VI. SUMMARY AND CONCLUSIONS 



The development of the conformal mapping problem was completed 

 for transforming the interior region bounded by two arbitrarily- 

 shaped curves and two parallel lateral boundaries into a rectangle 

 in the image plane. Specifically, the coastline and seaward boundary 

 curve taken as the 180-meter depth contour are transformed into the 

 image plane as constant values of n . The bicurve fitting procedure 

 was successively tested on five regions of the Continental Shelf in 

 the Gulf of Mexico and east coast of the United States. The distinct 

 advantages of the orthogonal curvilinear coordinate system over 

 rectilinear representations of the coast are the accuracy and ease 

 of specification of the boundary conditions. For long wave calcula- 

 tions, the numerical simulation of a free long wavelength gravity 

 wave in a quarter annulus demonstrated the superiority of modeling 

 the boundaries in curvilinear coordinates. 



A two-dimensional time-dependent numerical storm surge model 

 employing orthogonal curvilinear coordinates is presented. Since 

 the transformation is conformal, the associated modification of the 

 equations is minimized. The final coordinates allow for the greatest 

 resolution near the coast over the area of principal storm surge 

 development and modification. A second transformation is performed 

 independently on each of the curvilinear coordinates to provide the 

 desired spatial resolution in the shallower region near hurricane 

 landfall. 



Results of the numerical simulation in transformed coordinates 

 of the storm surges along the gulf coast induced by Hurricanes Carla, 

 Camille, and Gracie which crossed the east coast are presented. The 

 surface wind fields are analytically represented. Additionally, a 

 systematic procedure is employed to deform the symmetric wind fields 

 of Carla and Camille in the nearshore region, The deformed Carla 

 wind representation produces better agreement between the computed 

 surges and that observed than a similar simulation with a symmetric 

 wind. Although the pattern of windspeeds produced by the deformation 

 procedure is in good agreement with those provided by NWS, the 

 problem of specifying the proper orientation of the wind-stress 

 vector in the nearshore region requires further study. 



The surge simulations indicate that the model produces results 

 in good agreement with the observed peak surges and hydrographs, 

 especially in the area to the right of landfall. Moreover, the 

 results from the simulation of the Hurricane Gracie surge indicate 

 superior rendition in the curvilinear grid compared with that of the 

 Cartesian grid. Certainly one contributor to any discrepancy between 

 computed and observed water levels may result from the portrayal of 

 the coast as a wall. Proper inclusion of attached bays, lakes, and 

 flooding of low- lying areas should be pursued. 



130 



