Couldn't ingest precipitation from grib2 model file. #1138
-
|
Hello, I try to run GridStat using 1h precipitation accumulated from the initial time from a grib2 model file with the following parameters: MET doesn't find any precipitation records (while other records are read correctly, temperature, for example): What could be the problem? I saw that the precipitation record from a MET example file has the similar parameters: I attach my files in zip archive here. |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
|
Hi Anastasia, thanks for sending along some sample data to demonstrate the issue. That made it really easy to debug. I see that you'd like to run Grid-Stat to evaluate precipitation. When working with GRIB2 files, I use the "wgrib2" utility to dump an inventory of their contents. For example, here's what I see in one of the data files you sent: Note that there is no accumulated precipitation, named "APCP", present in that file. And that's why you're seeing this warning: However, TPRATE, for total precipitation rate, is present: But that naming convention (TPRATE) is not obvious in the output of the "grib_ls" command you ran, and I suspect that's the source of the confusion. Whenever getting going with new datasets, I always recommend running MET's plot_data_plane tool: Which creates this plot: Note that I had to specify "file_type=GRIB2;" since the input file uses the ".grb" suffix which MET interprets as being a GRIB version 1 file. If you renamed it as ".grb2" then you would no longer need that "file_type" option. MET interprets the ".grb2" suffix as meaning its a GRIB2 file. So you'll need to change "APCP" to "TPRATE" in your Grid-Stat config file. However, if you run Grid-Stat to compare ilfr_00010000s.cfoeps.grb to ilfr_00020000s.cfoeps.grb you'll get a different error message. Because while the first contains the total precip rate accumulated over 1 hour (A1), the second contains it accumulated over 2 hours (A2). I'm guessing that these are just two output timesteps from some model run. So both are forecasts, and I doubt you actually want to compare them directly with Grid-Stat. If you did want to do that for some reason, you'd need to specify the "fcst" and "obs" dictionaries separately in the config file. One would be TPRATE/A1 and the other would be TPRATE/A2. I'd recommend looking closely at your verifying observation/analysis data. Think about what accumulation interval is present there and adjust your forecast so that your can make a fair comparison. You may find MET's pcp_combine tool useful here. For example, if you want the accumulation between hours 1 and 2, you could run a subtract command like this: Also note that you can use the "convert(x)" function to convert your hourly rate to a total accumulation. Running plot_data_plane on your 2-hour accumulation is a good example of this:
Hope that helps clarify. |
Beta Was this translation helpful? Give feedback.
-
|
Dear John, |
Beta Was this translation helpful? Give feedback.
Hi Anastasia, thanks for sending along some sample data to demonstrate the issue. That made it really easy to debug.
I see that you'd like to run Grid-Stat to evaluate precipitation. When working with GRIB2 files, I use the "wgrib2" utility to dump an inventory of their contents. For example, here's what I see in one of the data files you sent: