Forcing Files

[gloeg002]

I am trying to create a forcing file and receive the following error :

Code: Select all

 NETCDF_INQ_VAR - requested variable:  xi_u
                  not found in input file:  /Users/lucasgloege/ROMS/model-001/forcing_file_nointerp.nc
                  call from:  get_varcoords.F

 GET_2DFLD   - error while reading variable: sustr   at TIME index =       1

Here is the result of ncdump -h :

Code: Select all

netcdf forcing_45001_nointerp {
dimensions:
	xi_u = 303 ;
	eta_u = 147 ;
	xi_v = 304 ;
	eta_v = 146 ;
	sms_time = 48 ;
variables:
	double sms_time(sms_time) ;
		sms_time:long_name = "surface momentum stress time" ;
		sms_time:units = "day" ;
		sms_time:field = "sms_time, scalar, series" ;
	float sustr(sms_time, eta_u, xi_u) ;
		sustr:long_name = "surface u-momentum stress" ;
		sustr:units = "Newton meter-2" ;
		sustr:field = "surface u-momentum stress, scalar, series" ;
		sustr:time = "sms_time" ;
	float svstr(sms_time, eta_v, xi_v) ;
		svstr:long_name = "surface v-momentum stress" ;
		svstr:units = "Newton meter-2" ;
		svstr:field = "surface v-momentum stress, scalar, series" ;
		svstr:time = "sms_time" ;
}

Any suggestions? I thought I had to add xi_u, eta_u, xi_v, and eta_v to my forcing file, but that appears to not be the case.

Thank you for your time.

[kate]

When providing such forcing files, there are two options:

1. If the eta_u, xi_u dimensions match that of ROMS, no interpolation is done. No extra geometry need be provided. Your values are consistent with Lm=302, Mm=145.

2. If the dimensions do not match, ROMS assumes that you need interpolation to be done. Your error call refers to get_varcoords, which is only called from the regrid routine. This will need more geometry, as stated at the top of get_varcoords.F:

Code: Select all

!  This routine reads the spatial positions of any NetCDF variable     !
!  having the attribute  "coordinates",  as specified by CF rules.     !
!  For example, in CDL syntax:                                         !
!                                                                      !
!       float my_var(time, lat, lon) ;                                 !
!               my_var:long_name = "my variable long name" ;           !
!               my_var:units = "my variable units" ;                   !
!               my_var:coordinates = "lon lat time" ;                  !
!               my_var:time = "time" ;                                 !
!                                                                      !
!  The following "coordinates" attribute is also allowed:              !
!                                                                      !
!               my_var:coordinates = "lon lat" ;                       !
!                                                                      !
!  That is, the time variable "time" is missing in the "coordinates"   !
!  attribute.                                                          !

[gloeg002]

Thank you Kate for clearing that up. I resolved the issue, it was a silly typo I should have caught.

However, now I am intrigued that ROMS will interpolate for me. Lets say I have wind speed data from 3 buoys on Lake Superior and I convert each time series to surface stress. How can ROMS interpolate the data from 3 points over the entire grid? What should I be reading about to do this? Also, what sort of interpolation scheme does ROMS use?

Thanks again for your help! I appreciate it.

[kate]

ROMS has limitations on what it can do. The input fields must be on a regular lat,lon grid extending at least as large as your ROMS grid. We use it with global atmospheric model results. There's some information here.