At this stage, it is more obvious why the depth grid had to be extended seaward beyond the intended locations of the open boundaries of the model. Referring for example to Figure 2.1, the depth grid extends well beyond the intended open boundary M-N. Consequently the Cartesian grid of Figure 3.1 has squares with valid water depths in this trans-boundary region. This means that clusters at or near the line M-N can grow to proper size, according to the specified area/depth law, and that interior nodes are correctly spaced even where an open boundary of the model is to be located. If the Cartesian grid reached only to the line M-N, clusters being assembled close to the open boundary might either have to be aborted or be too small or distorted in shape, resulting in poorly spaced nodes close to the boundary.
There is a disadvantage in the above procedure, however, since model nodes have been created in the region MNCB and the model grid formed when these nodes are triangulated will extend beyond the required open boundary M-N, just as the depth grid does. Removal of the surplus portion MNCB (and of the similar unwanted extensions PQED and RSAF at the other open boundaries) can be done either before or after triangulation, by the NODER or SPLITTER programs respectively. In this case, it will be done later with the SPLITTER, after triangulation, since that serves to introduce the latter program, which has other uses.