PointStat cannot find matched pairs

[smileMchen]

Software Version

MET-v12.2.0

Text

I run PointStat to verify a regional MPAS history file against GDAS. It runs successfully with output files of point_stat_cnt.txt and point_stat.stat. However, both files are empty except a header (e.g., VERSION MODEL DESC FCST_LEAD FCST_VALID_BEG etc). The log file suggests that "Number of matched pairs - 0 ". Below is what I did:

(1) create prepbufr.gdas.2017020400.nc by running PB2NC

(2) make sure MPAS datafile (history.2017-02-04_00.00.00.nc) has global attributes:
mesh_spec = "0.0"
Conventions = "MPAS"

(3) In PointStatConfig:
ugrid_dataset = "mpas"
ugrid_coordinates_file = "history.2017-02-04_00.00.00.nc"
fcst = {
field = [ { name = "t2m"; level = "Z2"; } ];
}
obs = {
field = [ { name = "TMP"; level ="Z0"; } ];
}

(4) point_stat history.2017-02-04_00.00.00.nc prepbufr.gdas.2017020400.nc PointStatConfig -v 10

Point_stat ran to the end but with empty content in the output file. The log file suggests that "Number of matched pairs = 0 "

Just wonder whether I missed something or did something wrong? Any inputs are highly appreciated.

Ming

[hertneky]

Hi Ming,

It sounds like you are comparing MPAS UGRID to GDAS using PointStat, but the log shows no matched pairs and empty output files. I think this may be related to this bug, where the forecast lead is incorrectly parsed from the input file. As a work-around, can you try adding 'set_attr_lead', such that it gets the lead correctly. See this section of the documentation.

[smileMchen]

Hi Tracy,

Thank you for getting back to me quickly. I will add "set_attr_lead" and keep you updated about the result.

Just one quick question: MPAS history file includes 3 time variables:

inital_time: Model initialization time
xtime: Model valid time
Time: seconds since initialization time (i.e., forecast lead time).

Which of the above three time variables are used?

Again, thank you for your help!

Best regards,
Ming

[smileMchen]

Hi Tracy,
I am sorry to tell that point_stat still didn't work after adding set_attr_lead. I will continue to work on this issue. meanwhile, please let me know if you have any suggestions. Thank you.
Ming

[hertneky]

Ok, so adding set_attr_lead does not produce any matched pairs. So there must be something else wrong here, either with the configuration or the input data.

In case you have not seen it, here are the metadata requirements for UGRIDs. Can you check to make sure your file has all the requirements?

And here is an example unit-test config for UGRD PointStat.

Please see the limitations for unstructured grids here.

If you are still having issues, feel free to send sample data and config file to us to investigate further.

[smileMchen]

Hi Tracy,
I have uploaded my MPAS data and config file (ming-met-issue.tar) to Google drive. Please take a look and let me know if you need more information. Thank you !
Ming

[hertneky]

Thank you for sending the data. Unfortunately, the historyRegCF.2017-02-04_00.00.00.nc file is just a link to a location on derecho that I do not have access to.

ls: cannot open directory '/glade/u/home/chenming/MyDoc/': Permission denied

Can you move it somewhere where I can access it? The other files are fine.

[smileMchen]

Hi Tracy,

I aplogize for the permission issue. Thank you for looking into my issue.

I have the data saved at /glade/derecho/scratch/chenming/METplus/Test/historyCF.2017-02-04_00.00.00.nc. This is a regional MPAS output file and I added 'Time' to make it CF-compliant.

Please let me know if you still cannot access it. Thank you.

Ming

[hertneky]

Hi Ming - Thank you for providing access to that file. I looked at the file and did not see any issues with your mpas data. Next I ran PointStat on your data and recreate the issue you see - that there are no matched pairs. From the log, I see that all observations were rejected:

DEBUG 2: Number of matched pairs   = 0
DEBUG 2: Observations processed    = 176493
DEBUG 2: Rejected: station id      = 0
DEBUG 2: Rejected: obs var name    = 145224
DEBUG 2: Rejected: valid time      = 0
DEBUG 2: Rejected: bad obs value   = 0
DEBUG 2: Rejected: off the grid    = 0
DEBUG 2: Rejected: topography      = 0
DEBUG 2: Rejected: level mismatch  = 31269

While the variable TMP exists in your processed gdas file, none of them are at the level that you requested: "L0". I also checked "Z2" and there were no matches at that level as well. I would therefore check your PB2NC processing to ensure you are saving the observations you want. You may use the level_category to specify level types you want to save, or leave it blank to process them all. Please see the level_category options under the PB2NC config file options.

Please let me know if this solves the issue.