Looking for ERA5/MERRA2 reanalysis data

[ndkeen]

I know this is possible and have heard that there is likely a directory somewhere on perlmutter for users to access ERA5 or MERRA2 reanalysis data to do this.

Would you be able to help me find where I can get access to this data?

[rljacob]

@paullric might know.

[chengzhuzhang]

yes, I would say to ask Paul, but under /global/cfs/cdirs/e3sm/diagnostics/observations/Atm/time-series you could found a limited subset of ERA5 and MERRA2 data that are used in E3SM Diags.

[jazweifel]

This is great! Noel asked this question for me after I reached out to NERSC inquiring about this reanalysis data location.

While we wait for a response from Paul I can add some context that I'm looking for this data to run a nudged E3SM simulation.

I poked around that directory above a bit but it looks like it has either monthly output or daily output for just one or two variables. I think I'll need daily ERA5/MERRA2 (u, v, t, q) data to nudge how I'm picturing. We were able to accomplish this in CESM with:

user_nl_cam:
met_data_file = '2000/MERRA2_0.9x1.25_20000701.nc'
met_filenames_list = '/glade/work/jzweifel/MERRA2/filenames_1975-2019_c200201_NOLEAP.txt'
met_data_path = '/glade/campaign/collections/rda/data/ds313.3/0.9x1.25/'

Anyone have an iea of where we could find daily data (u, v, t, q) and how to implement this in E3SM?

There's documentation about nudging in EAMxx but I'm just using E3SMv3.

Any help would be appreciated!

Jack

P.S. Also great to hear from you all again since seeing you at Berkeley last year!

[whannah1]

@jazweifel Just chiming in to say that the HICCUP tool I built can help with generating nudging data. However, the big caveat is that the data needs to be organized with all variables for a given time in the same file. Last time I checked - the data that is housed at NERSC is not organized in this way. Instead, it is grouped as one variable per file across many different times.

When I created HICCUP it was only meant to streamline the process of creating a single initial condition file at a time from ERA5+NOAA data for hindcast simulations. Using it to create nudging data seemed like a natural extension as long as you obtain the ERA5 data via the python interface from the servers in Europe - but many people don't want to bother downloading their own data for long nudging experiments. I don't have a good solution for this, and the number of people who need this type of experiment is very small, so modifying HICCUP to use the NERSC data is a very low priority.

Anyway, if you're interested, there should be a HICCUP template script included in the repo that you can look at, which is based on a previous nudging experiment I ran.

[mahf708]

my recommendation: you'd likely want to learn this stuff on your own and customize the nudging setup accordingly to suit your specific needs, or collaborate with someone/somegroup who will help you with that part. There's no generic out-of-the-box solution that will work for all users, and as such, any help you'd get would likely require a bit of specialization, which you won't find "for free"

[mahf708]

Since I am here, I will leave you with vague instructions regarding E3SM v3 nudging --- this is the stuff you need in your user_nl_eam:

! -- nudging settings -----------------
Nudge_Model              = .True.
Nudge_Path               = '/global/cfs/cdirs/e3sm/mahf708/ne30pg2L80'
Nudge_File_Template      = 'MERRA2_ne30pg2_%y-%m-%d-%s.nc'
Nudge_Times_Per_Day      = 4                                !! nudging input data frequency
Model_Times_Per_Day      = 48                               !! should not be larger than 48 if dtime = 1800s
Nudge_Uprof              = 2                                !! 2 if nudging tendency has a vertical profile
Nudge_Ucoef              = 1.
Nudge_Vprof              = 2
Nudge_Vcoef              = 1.
Nudge_Tprof              = 0
Nudge_Tcoef              = 0.
Nudge_Qprof              = 0
Nudge_Qcoef              = 0.
Nudge_PSprof             = 0
Nudge_PScoef             = 0.
Nudge_Beg_Year           = 0000
Nudge_Beg_Month          = 1
Nudge_Beg_Day            = 1
Nudge_End_Year           = 9999
Nudge_End_Month          = 1
Nudge_End_Day            = 1
Nudge_Vwin_Lindex        = 0.                               !! activated only when Nudge_Xprof = 2 !! X = U, V, T
Nudge_Vwin_Hindex        = 80.                              !! ""
Nudge_Vwin_Ldelta        = 0.1                              !! ""
Nudge_Vwin_Hdelta        = 0.1                              !! ""
Nudge_Vwin_lo            = 0.                               !! ""
Nudge_Vwin_hi            = 1.                               !! ""
Nudge_Method             = 'Linear'
Nudge_Loc_PhysOut        = .True.                           !! nudging tendency calculated before radiation
Nudge_Tau                = 6.0                              !! relaxation time scale, unit: h
Nudge_CurrentStep        = .False.
Nudge_File_Ntime         = 1                                !! should be the same as Nudge_Times_Per_Day or 1
Nudge_Allow_Missing_File = .False.

you may be able to infer what stuff means from the names, but there are all sorts of caveats, etc.

some nudging data can be found under $CFS/e3sm/mahf708/ne30pg2L80

My recommendation above stands though: learn this thoroughly on your own (which may be a major time sink) or collaborate more closely with someone who has thought about all these details, especially if they happen to be interested in the scientific topic of interest to you (people use nudging for all sorts of wicked things out there)

[mahf708]

and you should be able to download the reanalysis data on your own, then remap it to for the model of interest

[jazweifel]

Hey Naser, this was really helpful! I spent some time learning about the parameters available here and after a bit of configuring I was able to successfully use the MERRA2 data you pointed me to!

I just did a U, V, T, Q nudge globally like so:

! -- Global nudging of winds, temperature, and moisture ------------------
Nudge_Model         = .True.
Nudge_Path          = '/global/cfs/cdirs/e3sm/mahf708/ne30pg2L80/'
Nudge_File_Template = 'MERRA2_ne30pg2_%y-%m-%d-%s.nc'

! Time resolution
Nudge_Times_Per_Day = 8                     ! 3-hourly input data
Model_Times_Per_Day = 48                    ! 30-min model timestep

! Apply nudging globally (uniform profile)
Nudge_Uprof = 1
Nudge_Ucoef = 1.0
Nudge_Vprof = 1
Nudge_Vcoef = 1.0
Nudge_Tprof = 1
Nudge_Tcoef = 1.0
Nudge_Qprof = 1
Nudge_Qcoef = 1.0
Nudge_PScoef = 0
Nudge_PSprof = 0

! Activate only for a 5-day window
Nudge_Beg_Year  = 2009
Nudge_Beg_Month = 1
Nudge_Beg_Day   = 1
Nudge_End_Year  = 2009
Nudge_End_Month = 1
Nudge_End_Day   = 6                         ! Nudging ends just after day 5

! Method settings
Nudge_Method             = 'Linear'
Nudge_Loc_PhysOut        = .True.
Nudge_Tau                = 6.0              ! Relaxation timescale in hours
Nudge_CurrentStep        = .False.
Nudge_File_Ntime         = 1
Nudge_Allow_Missing_File = .False.

Thanks for all the help! Before I go do you know if there is anywhere else with this data and or more MERRA2 data on Perlmutter? We're looking to nudge for ten years. I could also just download some more MERRA2 data as well as you pointed out but wanted to see if it happened to already exist somewhere.

Thanks again!

Jack