3Q S * 



-d~ - ° • ™ 



This condition implies that along the left side of the grid, 



Q s *-U,j,n+1) = Q s *(3,j,'n+1) , (80) 



j = 3,5 • • • JM 



and along the right side, 



Q s JlM,j,n + l) = Q s *(IM-2,j,n+l) , (81) 



j = 3,5 • • • JM . 



The S*-directed transports for even j and i=2 or IM-1 are deter- 

 mined by the average of the two neighboring interior values. Addi- 

 tionally, the T*-directed transports are required along i=2 or IM-1 

 and are calculated from equation (56) where D is again the average 

 of the local fluid depth along the boundary. 



Special computations are required at seaward corner points (2,2) 

 and (3,1). Simultaneous equations are solved for Q T *(2,2,n+l) and 

 Qg*(3,l,n+1) with the approximation that Qy* (3, 1 ,n+l) is the aver- 

 age of the two neighboring interior values. Similar expressions are 

 used at the right-hand seaward corner points. 



Other conditions at the seaward and lateral boundaries were 

 experimented with for the Hurricane Carla simulations. These in- 

 cluded radiational conditions of the type discussed by Reid and 

 Bodine (1968) and Reid (1975). The main differences, as anticipated, 

 were close to the lateral boundaries where the "flow through" condi- 

 tion in equation (79) gave a more realistic rendition of the long- 

 shore flow. 



The basic numerical model was used successfully by Alvarez 

 (1973) . Further testing of the algorithm, where in some cases an 

 analytical solution was possible for comparison purposes, is pre- 

 sented in Appendix G. 



5. Wind and Pressure Fields. 



Of major importance in simulating the storm surge is the accurate 

 portrayal of the hurricane wind field on the computing grid. More- 

 over, the representation of the wind must be time-dependent. 



The Hydrometeorological Section of the National Weather Service 

 (NWS) provided charts of the surface winds at 10 meters above mean 



73 



