grid cell switches from dry to wet while (u,v) are NaN

[xwu]

hello,

I want report an issue that I encounter recently. I am running a realistic ROMS-only test that receives freshwater input from a small river (see attached figure). This test incorporates tides and atmospheric forcing (winds, heat flux etc). The grid spatial resolution is around 8-20m. Wet-dry is turned on. During flood tides, water enters into the watershed of the river and grid cells within the watershed can become wet. The minimum water depth for wet-dry is set as 0.25m.

The issue is, within the watershed a grid cell can switches from a dry to wet cell ('wetdry_mask_rho' is 0 at the first time step, and then becomes '1' at the next time step, also see red arrow in panels (a1) and (b1) in the attached figure, where '\eta' slightly increases and exceeds the 0.25m cutoff). However, within these two time steps, (u,v) at this grid cell location (adjacent to this rho-point) are all NaN values. This result indicates that, the change of '\eta' at this rho-point cell is not due to the (u,v)-driven water mass flux. Does ROMS adjust the '\eta' value numerically, or is there any other reason why the '\eta' value at this rho-point cell changes (like response to the air pressure?). I am glad to provide more information if necessary.

thanks,
Xiaodong Wu

[jcwarner]

the cells will become wet when ever the total water depth is > Dcrit. The water will move from all the processes in the model, such as pressure gradients from gravity or density, wind, horiz mixing, etc.

At the model init, how did you set the water level (zeta) on the grid? did you use an init file, or did you use ana_initial?
sometimes i set the initial water level everywhere to be slightly less than DCrit, this prevents any small round off issues that drive flows from creating wet cells at the start.

-j