115 



able for intersecting waves, is only accurate to fourth order, and only being applicable 

 in deep or transitional water. This precluded an analysis of the behavior of the so- 

 lution in shallow water. When the method is applied in shallow water, the lessons 

 learned in the two dimensional case should be applied, and a higher order solution, 

 perhaps to fifth order, should be used. 



It is also the case that the Ohyama solution is a single solution that is accurate 

 to fourth order globally. The local nature of the LFI method allows each separate 

 window in time to be represented by a unique solution, representing the entire record 

 with far more accuracy than a global solution of the same order. 



Choice of Window Width 



As mentioned in the previous section, it is important to keep the window width at 

 a minimum. For the short crested wave shown in Figure 5.27, the standard window 

 width was one tenth the zero-crossing period (O.lTj). It was necessary to widen the 

 window to 0.15T^ at three locations in order to find a solution. This indicates that 

 the standard window width is set close to the smallest size that would be eff^ective. 

 In general, it is necessary for the window to contain enough of the record to define 

 the local curvature, and include sufficient data points to reasonably expect to define 

 a high order solution. Most often, the limiting factor will be the sample rate of the 

 data. For example, the above examples are computed for a short crested wave with 

 a period of ten seconds. In those examples, the window width of 1/10 the period 

 (Is) was adequate to capture most of the wave. Unfortunately, field records are often 

 sampled at only IHz. At this sampling frequency, and a one second window, there 

 would be a maximum of only two data points in each window, and it may be necessary 

 to widen the window to include more data. 



Effects of Array Size 



The examples above are all computed from data sampled by a fairly small array 

 (Fig. 5.3). The 1.6m size of the array is about 0.01 of the wavelength of 10s wave in 

 deep water. That particular size was chosen because it has the same dimensions as 



