frequency relationship is required. This data base had been generated for the 

 study area during the conduct of a previous CERC study for Long Island, 

 New York. 



The goal of the Long Island study was to develop reliable stage -frequency 

 relationships for specific locations in the study area. The goal was accom- 

 plished through a numerical modeling effort which computed the propagation of 

 storm surge and tidal data from deep water into the shallow water study area. 

 Tidal and storm surge hydrographs were generated for every grid location as a 

 function of the specified deepwater boundary condition. Site specific 

 stage-frequency relation-ships were developed by simulating the shoreward 

 propagation of a data base of stochastically and historically generated 

 boundary conditions. 



The model used for the numerical simulations is the WES Implicit Flooding 

 Model (WIFM) described by Butler (1978). WIFM incorporates an alternating 

 direction implicit (ADI) finite difference algorithm to solve the depth 

 integrated shallow water wave equations at each cell of the computational grid 

 shown in Figure 4. The model was calibrated for tides to the primary M2 tidal 

 constituent and verified by reproducing an observed mixed tide condition. 

 Verification to both hurricanes and northeasters was achieved by simulating 

 documented storms of record and comparing the computed results to recorded 

 stage level observations. Details of the storm verification can be found in 

 Butler and Prater (1986). The generation of the study area data base is 

 described below. 



Hurricanes and northeasters were used as the storm surge boundary condi- 

 tions. Because of the basic differences in the characteristics of each storm 

 type, the two were treated separately. Wind speed and direction data were 

 specified for hurricanes according to the Standard Project Hurricane criteria 

 (National Weather Service 1979). A joint probability method was used for 

 establishing hurricane stage -frequency curves for the study area. This 

 procedure involved the identification of the following five storm parameters: 

 (1) central pressure deficit, (2) radius of maximum winds, (3) forward speed, 

 (4) direction of propagation, and (5) point of landfall. Each parameter was 

 assigned a probability based on the historical occurrence of hurricanes in the 



137 



