[Met_help] Problem running Grid Stat

[John Halley Gotway]

Stefan,

Unfortunately, I think you're out of luck.  MET currently supports the following 4 projection types: Lambert Conformal, Polar Stereographic, Plate Carree (lat/lon), and Mercator.  That covers the vast
majority of the data we encounter.  I'll pass along  your email to the person who's implemented our grid code to let him know about the orthographic projection you're using.

Also, I wanted to give you some background about the Point-Stat and Grid-Stat tools.  These two tool are very similar - they compute many of the same types of statistics.  The difference is in the
type of observations they handle.  Point-Stat compares a gridded forecast to point observations.  When doing so, it performs horizontal and vertical interpolation in computing matched pairs.
Grid-Stat, on the other hand, compare a gridded forecast to a gridded observation ON THE SAME GRID.  We do not perform interpolation in Grid-Stat since we require to the data to be on the same grid.
And it's the user's responsibility to put the data on the same grid - which is pretty easy when using GRIB data.

Hope that helps,
John Halley Gotway
johnhg at ucar.edu

Stefan Kiesenhofer wrote:
> Hi, John,
> 
> Sorry, I didn't make my question clear.  What I mean is that our data
> resides on an orthographic projection plane where the plane contacts the
> surface of the earth at 48.8 degrees north and 7 degrees east. The mesh size
> on the plate is constant 8 km, whereas every grid point x, y have been
> converted into lat/lon. Therefore the data looks pretty garbled.
> So I tried to find out the correspondence between the contents of the NetCDF
> attributes and the contents of the GRIB table
> (http://www.nco.ncep.noaa.gov/pmb/docs/on388/tabled.html) for SPACE VIEW
> PERSPECTIVE OR ORTHOGRAPHIC grids.
> 		
> 		:Projection = "Orthographic" ;
> 		:lap_deg = "48.800000 degrees_north" ;
> 		:lop_deg = "7.000000 degrees_east" ;
> 		:delta_x_km = "8.000000 km" ;
> 		:delta_y_km = "8.000000 km" ;
> 		:nx = "141 grid_points" ;
> 		:ny = " 99 grid_points" ;
> 
> Unfortunately, when running grid_stat now, this content leads to 
> "ERROR: read_netcdf_grid() -> Projection type Orthographic not currently
> supported."
> 
> The forecast and the observation files reside on the same grid, the
> observation file is greater because of the spatial differences between
> several models. I thought, grid_stat is in search of matching pairs? 
> 
> Thanks' for your help,
> Stefan
> 
> 
> 
> 
> -----Ursprüngliche Nachricht-----
> Von: John Halley Gotway [mailto:johnhg at ucar.edu] 
> Gesendet: Freitag, 11. Dezember 2009 20:45
> An: Stefan Kiesenhofer
> Cc: met_help at ucar.edu
> Betreff: Re: [Met_help] Problem running Grid Stat
> 
> Stefan,
> 
> I need to tell you about 3 things:
> 
> First, to answer your question about projections.  You're using an
> "equidistant" projection, which we assume mean "cylindrical equidistant".
> That's also referred to as a Plate Carree or a Lat/Lon
> projection.  In your NetCDF file, please add the following as global
> attributes:
> 
> :Projection = "LatLon" ;
> :lat_ll_deg = "-90.000000 degrees_north" ;
> :lon_ll_deg = "0.000000 degrees_east" ;
> :delta_lat_deg = "1.000000 degrees" ;
> :delta_lon_deg = "1.000000 degrees" ;
> :Nlat = "181 grid_points" ;
> :Nlon = "360 grid_points" ;
> 
> Where you ...
>    - Set "lat_ll_deg" and "lon_ll_deg" to define the lower-left corner of
> your grid.
>    - Set "delta_lat_deg" and "delta_lon_deg" to define the change in lat/lon
> values.
>    - Set "Nlat" and "Nlon" to define the grid dimensions.
> Hopefully that'll do the trick and allow MET to read these files.
> 
> Second, your forecast and observation files appear to reside on different
> grids.  The forecast data, aladfr_euro, is on a grid of dimension 141x99.
> The observation data, veraxx, is on a grid with of
> dimension 166x136.  In order to compare these files in MET, you'll need to
> place the gridded data on a common grid.  Unfortunately, MET doesn't provide
> any utilities for regridding data.  It's the
> user's responsibility to do this.  The "copygb" tool may be used to regrid
> GRIB files, but since you're using NetCDF, that won't help you.
> 
> Third, I used the ncview utility to look at the contents of the forecast and
> observation file you sent.  The data looks pretty garbled.  It appears that
> there may be an issue in how the data is
> getting packed into this NetCDF file.
> 
> Hope that helps.
> 
> John Halley Gotway
> johnhg at ucar.edu
> 
> Stefan Kiesenhofer wrote:
>> Dear MET Help,
>>
>>  
>>
>> When I run Grid-Stat I get  " NetCDF: Attribute not found".
>>
>> I'm using netcdf files created by "ncgen" from CDL files. After reading
> the
>> Met help Email archive, I tried to make my files look like the output of
> the
>> PCP-combine tool.
>>
>> The problem is I think the grid definition. Lat and Lon are computed from
> an
>> equidistant x, y grid. Attached you can see my forecast file
> (aladfr_euro.)
>> and my observation file (veraxx.)
>>
>> How I can define my grid?
>>
>>  
>>
>> Thanks,
>>
>>  
>>
>> Stefan Kiesenhofer
>>
>>  
>>
>>  
>>
>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>
>> Department of Meteorology and Geophysics A - 1090 Vienna, 
>>
>> Althanstrasse 14, UZA II
>>
>>  
>>
>>
>>
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Met_help mailing list
>> Met_help at mailman.ucar.edu
>> http://mailman.ucar.edu/mailman/listinfo/met_help
>