Using MET to calculate 500mb ACC scores #1113
-
|
Good Day, I have for years used MET to calculate 500 mb Height ACC scores from gridded model output. However, recently I upgraded from version 8.1.2 to 9.1.3 and ran into an issue. I have previously used grib1 versions of: I have attached by GridStatConfig file and also pasted below my output. Any help would be appreciated. Thanks. This is the output I get using 9.1.3: |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 1 reply
-
|
Hello, and thank you for your question. I took a look at the grib2 file you attached and confirmed that the dates of the grib2 file match your output. From the output and error provided, it would seem like one of the entries in the grib2 file, most likely 19840925 for the HGT/P500 entry, has a valid date being read by MET as a bad date, so it's defaulting to 19700101. I'm going to include @JohnHalleyGotway on this issue as he should be able to provide better insight on what's causing the |
Beta Was this translation helpful? Give feedback.
-
|
@awhuffma, thanks for sending the sample climatology data file. I took a look at the records in there and do NOT see any issues there. To summarize, the z500.cfsr.mean.clim.daily.1982.2010.grb2 file contains 1464 GRIB2 records, all for 500mb height. The records are spaced 6 hours apart during 1984 (4 per day for 366 days including 2/29). You're running Grid-Stat to compare gfs.t00z.pgrb.0p50.f000 to 2021082500gdas.grib. However, as the log message indicates, you're using an old configuration file version: When the major version number of the code (i.e. 9) differs from the major number listed in the config file (i.e. 8), we print that error message to alert users that the configuration options may have changed. And between MET versions 8.1.2 and 9.1.3, the way you configure information about climatology data changed. I see that you simply changed the version number listed in the config file without actually changing the contents. In general, I'd recommend diffing your v8.1.2 configuration file against the default Grid-Stat config file for v9.1.3. It should be pretty easy to see what's changed and then you can migrate the specifics of your v8.1.2 configuration into a file for v9.1.3. Comparing the default Grid-Stat config file between versions 8.1.2 and 9.1.3, you'll find that inside the "climo_mean" dictionary: You can read more about the climo_mean settings in this section of the MET User's Guide. But for this data, that is daily and spaced 6-hours apart, you should use: Do note that there are other changes between the Grid-Stat config file between these versions. So I'd recommend looking more closely at those. I'm not sure this will fix the error you're getting, but we should definitely update the config file before debugging any further. Please let me know if/how the behavior changes after you update the config file. I do realize this is a hassle and apologize for that. Whenever possible, we do make config file changes backward compatible. But in this case, doing so confused the logic of the code considerably. |
Beta Was this translation helpful? Give feedback.
-
|
John, Thanks so much for the reply. This did fix my issue after I updated my config file. Should have thought of that myself. Thanks again for the help! Allan |
Beta Was this translation helpful? Give feedback.
@awhuffma, thanks for sending the sample climatology data file. I took a look at the records in there and do NOT see any issues there. To summarize, the z500.cfsr.mean.clim.daily.1982.2010.grb2 file contains 1464 GRIB2 records, all for 500mb height. The records are spaced 6 hours apart during 1984 (4 per day for 366 days including 2/29).
You're running Grid-Stat to compare gfs.t00z.pgrb.0p50.f000 to 2021082500gdas.grib. However, as the log message indicates, you're using an old configuration file version:
When the major version number of th…