identify areas for testing #27
Assignees
Comments
|
I need to find out if it even works for all or most WRF files when actually
called. It is only implemented in production for the UFS and its
derivative products which do not do ANYTHING with WRF but it
is still needed because UPP has its WRF hooks. It may not even work for
the UPP forks prior to 2018 that process several generations of WRF files.
(UPP now properly branches and tags but the old forks remain)
It will take some time to run this down but note we need to clean this out
or at least document where we stand with WRF so that's time well spent and
I'll start working on it. An example of
such a fork is what was submitted to process our HRRR forecast files, short
range rapid update forecast cycle.
The test for successful UFS workflow workflow processing is successful
BUILD of UPP.
Yeah we keep sticking our finger into our software haystack and coming out
stiff with needles. However with better software management since 2017 and
consolidation of our dynamic core around UFS-weather model
with the FV3 dynamic core, the situation is much improved from four years
ago.
Hi\uiya and Wen. Do you know anything about compatibility of post versions
that process HRRR and WRF-ARW forecasts with the current UPP repository.
Are there tags to support WRF and HRRR or
did these split off as forks (disk copies to NCO) when the several WRF
systems were implemented
Jacob, Geoff, and Eric? Where do the post processors for HRRR and WRF-ARW
or NAM, come from?
…On Fri, Oct 30, 2020 at 6:35 AM Edward Hartnett ***@***.***> wrote:
We need to identify some tests that we are going to write.
I guess the best thing to start with is a WRF data file, preferably a
small one. Then we can build tests around that. @GeorgeVandenberghe-NOAA
<https://github.com/GeorgeVandenberghe-NOAA> can you come up with a test
file?
It should be less than 50 MB in size...
—
You are receiving this because you were assigned.
Reply to this email directly, view it on GitHub
<#27>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ANDS4FW3GUIMTP6LOIGVMR3SNKJFZANCNFSM4TE2A26Q>
.
--
George W Vandenberghe
*IMSG* at NOAA/NWS/NCEP/EMC
5830 University Research Ct., Rm. 2141
College Park, MD 20740
[email protected]
301-683-3769(work) 3017751547(cell)
|
|
George, I believe DTC runs their UPP regression tests which include post
processing WRF output.
Copying Kate in this email.
Huiya
On Fri, Oct 30, 2020 at 7:58 AM George Vandenberghe - NOAA Affiliate <
[email protected]> wrote:
… I need to find out if it even works for all or most WRF files when
actually called. It is only implemented in production for the UFS and its
derivative products which do not do ANYTHING with WRF but it
is still needed because UPP has its WRF hooks. It may not even work for
the UPP forks prior to 2018 that process several generations of WRF files.
(UPP now properly branches and tags but the old forks remain)
It will take some time to run this down but note we need to clean this out
or at least document where we stand with WRF so that's time well spent and
I'll start working on it. An example of
such a fork is what was submitted to process our HRRR forecast files,
short range rapid update forecast cycle.
The test for successful UFS workflow workflow processing is successful
BUILD of UPP.
Yeah we keep sticking our finger into our software haystack and coming out
stiff with needles. However with better software management since 2017 and
consolidation of our dynamic core around UFS-weather model
with the FV3 dynamic core, the situation is much improved from four years
ago.
Hi\uiya and Wen. Do you know anything about compatibility of post
versions that process HRRR and WRF-ARW forecasts with the current UPP
repository. Are there tags to support WRF and HRRR or
did these split off as forks (disk copies to NCO) when the several WRF
systems were implemented
Jacob, Geoff, and Eric? Where do the post processors for HRRR and
WRF-ARW or NAM, come from?
On Fri, Oct 30, 2020 at 6:35 AM Edward Hartnett ***@***.***>
wrote:
> We need to identify some tests that we are going to write.
>
> I guess the best thing to start with is a WRF data file, preferably a
> small one. Then we can build tests around that. @GeorgeVandenberghe-NOAA
> <https://github.com/GeorgeVandenberghe-NOAA> can you come up with a test
> file?
>
> It should be less than 50 MB in size...
>
> —
> You are receiving this because you were assigned.
> Reply to this email directly, view it on GitHub
> <#27>, or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/ANDS4FW3GUIMTP6LOIGVMR3SNKJFZANCNFSM4TE2A26Q>
> .
