Problem getting Point-Stat to produce matched pairs for ADPUPA point obs #2498
-
I am having problems getting Point-Stat to produce matched pairs for ADPUPA point observations. My observations are as follows: The UPP post-processed GRIB WRF forecast file contains the following variables to match with the above observations: The ADPSFC variables for surface verification are: The Point-Stat MPR file only has the ADPSFC pairs, but no ADPUPA pairs. I'm trying to figure out why no ADPUPA pairs. I will provide my Point-Stat config file (PointStatConfig_v11.1.0_WRF_LES_twr10_obs_test11.txt), the forecast file (WRFPRS_d04.09_00), the Point-Stat log file (WRF_twr10_test11) and the NetCDF point observation file (d04.2022-07-05_12:00:00.0000_09FEB2024_test9.nc), and the MPR file (point_stat_WRF_twr10_test11_090000L_20220705_120000V_mpr.txt) to Julie Prestopnik via the Army file server (SAFE). This test was conducted to see if the UPP GRIB forecast file containing the above output can be used to verify ARL tower data using surface verification and vertical level verification. This relates to GitHub issue:#2487 Thanks. |
Beta Was this translation helpful? Give feedback.
Replies: 4 comments 12 replies
-
Hi @jwraby if you're using Python embedding, can you also share your Python script? Thanks! |
Beta Was this translation helpful? Give feedback.
-
@jwraby I see you have some questions about how Point-Stat constructs matched pairs. You are seeing MPR output when verifying forecasts of 2-meter relative humidity, 2-meter temperature, and 10-meter winds against observations with message type ADPSFC. However, you're not seeing MPR output when verifying 2-meter winds or 10-meter temperature against observations with message type ADPUPA. Hopefully I can provide some information to help clarify what's going on. First, the Point-Stat tool provides options for handling the "level" information when matching forecast data to point observations. And those options are what need to be clarified here. Apologies, if I restate anything already obvious to you. The Point-Stat configuration file includes separate dictionaries to specify information about the gridded Each Here's the usage statement from ascii2nc which describes the 11-column point observation format:
In the sample data you sent, the 9th column (i.e. But wait, if Point-Stat only uses point observations that match the requested levels, why are you getting MPR output for the ADPSFC message type? The answer is the message_type_group_map configuration option. By default, that setting includes this entry:
And those three message types are assumed to indicate that these are point observations at the SURFACE. To make surface verification easier, Point-Stat includes a rule that when verifying a vertical-level-type forecast (i.e. If you were to add ADPUPA to the list of surface message types, then you'd also see matched pairs for ADPUPA in the output. But I would not recommend that. ADPUPA is the PREPBUFR mnemonic indicating upper-air observations. Generally, they contain sounding data from weather balloons... and that's definitely not surface data. It sounds like you'd like to know if/how you can verify 2-meter winds and 10-meter temperature against ADPUPA point observations. I'm guessing the real problem here is how the Height of those observations is defined. The forecast 2-meter wind and 10-meter temperature are defined as meters above-ground-level (i.e. AGL). I'm guessing the point observations are defined as meters above-mean-sea-level (i.e. MSL). And MET does not currently include any direct conversion between MSL and AGL for point observations. I certainly could see it could be enhanced to do so, using topography data provided by the Big picture, I haven't really encountered folks verifying 2-meter winds and 10-meter temperature against ADPUPA point observations in the past. Instead, they verify forecast at pressure levels against ADPUPA point observations. For example, people often verify an 850 mb temperature forecast ( Hope that helps clarify. What other questions do you have? |
Beta Was this translation helpful? Give feedback.
-
@DanielAdriaansen. Thanks for checking. I may have contributed to the confusion by creating a new discussion, but if the problem can be resolved, then it paves the way for a better way to handle the tower data for the use case. |
Beta Was this translation helpful? Give feedback.
-
Thank you, Julie.
|
Beta Was this translation helpful? Give feedback.
@jwraby I see you have some questions about how Point-Stat constructs matched pairs. You are seeing MPR output when verifying forecasts of 2-meter relative humidity, 2-meter temperature, and 10-meter winds against observations with message type ADPSFC. However, you're not seeing MPR output when verifying 2-meter winds or 10-meter temperature against observations with message type ADPUPA. Hopefully I can provide some information to help clarify what's going on.
First, the Point-Stat tool provides options for handling the "level" information when matching forecast data to point observations. And those options are what need to be clarified here. Apologies, if I restate anything already obvio…