grid_stat not reading polygon

[carlos9917]

Hello

I am trying to run grid_stat for a group files like the two attached below.
I defined a region with a polygon, but grid_stat keeps saying the polygon NORTH is not defined.
DEBUG 4: parse_grid_mask() -> parsing grid mask "FULL"
DEBUG 3: Processing grid mask: NORTH
DEBUG 4: parse_grid_mask() -> parsing grid mask "NORTH"
ERROR :
ERROR : parse_grid_mask() -> the mask_grid requested "NORTH" is not defined.
ERROR :

I tried using the polygon to produced a masked region using
gen_vx_mask and that worked, so I guess gen_vx_max could read the polygon

gen_vx_mask obs.nc -type poly polygons/north.poly masked_region.nc

but for some reason grid_stat does not seem to be able to. Then I tried passing the masked region using the nc file generated by gen_vx_max but that did not work either. I am not sure what am I doing wrong.
I am attaching my config file and the input files I am using.

[carlos9917]

I just uploaded the data files with the google form https://docs.google.com/forms/d/e/1FAIpQLSd_DFPopkOCX1NDrIFf1kKjkwdvrrIvUQOgNJ1-5T2Y8E67Zw/viewform

[JohnHalleyGotway]

Hi @carlos9917, thanks for sending along your sample data. You are, in fact, the first user to provide data to us through that new Google Form. Seems like it worked well! Although with small files like this, another option is tar/zipping them and attaching them directly to the discussion.

For example, if you have all the files to send in a directory named for_2895, you'd run:

tar -cvzf for_2895.tgz for_2895

And then attach that for_2895.tgz file directly onto this discussion. For "small" files, either way is fine.

On to the issue at hand...

First, I did run Gen-Vx-Mask on the data you sent, but noticed that the NORTH mask DOES NOT actually overlap with your evaluation domain. You can see that in the -v 3 log messages and/or by plotting the resulting field:

> gen_vx_mask obs.nc north.poly north_mask.nc -type poly -v 3
...
DEBUG 2: Parsed Lat/Lon Mask:	NORTH containing 13 points
DEBUG 3: Polyline Masking:	0 of 800000 points inside
> plot_data_plane north_mask.nc north_mask.ps name="NORTH"; level="(*,*)";

The result is a constant field of 0's.

I always look for background map data to tell me where on earth this is, but don't see any. Looking at the lat/lon values in the north.poly, this sure is a small region!

Next I compared the projection info from the obs file to the north.poly lat/lon values:

> ncdump -h obs.nc
...
:projection = "+R=6371000 +lat_0=80 +lat_1=80 +lat_2=80 +lon_0=-34 +no_defs +proj=lcc +type=crs +units=m +x_0=0 +y_0=0" ;
> head -2 north.poly
NORTH
-33.993204 80.008338

Looking at the range of lat/lon values, I'm wondering if you've listed the north.poly points as (lon, lat) when MET expects them to be (lat, lon). So I swapped the columns, reran gen_vx_mask, and got a much better result:

> cat north.poly | awk '{print $2 $1}' | sed -r 's/^ //g' > north_swap.poly
> gen_vx_mask obs.nc north_swap.poly north_swap_mask.nc -type poly -v 3
...
DEBUG 3: Polyline Masking:	83036 of 800000 points inside
> plot_data_plane north_swap_mask.nc north_swap_mask.ps name="NORTH"; level="(*,*)";

As for how to reference the Gen-Vx-Mask output in the Grid-Stat config file, I see you're using:

mask = {
   grid = [ "FULL" ,"NORTH"];
   poly = ["./north.poly"];
   //poly = ["masker_region.nc"];
}

Please just list the path to the Gen-Vx-Mask output file in the poly setting, not the grid one, like this:

mask = {
   grid = [ "FULL" ];
   poly = ["./north.poly", "/path/to/north_mask.nc" ];
}

Although, either north.poly or north_mask.nc should both work fine, but you should use one of them, not both. Otherwise, you'll see the same statistics for VX_MASK = NORTH written to the output files twice.

Hope that helps.

[carlos9917]

Thank you very much for your help! Now it works :-).