USDA-ARS-NWRC / smrf

SMRF was designed to increase the flexibility of taking measured weather data, or atmospheric models, and distributing the data across a watershed.
Other
12 stars 4 forks source link

Docker - Install weather_forecast_retrieval via pip install. #147

Closed jomey closed 4 years ago

jomey commented 4 years ago

The weather_forecast_retrieval (WFR) is a pip install package now and simplifies the Docker image. Updates to WFR also requires a rebuild of the gold images.

Fixes #146

Changes to Gold files

Only the HRRR gold files were affected by this change. Here Histograms for each file that showed a change: gold_hrrr_hist_air_temp nc gold_hrrr_hist_net_solar nc gold_hrrr_hist_precip_temp nc gold_hrrr_hist_thermal nc gold_hrrr_hist_vapor_pressure nc gold_hrrr_hist_wind_direction nc gold_hrrr_hist_wind_speed nc

jomey commented 4 years ago

Note that there are some warnings now, when running the test suite triggered from the current version of WFR:

/usr/local/lib/python3.6/dist-packages/weather_forecast_retrieval/hrrr.py:716: FutureWarning: The pandas.np module is deprecated and will be removed from pandas in a future version. Import numpy directly instead
1914  data = data.assign_coords(x=pd.np.arange(0, len(data['x'])) * 3000)

Those will be addressed once a new version with the latest from the master branch on WFR is cut.

scotthavens commented 4 years ago

We have been trying to document changes to the gold files so we have explicit understanding . For example, PR #125 showed the histogram for net_solar since that was the only variable that was changing when the gold files were updated. Can you add something similar so we all have an idea of how much the gold files changed?

scotthavens commented 4 years ago

You can use nc_stats that will print out a summary difference for two files. It's in vhdr at the moment.

jomey commented 4 years ago

There is some documentation of the difference between the two different WFR version with issue USDA-ARS-NWRC/weather_forecast_retrieval#9. Should I plot another histogram for the gold files explicitly? Any standard code checked in to ensure similar standards?

scotthavens commented 4 years ago

There is some documentation of the difference between the two different WFR version with issue #9. Should I plot another histogram for the gold files explicitly? Any standard code checked in to ensure similar standards?

Check out the comment above. In the long run this nc_stat will be ran when making the gold files so it's automatic.

scotthavens commented 4 years ago

The test outputs are compared to the gold files so that we know if anything changed in the way we model. For example, if we make some code improvements to solar radiation, it the tests error because they don't match the gold file, we'll know that the code improvement broke SMRF.