>
--
George W Vandenberghe
*IMSG* at NOAA/NWS/NCEP/EMC
5830 University Research Ct., Rm. 2141
College Park, MD 20740
***@***.***
301-683-3769(work) 3017751547(cell)
--
--------------------------------------------------------------
Physical Scientist
Environmental Modeling Center, NCEP, NWS, NOAA
301-683-3716
Office 2069
5830 University Research Court
College Park. MD 20740
|
|
This wrf_io lib is derived by the DTC UPP team from the WRF package for reading wrf io format data e.g. hrrr, wrf in UPP . I invite |
|
I need to come up with a unit test for the wrf-io library as a part of the
much larger requirement to (finally) implement unt tests for all of
NCEPLIBS. I have a plan for sigio and sfcio but am not sure how
compatible our several wrf versions are.
On a side note, I have used UPP builds to assess the sanity of an NCEPLIBS
distribution since late calendar 2015 because UPP uses a lot of those
libraries. In late calendar 2015 I started a project under the guise
of "general NCEP optimization support", to get NCEPLIBS and third party
libraries working on all NOAA platforms and looking like NCO's
implementation, at one point describing this to Arun as "NCEP in a can".
Without chest beating I'll just say that's been very successful and a lot
of other work has built on it.
…On Fri, Oct 30, 2020 at 8:25 AM WenMeng-NOAA ***@***.***> wrote:
This wrf_io lib is derived by the DTC UPP team from the WRF package for
reading wrf io format data e.g. hrrr, wrf in UPP . I invite
@fossell <https://github.com/fossell> in this thread. She might provide
more information.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#27 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ANDS4FVPGJ73RS7XWVAEO5LSNKWE3ANCNFSM4TE2A26Q>
.
--
George W Vandenberghe
*IMSG* at NOAA/NWS/NCEP/EMC
5830 University Research Ct., Rm. 2141
College Park, MD 20740
[email protected]
301-683-3769(work) 3017751547(cell)
|
|
Thanks. That is good. I will correct my earlier emails to the NCEPLIBS
team
This puts the onus back on me to develop a unit test suite for it.
(probably for sigio and sfcio also) I have suggested gfsio not be unit
tested
since no one at all executes it but we have to keep the library API because
GFS spectral based codes are full of hooks using
it even though those are never executed. It still needs to be build
tested which means things that have it's API have to build.
I did examine UPP for possible removal of gfsio. It would take modest
effort (about an hour, I did it with an old fork) but at the end of this we
would still have it in GFS based ensemble
codes and probably other stuff I don't know about and cleaning out or even
finding all of those would take a long time
On Fri, Oct 30, 2020 at 10:18 AM Wen Meng - NOAA Affiliate <
[email protected]> wrote:
… The UPP uses wrf-io library for all implementations, e.g. GFS V16, GEFS
V12, HRRR4, RAP5.
Wen
On Fri, Oct 30, 2020 at 10:12 AM George Vandenberghe - NOAA Affiliate <
***@***.***> wrote:
> Will UPP use the wrf-io library for all implementations going forward or
> do some implementations still
> need the full WRF Vx.y built alongside and prior?
>
> On Fri, Oct 30, 2020 at 10:01 AM Wen Meng - NOAA Affiliate <
> ***@***.***> wrote:
>
>> George,
>>
>> The UPP for hrrr v5 and rap v4 were implemented from a UPP repository
>> belonging to GSL on VLAB. Last Summer, we completed GSL UPP
>> repos. merged to our EMC UPP repos. on github. In the future GSL will
>> use EMC UPP repos. for development.
>>
>> Wen
>>
>> On Fri, Oct 30, 2020 at 9:12 AM George Vandenberghe - NOAA Affiliate <
>> ***@***.***> wrote:
>>
>>> Am I correct that UPP for HRRR and WRF-ARW was implemented in
>>> production as copied snapshots or was it implemented from the UPP branch in
>>> our svn, vlab, or git repositories? I am referring to HRRR implemented in
>>> 2017 or 2018, not the earlier one in 2012. I don't know even that much
>>> about WRF-ARW.
