search

USDA-ARS-NWRC / awsm

Automated Water Supply Model (AWSM) was developed at the USDA Agricultural Research Service. AWSM was designed to streamline the workflow used to forecast the water supply of multiple water basins.
Other
9 stars 2 forks source link

Update smrf cfg #57 #58

Closed micahjohnson150 closed 4 years ago

micahjohnson150 commented 4 years ago

The new cfg updates in smrf had implications fro awsm and a few references awsm make to it. This PR is updating AWSM to reflect changes that occurred in SMRF #98

Most of the changes in the AWSM code are related to item name changes in the SMRF gridded dataset section and the changes in the tests are for updating the config files.

Currently I have not had to change the test netcdfs. I DID have to update the gold_config.ini to reflect the same changes AND I confirmed that it still produces a matching dataset.

micahjohnson150 commented 4 years ago

@scotthavens this is pending SMRF PR #133

micahjohnson150 commented 4 years ago

Ok I did a complete documented check on all the variables for the em.nc and snow.nc files. Here is what I did:

  1. awsm gold_config.ini
  2. ncdiff output_file gold_file diff.nc for both snow and em files
  3. nc_stats diff.nc for all variables in each file

Attached is the output from this. TL;DR Zeros across the board! gold_diff_report.txt

micahjohnson150 commented 4 years ago

So I think what is happening is that snowav is uninstalling spatialnc 0.12 for 0.10 which still has older config options. @robertson-mark is going to remove the ipw code that uses spatialnc from snowav so that this chunk of code can pass. I am awaiting a new version of snowav that is sans spatialnc or with a newer requirement.

robertson-mark commented 4 years ago

snowav==0.11.5 no longer requires spatialnc (and doesn't process ipw files)

scotthavens commented 4 years ago

I was able to build the docker image locally but Travis is having problems with gcc in pysnobal.

robertson-mark commented 4 years ago

Everything looks good to me, but I held off approval because of the failing build.

micahjohnson150 commented 4 years ago

@scotthavens I got it all wrapped up. A few issues were resolved. The conclusion was it was never meant to pass while threading which got changed early on. So there is still some kind of configuration issue relating to running less than 1 day, with threading and make_nc=False (I think). So we will need to investigate more. I was never able to replicate it locally unfortunately even after scorching docker and virtualenvs. The good news is though that this test is now covering what it was originally meant to test.

Anyways if you get me the thumbs up here (since its been a fair amount of time since your approval) I will merge it in and squash all my unflattering desperate commits....

micahjohnson150 commented 4 years ago

I will hold off on the test to avoid failures. Made an issue and reference this pull request. CLOSING AND MERGING! WAHOO!