zgrid boundaries in floats output file

[colberg]

Hi all,
we are running a ROMS simulation with

define FLOATS
define FLOAT_VWALK

we have 30 vertical layers defined and hence values for zgrid should be between 0, 29.
However, the model computes values below zero (up to -400) and greater than 30. This happens mostly in shallow areas related to river channels and coastal points. However, the depth values of affected floats is not unreasonable.

When we

undef FLOAT_VWALK

the zgrid variables behaves as expected. So the vertical random walk appears to be the culprit. Has anyone encountered similar behaviour ?

thanks
Frank

[wilkin]

What turbulence closure scheme are you using? That affects how Akt is computed that sets the random velocity.

Is FLOAT_STICKY #defined? That affects whether floats reflect from the surface and bottom. You might experiment with setting this to see what happens - it could help diagnose the issue. Around line 973 in step_floats.F you'll see that it appears the extreme zgrid values you report are explicitly prevented.