>>>
>>>
>>>
>>>
>>>
>>> --
>>>
>>> George W Vandenberghe
>>>
>>> *IMSG* at NOAA/NWS/NCEP/EMC
>>>
>>> 5830 University Research Ct., Rm. 2141
>>>
>>> College Park, MD 20740
>>>
>>> ***@***.***
>>>
>>> 301-683-3769(work) 3017751547(cell)
>>>
>>>
>>
>> --
>> Wen Meng
>> IMSG at NOAA/NWS/NCEP/EMC
>> 5830 University Research Ct., Room 2070
>> College Park, MD 20740
>> ***@***.***
>> 301-683-3779
>>
>
>
> --
>
> George W Vandenberghe
>
> *IMSG* at NOAA/NWS/NCEP/EMC
>
> 5830 University Research Ct., Rm. 2141
>
> College Park, MD 20740
>
> ***@***.***
>
> 301-683-3769(work) 3017751547(cell)
>
>
--
Wen Meng
IMSG at NOAA/NWS/NCEP/EMC
5830 University Research Ct., Room 2070
College Park, MD 20740
***@***.***
301-683-3779
--
George W Vandenberghe
*IMSG* at NOAA/NWS/NCEP/EMC
5830 University Research Ct., Rm. 2141
College Park, MD 20740
[email protected]
301-683-3769(work) 3017751547(cell)
|
|
@GeorgeVandenberghe-NOAA we can develop the test suites as a team. ;-) You are not on your own. So for wrf_io, the first step is to find one of the kind of WRF files we are expected to be able to read, and then we can build some tests around that. sigio and sfcio we will address in their projects, but we will start in the same place for their read code: we need a test file that we can start with. ;-) |
|
Our UPP regression tests include hrrr test which use hrrr model output (WRF file). |
|
@WenMeng-NOAA can you attach a copy of that WRF file to this issue? |
|
@edwardhartnett That wrf file is too large to attached in this issue. If you have Hera or WCOSS account, I will point you the location. |
|
Yes, please do, on hera. |
|
Please see /scratch1/NCEPDEV/stmp2/Wen.Meng/edward for hrrr model output on Hera. |
|
I'm a little late to this thread. The best course of action is what has already been initiated, using HRRR model output from EMC regression tests. While DTC does have some WRF data for testing, we are phasing out support for WRF models so this is no longer part of our work plan. |
|
@fossell the goal is to also have independent testing of the wrf_io library within this repo. This does not affect testing of the UPP, which is a separate effort. The UPP uses many of the NCEPLIBS libraries. We have a refactor underway to add testing and documentation to many of them. This issue is part of that effort. The goal is to ensure that we're putting out a wrf_io library that we know works. Like all the NCEPLIBS libraries, it can and will be further tested at the UPP level, and that's great. Well the file in question is >10 GB, so that doesn't work very well for testing. Are there any WRF files that we need to read which are smaller? |
|
@GeorgeVandenberghe-NOAA will attempt to do this in three steps: THis will be started sometime in the next couple of weeks, we hope, after WCOSS2 problems have been sorted out. |
|
It's WCOSS2 I am dealing with.
…On Tue, Nov 10, 2020 at 3:55 PM Edward Hartnett ***@***.***> wrote:
@GeorgeVandenberghe-NOAA <https://github.com/GeorgeVandenberghe-NOAA>
will attempt to do this in three steps:
1 - synthesize a file using the write functions in the library, then read
it.
2 - Take some use cases from UPP and make them into tests within wrf_io
repo.
3 - Collect some test files (which will be large) and put them on FTP
site, to be downloaded by script and used it testing.
THis will be started sometime in the next couple of weeks, we hope, after
WCOSS problems have been sorted out.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#27 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ANDS4FVNJM6T6T3ISDUUOXLSPGSCDANCNFSM4TE2A26Q>
.
--
George W Vandenberghe
*IMSG* at NOAA/NWS/NCEP/EMC
5830 University Research Ct., Rm. 2141
College Park, MD 20740
[email protected]
301-683-3769(work) 3017751547(cell)
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
We need to identify some tests that we are going to write.
I guess the best thing to start with is a WRF data file, preferably a small one. Then we can build tests around that. @GeorgeVandenberghe-NOAA can you come up with a test file?
It should be less than 50 MB in size...
The text was updated successfully, but these errors were encountered: