search

E3SM-Project / E3SM

Energy Exascale Earth System Model source code. NOTE: use "maint" branches for your work. Head of master is not validated.
https://docs.e3sm.org/E3SM
Other
334 stars 334 forks source link

Add ERA5 (hourly and 6hrly) datm forcing option #6234

Closed chloewhicker closed 2 weeks ago

chloewhicker commented 3 months ago

[BFB]

rljacob commented 3 months ago

How large is this new forcing data set? Is it available on the blues data server?

bishtgautam commented 3 months ago

@chloewhicker can confirm the following finding. There is an hourly and six-hourly version of ERA5 on Perlmutter

  1. Hourly data /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.0.25d.v5.c180614

    1.5T    /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.0.25d.v5.c180614/lwdn
1.5T    /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.0.25d.v5.c180614/pbot
3.0T    /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.0.25d.v5.c180614/prec
5.9T    /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.0.25d.v5.c180614/swdn
1.5T    /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.0.25d.v5.c180614/tbot
1.5T    /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.0.25d.v5.c180614/tdew
5.9T    /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.0.25d.v5.c180614/wind

  2. Six-hourly data /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.6HRLY.0.25d.v5.c180614

    du -sh /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.6HRLY.0.25d.v5.c180614/*
100M    /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.6HRLY.0.25d.v5.c180614/domain.lnd.era5_721x1440_rdrlat_EC30to60E2r2.221115.nc
253G    /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.6HRLY.0.25d.v5.c180614/lwdn
253G    /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.6HRLY.0.25d.v5.c180614/pbot
506G    /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.6HRLY.0.25d.v5.c180614/prec
1007G   /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.6HRLY.0.25d.v5.c180614/swdn
253G    /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.6HRLY.0.25d.v5.c180614/tbot
253G    /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.6HRLY.0.25d.v5.c180614/tdew
1007G   /global/cfs/cdirs/e3sm/inputdata/atm/datm7/atm_forcing.datm7.ERA.6HRLY.0.25d.v5.c180614/wind

We should at least do the following:

  1. Copy 1-yr of hourly and six-hourly data on each supported machine and e3sm inputdata server, and
  2. Add two new tests that use hourly and six-hourly ERA5 dataset.
chloewhicker commented 3 months ago

Thanks Gautam. The datm forcing files are in the cfs on Perlmutter but have not been uploaded to blues. If the hourly data is too large we can move forward with just the 6 hourly option. I ran some tests and the the 6 hourly resolution only results in minor discrepancies compared to the hourly forcing.

rljacob commented 2 months ago

@bishtgautam is your table above the full data set or just one year? @chloewhicker please add the tests that @bishtgautam mentioned.

bishtgautam commented 2 months ago

@rljacob The table listing the size of the data is for the whole data set.

@chloewhicker In case you aren't familiar with adding tests, I can add a test in your branch.

rljacob commented 2 months ago

Then go ahead and add a year's worth of data to the input data server.

bishtgautam commented 2 months ago

Copying @chloewhicker's original first message in the PR:

We've implimented ERA5 as a new datm forcing option for ELM simulations. This PR includes both the hourly ERA5 forcing (ELMERA5) and the 6 hourly forcing (ERA5_6HR) from 1980-2021. The 6 hourly forcing is significantly faster than the hourly.

More information about the implementation process can be found here: https://acme-climate.atlassian.net/wiki/spaces/PSC/pages/3709730833/ERA5+Offline+Forcing+Implementation+and+Evaluation

and information about a 1k IG spin up simulation using the 1980s ERA5 datm can be found here: https://acme-climate.atlassian.net/wiki/spaces/FAN/pages/4096557191/Deep+Firn+1k+Spin+Up

These modifcations only change the output if one selects the ERA5 option in env_run.xml.

These modifications are complete and have been shown to work properly within E3SM (see simulation information here https://acme-climate.atlassian.net/wiki/spaces/PSC/pages/3709730833/ERA5+Offline+Forcing+Implementation+and+Evaluation and here:https://acme-climate.atlassian.net/wiki/spaces/FAN/pages/4096557191/Deep+Firn+1k+Spin+Up ).

The relevent code modifications can be found here: https://github.com/chloewhicker/E3SM/tree/cwhicker_m20240206_2_datm-era5

The implimentation of an ERA5 forced data atmosphere will improve SMB estimates in ELM as ERA5 precipitation has been shown to be more highly correlated with with in situ measurements of net accumulation than other widely used reanalyses (WFDE5, CRUNCEP, and GSWP3) (Schneider et al., 2023). The ERA5 data forcing option includes 0.25 degree, hourly, and 6 hourly resolution data from 1980-2021, where as other currenly avaiable datm forcing options only extend to 2005 (CLM_QIAN, the current default option) and 2017 (CRUNCEP) and have lower spatial and temporal resolution.

The text in the first message will be reduced as it will be copied/pasted in the commit message.

rljacob commented 1 month ago

@bishtgautam can you merge this?

bishtgautam commented 1 month ago

@rljacob Even a single year of EAR5 hourly data is quite large (~430GB). Let me see if I can use testmods to reduce the data size required by a factor of 12 as the 5-day tests don't need 12 monthly files.

rljacob commented 4 weeks ago

@bishtgautam is this ready now?