and Nags Head, both sites on the open ocean, show 

 the best agreement. For sites in the bay, the effects of wave setup and 

 shoreline approximations play an important role in the agreement between 

 computations and field records. An added factor is the speed at which Hur- 

 ricane Donna moved across the North Carolina coast. The eye of the storm 

 traversed the distance from Cape Tear to Cape Henry in about 8 hr, causing 

 rapid changes in wind speed and direction for sites in Pamlico and Albemarle 

 Sounds. The rapid movement of the storm, along with the shoreline simpli- 

 fications of the Neuse and Pasquotank Rivers, caused some phase difference 

 between computations and the prototype records at Cherry Point, Minnesott 

 Beach, and Elizabeth City. Columbia, a site up a narrow recess of Albemarle 

 Sound, also has a phase shift. The prototype records for Hatteras and 



Nags Head show the effect of wave setup at the peak of Hurricane Donna 

 (hour 22) but these records agree well with computations. Despite the shore- 

 line approximations for Stumpy Point Bay in the offshore grid, the results 

 match well at this site. The computations at Englehard overpr«;dict the 

 surf,e, but the results agree at hour 22, the time of maximum f].ow at Oregon 

 Ir.let. 



47. Plates 11-16 illustrate that the offshore model can simulate both 

 historical northeasters and hurricanes to a reasonable accuracy, with a mini- 

 mal number of grid cells. Considering the accuracy with which the datum 

 shift adjustments in Table 3 are estimated, the comparisons of computations 

 with field data for the offshore model are quite good. • , 



31 



