Wrf forward operators #708
Conversation
Fixes Q2 to SH2 conversion bug and also improves code comments of when operating on Q2(vapor mixing ratio) and SH2 (specific humidity)
Updating 2M temp obs definition such that it horizontally interpolates on WRF surface output (T2) instead of interpolating on bottom level of WRF 3D temperature (T)
Adding docs to FO bug fixes and also providing description of linkage between observation types and required WRF field outputs
braczka
added
Documentation
…F_forward_operators
| The number of WRF output fields required to support an observation type can vary. For | ||
| observation types where there is a direct analog to a WRF output field, the forward | ||
| operator consists of only spatial interpolation, thus requires only a single output | ||
| variable (e.g. METAR_TEMPERATURE_2_METER). For observation types that require multiple |
There was a problem hiding this comment.
I think it is worth being clear here about model_interpolate vs forward operators.
models/wrf/tutorial/README.rst
Outdated
|
|
||
|
|
||
| A critical piece of observation metadata includes the observation type | ||
| (``METAR_TEMPERATURE_2_METER``) which is linked to the observation quantity |
There was a problem hiding this comment.
quantity rather than observation quantity:
| (``METAR_TEMPERATURE_2_METER``) which is linked to the observation quantity | |
| (``METAR_TEMPERATURE_2_METER``) which is linked to the quantity |
models/wrf/readme.rst
Outdated
| - The WRF ``model_mod.f90`` file reads WRF netCDF files directly to acquire the model state | ||
| data. Earlier versions required ``wrf_to_dart`` and ``dart_to_wrf`` programs which | ||
| are no longer necessary. |
There was a problem hiding this comment.
It is a bit weird to talk about the .f90 file reading anything.
| - The WRF ``model_mod.f90`` file reads WRF netCDF files directly to acquire the model state | |
| data. Earlier versions required ``wrf_to_dart`` and ``dart_to_wrf`` programs which | |
| are no longer necessary. | |
| - The WRF ``model_mod`` reads a WRF netCDF file for each WRF domain to acquire the model state | |
| meta data, static data such as surface elevation. |
It would be good to be clear hear that model_mod does not read state, only the meta data (variable, size, dimensions). Model_mod does read in static data.
I think it is safe to remove the note about old versions.
There was a problem hiding this comment.
This looks good on my end. There are 2 changes that could lead to changes in results within the WRF model_mod: 1) the corrected conversion from vapor mixing ratio to specific humidity and 2) when a temperature observation is defined as a 'surface' obs it will interpolate on the T2 WRF variable as default, instead of interpolating on bottom level of THM/T. Agreed this PR description does not do a great job of describing the bug fixes-- which were found during the documentation process.
The obs_diag.csh fix, in hindsight, is unrelated -- but is a bug that needed to be fixed while running the WRF tutorial -- which is the test case for these changes. I was able to implement the WRF tutorial successfully. Also included a fix with the WRF driver.csh code which likes to resubmit assim_advance.csh jobs spuriously. Seems to be related with transition to Derecho and I previously tried to address it with sleep statements which was suboptimal and didn't always work. I turned off the resubmission capability and replaced with a warning instead.
@hkershaw-brown I think I addressed all your comments so far -- if the science changing fixes need more documentation I can provide that. It's tough to detect physical differences in the tutorial result based on the surface temperature updates because of the brevity of the tutorial (2 assimilation cycles). If I should reorganize the PR description or organization, let me know......
driver.csh has ability to check for assim_advance jobs that fail, and resubmits. Switch to derecho has led to false detections, which leads to errors. Removed resubmission and replaced with warning
Alternative to modifying METART obs_def
|
@braczka sterling work on this Brett! I'll take a look next week. Have a great weekend |
|
Thank you --- forgot to mention too this PR will not address the WRF boundary perturbation documentation. I am still sorting through that and it is not important for functionality. Because this PR includes bug fixes which affect science results this should have more urgency and be pushed through. I will likely address WRF boundary perturbation documentation more fully when designing a nested WRF-DART tutorial, coming soon. |
Description:
Adds bug fixes to implementation of surface temperature and specific humidity forward operator calculations. Adds documentation for these fixes, and also provides more description of the forward operator process in general and linkage between the observations and the required WRF field outputs. Documentation added to both WRF Tutorial and WRF model
page.
Also addresses minor diagnostic script bug for WRF tutorial.
Fixes issue
Addresses bug #700, bug #672. Loosely related to #404
Types of changes
Documentation changes needed?
Tests
Ran WRF-Tutorial to verify changes work.
Checklist for merging
Checklist for release
Testing Datasets