| |t:==:t| Feature | Relative cost |
| Bathymetry | 100 if draft < depth, 10 if draft = depth, otherwise 0 |
| Land | 1000 (impossible to cross) |
| Mudflats | 500 |
| Shoreline | cost = F(distance to shore) |
| Residences | cost = F(distance to residence ) |
| Seagrass | 20 |
| ICW | -5 if in ICW, otherwise 0 |
| No-wake zones | 10 |
| |b:==:b| |
To determine the cost of a boat to navigate the bay, it was first necessary to calculate the relative costs of relevant environmental features in the bay. Table 4.13 shows the relative costs of traversing various environmental features in the bay. The origin and destination pairs and the cost surface are processed in a raster GIS; all possible paths are considered between the two points, and the path that incurs the least cost in transit is selected for output.
No data sets were available that could directly supply values for the various items in the cost table; subjective estimates of the costs of each feature were obtained to construct test cost surfaces, and the values were adjusted until the boat paths produced something that approximated reasonable boat path (did not cross land, avoided shallow areas, etc.). Absolute values of the costs are not important by themselves, rather the difference in costs between features. It is logical to presume that land is much more costly to cross in a boat than open water, but there is no empirical justification for making that cost 1000 times greater.
The first few items show features that are absolutely necessary for navigation. First, the water has to be deep enough to float the boat (deeper than the boat's draft). A prohibitively high cost is applied to all areas where the water is too shallow, a moderate cost when the draft is equal to the water depth, and low cost if the water is deep. Similarly, a prohibitively high cost is applied to land to ensure that no boat attempts to cross over land to achieve its destination. Shallow water (shallower than the draft of the boat) is given a cost less than that of land, since there are possible scenarios where a boat can cross shallow areas for short distances (boat may have a retractable rudder or drive shaft, boat may be able to cross on a plane, or tidal variations could create temporary passage over such an area: all of these scenarios are abnormal navigation conditions). There are no feasible scenarios that were considered in this simulation that allow for a boat to cross dry land.
The remaining features have lower costs, but are included in portions assumed to be proportional to their influence on boater behavior. There is a cost of passing close to shore and to residences proportional to the distance from them. This represents the boater's desire to stay away from land and respect people's privacy if all other factors are equal. Seagrass beds and no-wake zones also have a moderate cost of transit; these are equivalent to inconveniences, which can be traversed if needed but avoided when alternatives are available. The ICW has a negative cost, which is effectively a discount to boats that travel along it. This translates to a relative peace of mind for the boater to navigate a marked channel. The benefit of the marked channel is minor in comparison to the other environmental costs.
All of the environmental features were translated into raster GIS coverages at 10m resolution per pixel and stored as a GRASS GIS coverage. As each origin-destination pair was identified for transit in the schedule, the information was given to a script in GRASS GIS to calculate the cost surfaces and the least cost path between the points.
All environmental features except bathymetry were considered to be identical for the various kinds of boats; i.e., a sail boater was assumed to have the same attitudes as a speed boater about infringing on a residential area or crossing a seagrass bed. This could be improved by assigning varying costs to boaters for crossing environmental features to reflect their relative reluctances to cross seagrass beds, for example. However, such alternatives were not explored in this study.
After the GRASS GIS functions had determined the least cost path between the origins and destinations, the path was written as a separate coverage or data layer. The various data layers then were written as a static representation of boat activity to a series of maps for analysis.
The resolution and size of the area chosen for this study resulted in a bottleneck for the simulation runs at this point. With the available computing equipment (a PC with a pentium 100MHz processor and 32M of RAM, running linux 1.3.93 and grass 4.1), two hours was typically required to determine a single path from an origin to a destination. This meant that a single run of this simulation required several days to complete. A more efficient way of determining paths that the boats take is one of the most important design considerations for future work with this model.