Grid-Stat config file interp setting

[jwraby]

Is it possible to set the interp to "NONE" to prevent smoothing? The User's Guide provides the following possible settings:
FCST,OBS,BOTH.
If one cannot use "NONE" is there a reason for not allowing users to use unsmoothed fields for verification?

[JohnHalleyGotway]

Hi @jwraby, I see you have a question about the interp option in the Grid-Stat configuration file. You want to make sure that no smoothing is actually being performed on the data. And that is indeed the case.

Here's the default interp setting from the default Grid-Stat config file:

interp = {
   field      = BOTH;
   vld_thresh = 1.0;
   shape      = SQUARE;

   type = [
      {
         method = NEAREST;
         width  = 1;
      }
   ];
}

As you can see on this line of the smooth_field() function, calling it with nearest neighbor with a width of 1 simply returns and does no actual work.

I can see how this it may be more intuitive to support a field = NONE setting, but the good news is that it is already behaving as you'd expect.

Do you think it'd be worthwhile to add support for field = NONE and update the default Grid-Stat configuration file to use that?

[jwraby]

OK. Ran into a problem when running plot_data_plane. I specified the input as the Grid-Stat ncpairs output file, but the name of the field contains white space:
ERROR : VarInfo::set_magic() -> embedded whitespace found in the name "2mAGL_TMP_ANAL at Surface" or level "(,)" string. I obtained the long_name using ncdump on the ncpairs file.

The field name originates from the attrs dictionary contained in the python interpolation script as follows:
attrs = {

'valid': '20161112_120000',
'init': '20161112_000000',
'lead': '000000',
'accum': '000000',

'name': 'Z2TMP',
'long_name': '2mAGL_TMP_ANAL',
'level': 'Surface',
'units': 'kelvin',

'grid': {
'type': 'Lambert Conformal',
'hemisphere': 'N',
'name': 'lambert',
'scale_lat_1': 25.0,
'scale_lat_2': 25.0,
'lat_pin': 31.41,
'lon_pin': -107.56,
'x_pin': 0.0,
'y_pin': 0.0,
'lon_orient': -106.0,
'd_km': 2.5,
'r_km': 6367.47,
'nx': 79,
'ny': 79,
}

When I ran Grid-Stat using python embedding, the ncpairs file was generated with the long_name containing the whitespace, but I'm not sure why this happened. Can you see the issue in the info I put in the above dictionary? Thanks.

[JohnHalleyGotway]

John, I don't see anything immediately obvious. You could send a sample data file along with the command line you ran to produce the error message. If I can replicate the runtime error locally, I should be able to figure out what's going on.

[jwraby]

Belay my last reply on the issue with field name containing whitespace. I had an operator error with using field names from the wrong ncpairs file. When I used the correct ncpairs field name it worked and the plot was generated by plot_data_plane. Furthermore, this plot shows the expected jagged appearance of the contours for the grid spacing of 2.5 km and (79,79) array dimensions! This means pretty conclusively, that the appearance of the plot generated by IDV was misleading me about the cause of the problem. Like you were suggesting, it appears that the plot settings are indeed causing the issue. Particularly, the settings in IDV need to be revised so that the expected jaggedness can be resolved. Thanks for your help on this!

[jwraby]

I concur with your recommendation to add support for field = NONE and update the default Grid-Stat configuration file to use that? It wasn't obvious to me that there was the option to perform verification on unsmoothed fields from the Grid-Stat config file nor from the User's Guide. Thanks for considering this enhancement.

[JohnHalleyGotway]

@jwraby, FYI, I wrote up this issue (dtcenter/MET#2510) to capture this requested change to the Grid-Stat config file. Not sure how it'll slot into our competing development priorities, but I just wanted to give you a heads up. I'll lock this discussion to prevent future posts. But feel free to comment directly on dtcenter/MET#2510 if you have any anything else to share about that potential change.