I'm having great fun routing the Three Bridge Fiasco for tomorrow, where there are three marks about 7 miles apart in a rough triangle, in SFBAY. The course is go around all marks, cross the start/finish line at start and finish in either direction... 

Question: where the wind and currents are notoriously variable, and the shoreline, islands, bridges, and exclusion zones make some of the available channels small, what is the effect of changing the grid size and time resolutions ?

Currently successful at 0.25 nm and 30 min. Not sure if there's any point in being that small, or would coarser not only run faster, but give equal optimization in a practical sense. 

Should those values be proportional to the expected Boat speed ? e.g if I am using 0.25 nm and go 4 kts, then should be at ~5 min time steps? 

The routing is coming out with 3-5 min steps already, which are incremental distances close to the grid..