[Met_help] Several questions about MET

John Halley Gotway johnhg at rap.ucar.edu
Wed Jun 18 11:03:31 MDT 2008


Let me answer your second question first.

It's fine to use data on any grid with MET - even a user defined grid - so long as it's a lat/lon, lambert conformal, or polar stereographic projection.  We require you to specify a masking region(s) 
to define the area of which you'd like to accumulate statistics.  Often users like to look how the statistics vary in different regions of the domain.  If you'd like to score over the entire domain, 
then yes, your polygon idea is what I'd suggest.  And no, you do not need to trace the domain.  I'd suggest choosing 4 lat/lon corner points which fully enclose your domain.  They don't have to be the 
exact corner points - they could be further out than that.  Perhaps in future releases, we should support a special entry for the "mask_grids" configuration item that would instruct Grid-Stat to score 
over the entire domain, rather than having to mess with defining a polyline as I described above.

Regarding your first question, I would say that no, this probably doesn't have to do with GSL being compiled with GNU, as opposed to PGI.  That may or may not be a problem down the road.  I haven't 
tried it.

Could you do me a favor and send me those two files:

I'll try to figure out where that error message is coming from.

John Halley-Gotway
johnhg at ucar.edu

Jan Ploski wrote:
> Hello,
> I have several questions about METv1.0 that are not covered by the
> FAQ/known issues sections of the web page. I'm interested in getting
> grid_stat working and having some initial setup problems.
> 1. Running the tests after the build produces the error reproduced
> below. I also get the same error when trying to run grid_stat. I suspect
> that it may have something to do with my compilation settings.
> Everything except for the GSL library was compiled with PGI - GSL was
> compiled with GNU - could this be the cause? Which steps would you
> recommend to diagnose this problem further?
> *** Running GRID_STAT using netCDF input for both forecast and
> observation ***
> Forecast File: ../out/pcp_combine/sample_fcst_12L_2005080712V_12A.nc
> Observation File: ../out/pcp_combine/sample_obs_2005080712V_12A.nc
> Configuration File: config/GridStatConfig_APCP_12
> NetCDF Precipitation (min, max) = (0, 38.8841)
> NetCDF Precipitation (min, max) = (0, 104.1)
> ncvardef: ncid 14: Attribute or variable name contains illegal characters
> 2. We'd like to run MET over WRF output for a user-defined domain (over
> Europe/Germany in Lambert Conformal projection). I understand that
> grid_stat can accept GRIB from wrfpost as input. However, I am required
> to specify a grid number for masking in the configuration file. The grid
> number 255, as produced by the wrfpost, is rejected, even if I include
> LATITUDE/LONGITUDE fields in the GRIB file. Do I assume correctly that
> MET cannot process user-defined grids and that I *have* to regrid output
> to one of the predefined grids to use MET? Or would it be reasonable to
> define a polygon describing my custom grid? If so, would it be
> sufficient for the polygon to consist of the extreme (lat, lon)
> coordinate pairs of my grid, should it trace all the outer grid points,
> or do the locations of inner points also play a role? Why is specifying
> a mask (either grid number or polygon) mandatory anyway - does the input
> GRIB file not contain enough information?
> Best regards,
> Jan Ploski
> _______________________________________________
> Met_help mailing list
> Met_help at mailman.ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/met_help

More information about the Met_help mailing list