Why are particles advecting where there are no velocity values?

[brucelaughlin]

I seem to have particles entering and moving around within the San Francisco bay, but my ROMS files have the SF Bay masked. I understand that Opendrift uses a high resolution coastline file for plotting, but what might explain advection in areas where I'm not providing velocity values?

Please see the included plots; one is an Opendrift plot of trajectories, and one is of a u velocity field for roughly the same region.

Also, from the log, it appears that the land mask from the ROMS files is being used:
opendrift.models.basemodel:936: Using existing reader for land_binary_mask