Incorrect NetCDF attribute "formula_terms" when SED_MORPH is defined

[rsignell]

The CF Conventions use the "formula_terms" attribute to associate terms in the "ocean_s_coordinate" equation to actual variable names in the NetCDF file. Currently when SED_MORPH is defined, ROMS no longer writes the variable "h" to the history and averages files, but writes a new variable "bath" that is a function of time as well as horizontal coordinates.

If "bath" is written to the NetCDF file, then the "formula terms" attribute for "s_rho" and "s_w" needs to point to "bath" instead of "h".

So change:

s_rho:formula_terms = "s: s_rho eta: zeta depth: h a: theta_s b: theta_b depth_c: hc" s_w:formula_terms = "s: s_w eta: zeta depth: h a: theta_s b: theta_b depth_c: hc"

to s_rho:formula_terms = "s: s_rho eta: zeta depth: bath a: theta_s b: theta_b depth_c: hc" s_w:formula_terms = "s: s_w eta: zeta depth: bath a: theta_s b: theta_b depth_c: hc"

I'm working with the CF group and the NetCDF java folk to allow this more general representation of depth (as time dependent).

-Rich

[arango]

Yes, good catch. The application program now needs to figure out what time record of the bathymetry to use.