Lagrangian floats

[onken]

I have inserted 67 Lagrangian floats in a ROMS model in shallow water
(max water depth around 40 m). I expected the floats to move along
(potential) density surfaces in order to conserve isopycnal potential
vorticity, but several floats penetrate the density surfaces.
Here are my questions:
(1) The vertical position of the density surfaces has been computed from the
variable "rho" in the ROMS output. Is "rho" in-situ density or
potential density?
(2) How is the vertical displacement of the floats
computed? Along density surfaces or s-surfaces? Or don't the floats
know anything about density or vertical coordinates?

Thanks,
Reiner

[kate]

rho is in situ density which is what we need for horizontal pressure gradients.

From a floats.in file:

T Float trajectory type (1: Lagrangian, 2: isobaric, 3: Geopotential)

The Lagrangian ones move with the local velocity. The others behave differently.