Aerosol effects in OpenIFS 43r3?

AIFS, IFS and OpenIFS OpenIFS
, , ,
1

Good morning

I've been digging into understanding how aerosols are treated in OpenIFS 43r3v1. Some output from my NODE file is at the end of this post.

My questions boil down to:

1) Is there any treatment of aerosol indirect effects, i.e. aerosol-cloud interactions? I only see direct impacts on SW and LW forcing.

2) Aerosols are just a monthly climatology. Can I somehow use time-varying aerosol forcing for e.g. AMIP simulations of 1979-2014? Something like MACC SPv2 would be fantastic. Can I use something for pre-industrial conditions, when there was definitely sea salt and dust, but very little anthropogenic emissions?

So far, my AMIP simulations for 1979-2014 show some cooling of global-mean surface temperature around 1991-1993, which would be consistent with the Pinatubo eruption. But I have a feeling that this is a result of the observed SSTs and not any aerosol effect. Does that make sense?

Many thanks
Joakim

From my NODE file:

General settings:
Data files expected in "ifsdata/"
Clear-sky calculations are ON (do_clear=T)
Saving intermediate radiative properties OFF (do_save_radiative_properties=F)
Saving spectral flux profiles OFF (do_save_spectral_flux=F)
Saving surface shortwave spectral fluxes ON (do_surface_sw_spectral_flux=T)
Gas model is RRTMG-IFS (i_gas_model=1)
Aerosols are ON (use_aerosols=T)
Longwave derivative calculation is ON (do_lw_derivatives=T)

Reading NetCDF file ifsdata//aerosol_ifs_rrtm_43R3.nc
Reading mass_ext_sw_hydrophobic(14,14)
Reading ssa_sw_hydrophobic(14,14)
Reading asymmetry_sw_hydrophobic(14,14)
Reading mass_ext_lw_hydrophobic(16,14)
Reading ssa_lw_hydrophobic(16,14)
Reading asymmetry_lw_hydrophobic(16,14)
Reading mass_ext_sw_hydrophilic(14,12,5)
Reading ssa_sw_hydrophilic(14,12,5)
Reading asymmetry_sw_hydrophilic(14,12,5)
Reading mass_ext_lw_hydrophilic(16,12,5)
Reading ssa_lw_hydrophilic(16,12,5)
Reading asymmetry_lw_hydrophilic(16,12,5)
Reading relative_humidity1(12)
Closing NetCDF file ifsdata//aerosol_ifs_rrtm_43R3.nc
Aerosol mapping:
1 maps on to hydrophilic type 1: Sea salt, bin 1, 0.03-0.5 micron, OPAC
2 maps on to hydrophilic type 2: Sea salt, bin 2, 0.50-5.0 micron, OPAC
3 maps on to hydrophilic type 3: Sea salt, bin 3, 5.0-20.0 micron, OPAC
4 maps on to hydrophobic type 7: Desert dust, bin 1, 0.03-0.55 micron, Woodward 2001, Table 2
5 maps on to hydrophobic type 8: Desert dust, bin 2, 0.55-0.90 micron, Woodward 2001, Table 2
6 maps on to hydrophobic type 9: Desert dust, bin 3, 0.90-20.0 micron, Woodward 2001, Table 2
7 maps on to hydrophilic type 4: Hydrophilic organic matter, OPAC
8 maps on to hydrophobic type 10: Hydrophobic organic matter, OPAC (hydrophilic at RH 20-30%)
9 maps on to hydrophobic type 11: Black carbon, Olivier Boucher
10 maps on to hydrophobic type 11: Black carbon, Olivier Boucher
11 maps on to hydrophilic type 5: Ammonium sulfate, GACP
12 maps on to hydrophobic type 14: Stratospheric sulfate (hydrophilic ammonium sulfate at RH 20-30%)
Reading NetCDF file ifsdata//mcica_gamma.nc
Reading fsd(81)
Reading x(1000,81)
Closing NetCDF file ifsdata//mcica_gamma.nc
Tropospheric bacground uses aerosol type 8
Stratospheric bacground uses aerosol type 12


2

Hi Joakim,
1) There is no treatment of aerosol-cloud interactions in the IFS.
2) Not in the default setup for medium range forecasts, but something like this will have been used for reanalysis, although I don't know if such settings are available in OpenIFS - perhaps Paul Berrisford would know?

I am pretty sure that there is no representation of date-dependent stratospheric aerosol in the IFS so I would think that any Pinatubo effect would be seen only via SSTs.
Note that we are currently working on a version of the aerosol climatology that can represent longer term trends in anthropogenic aerosol species, but this wouldn't be available until 49r1 at the earliest.
Regards,
Robin.

3

Hi Joakim,

Reanalysis started using variations in tropospheric sulphate and stratospheric, volcanic aerosol in ERA-20CM, see sections 3.2.4 and 3.2.5 in:

https://rmets.onlinelibrary.wiley.com/doi/10.1002/qj.2528

I don't know if these settings are available to use in OpenIFS, sorry.

Regards,
Paul

4

Hi Joakim, interactive aerosols are being worked on by several people in the OpenIFS/AC project and there is ongoing work to implement the CMIP6 aerosols for EC-Earth4 which is based on OpenIFS. Sadly I cannot tag people but hopefully Markus will catch this thread. MACC SPv2  will probably not make it into openifs...

5

Hi Joakim, as part of the ongoing OpenIFS/AC project both KNMI and FMI are indeed implementing interactive aerosol processes into OpenIFS 43r3. 

The aerosol properties are however not passed to the model through external time-varying forcing files (which you seem to be referring to?) but it is done by including an interactive aerosol code in the model, together with either a minimal or a more elaborate interactive atmospheric chemistry package. 

I didn't have the impression that this is what you were looking for, but if this is of interest to you (or if you want to know more about CMIP6-type transient simulations with EC-Earth4) I would suggest to get in touch with Twan Van Noije at KNMI. 

6

Hi all

Thanks for the replies!

Sounds like there are some versions of IFS that have used time-varying aerosols at some point.
I also came across Chris Roberts' paper from 2018 where he says:

Tropospheric aerosol forcing is specified using version 2 of the Max Planck Institute Aerosol Climatology Simple Plume model (MACv2-SP Stevens et al., 2017), which was implemented within the IFS radiation scheme specifically for this configuration. MACv2-SP enables the prescription of the anthropogenic aerosol optical properties and associ- ated Twomey effect, and consists of nine spatial plumes re- lated to the major anthropogenic sources. The amplitudes of each plume are scaled through time to provide region- specific variations in tropospheric aerosol through the his- torical period.

Volcanic forcing is specified as a total stratospheric aerosol optical depth (SAOD) that varies with time and latitude.

That sounds like exactly what I'm looking for. And he was using 43r1 which is a lot closer to 43r3 that cy38 used for ERA-20C (I think?). Is that model version around somewhere and would it be a not-too-difficult merge?

Many thanks
Joakim

7

The radiation code changed in 43r3, and a new aerosol climatology was introduced. This changed the way aerosols are specfied (requiring aerosol mass per model layer, rather than optical depths), so work from earlier cycles does not carry over easily. The IFS does retain the ability to read in and use time-varying volcanic aerosol, and the optical depths at 525nm are converted to mass densities assuming the optical properties are those of tropospheric sulphate aerosols. This is how the latest cycles work, you can check for 43r3 by searching for STRAT_BG_AER_MASS_EXT in module/radiation_setup.F90 and phys_radi/radiation_scheme.F90. If you find these you can hopefully trace how it all works. Volcanic aerosol arrays are set in routine suecaec in module/yoeaerc.F90, and there are several options which are coded, including reading in data from a file. Our scripts define two files: cmip6_stratospheric_aerosol_aod.dat and CMIP5 volcanic_giss.data, but only vertically integrated amounts are used (and then re-projected to model levels using functions).

At ECMWF I am presently working (in the context of the H2020 CONFESS project) on upgrading how we handle the vertical distribution of volcanic aerosol, but plan to keep this basic mechanism for converting optical depth (which is what the observed datasets contain) to mass, so that when the radiation code calculates the optical depth, it matches the original input data.

For tropospheric aerosols, as Robin mentioned, ECMWF are working on a time-varying climatology based on simulations with the CAMS aerosol and chemistry schemes. We hope that these will become available in Cy49r1 in time for ERA6 and SEAS6 in 2024. Note that in pre-industrial times there was still a lot of atmospheric chemistry going on, and both natural and human induced aerosol emissions of many sorts. It was our inability to separate our existing aerosol climatology into "background" and "anthropogenic" that led us to ditch the MACv2-SP approach in favour of direct simulation of everything, despite the extra complexity. Our new approach also gives much more realistic spatial distributions than MACv2-SP. We are still in the middle of this work, though, and don't have a finished product yet.

8

Many thanks for the reply, Tim! I really appreciate you taking the time. And I can see now that copy-pasting code from 43r1 to 43r3 won't work.

I will dig into the code to see if I can understand the volcanic aerosol treatment and the various options. Reading in cmip6 aerosol AOD sounds very interesting. Might get back to this thread if I get lost completely...

On a side note, rumour has it someone in Helsinki is working on OpenIFS-HAM, i.e some form of the HAM-MOZ chemistry scheme (from ECHAM) is going into OpenIFS. Not sure if anyone from ECMWF is involved in this project.
If I learn something about this, I can post it here for the sake of completeness.

Many thanks again! Wishing you all a good weekend!
Joakim