ERA-5 forced ACCESS-OM2 simulations

Hi everyone,

This is a post to document some details for the ERA-5 forced ACCESS-OM2 runs that I have completed to a state which I’m happy to share. I’m hoping others with an interest can jump in and do some analysis. Hopefully the runs will be useful to someone!

Note that these runs are still somewhat of a work in progress. There are a number of known issues which have not been addressed yet (see @aekiss’s post at Documentation of the steps taken and problems that have been solved are at Support ERA5 forcing · Issue #242 · COSIMA/access-om2 · GitHub. That said, the results I’ve looked at so far seem reasonable.

I’ve done 4 ERA-5 runs (RYF and IAF, 1-degree and 1/4-degree), with 4 equivalent JRA55-do forced runs for comparison, with details listed in the below table.

Initial analysis of some basic metrics can be found here for the IAF and here for the RYF. Please feel free to expand on these and share your findings here!

Thanks to @aekiss, @russfiedler, Nic Hannah, @Aidan and probably a bunch of other people I’ve missed for their contributions.

Run name Time period Data location in /g/data/ik11/outputs/ Configuration
IAF Runs
025deg_era5_iaf 1980 to end of July 2023 access-om2-025/025deg_era5_iaf GitHub - rmholmes/025deg_era5_iaf
025deg_jra55v150_iaf 1980 to 2020 access-om2-025/025deg_jra55_iaf_era5comparison/ GitHub - rmholmes/025deg_jra55_iaf at jra55v150
1deg_era5_iaf 1960 to 2019 access-om2/1deg_era5_iaf GitHub - rmholmes/1deg_era5_iaf at ryan_testing
1deg_jra55_iaf 1960 to 2020 access-om2/1deg_jra55_iaf_era5comparison GitHub - rmholmes/1deg_jra55_iaf at era5comp
RYF Runs
1deg_era5_ryf 100 years access-om2/1deg_era5_ryf GitHub - COSIMA/1deg_era5_ryf
1deg_jra55_ryf 100 years access-om2/1deg_jra55v14_ryf GitHub - rmholmes/1deg_jra55_ryf at era5comp
025deg_era5_ryf 21 years access-om2-025/025deg_era5_ryf GitHub - COSIMA/025deg_era5_ryf
025deg_jra55_ryf 20 years access-om2-025/025deg_jra55_ryf_era5comparison GitHub - rmholmes/025deg_jra55_ryf at era5comp

Awesome, thanks @rmholmes!

Great work Ryan! I took a look at the notebooks, what would you say are the best and worst features of the ERA5 forced experiments?

Some things I noticed (as I am sure anyone else who took a look):


  • Scalars all look pretty similar, except for global mean temperature where ERA5 starts drifting earlier, but rates of drift look similar

  • Drake passage transports in 1-deg ERA5 never go as low and recover more quickly. 0.25 look the same

  • Arctic ice much lower for ERA5, though trends look similar.

  • Antarctic ice higher for ERA5, though trends look similar

  • Arctic summer minimum much lower for ERA5

  • Over-turning metrics didn’t plot correctly


  • Scalars all look pretty similar, except for 1-deg global mean temperature where ERA5 drifts down but stabilises and then rates of drift look similar

  • 1 degree ACC Transport is a bit odd, where ERA5 takes off, then comes back down around 1980 and basically shadows the JRA55 value until they start to diverge a little from 2010

  • Antarctic ice slightly higher for ERA5, though trends look similar. Winter maximum a fair bit higher for ERA5
  • Arctic ice lower in ERA5, and summer minimum much lower for ERA5 (Arguably too low?)

Edit: included @rmholmes figures inline, which I should have done originally

Thanks @Aidan.

The relatively consistent drifts in global T is one thing that surprised me. Some of the adjustments in JRA55-do were made I think to reduce drift, and so I was expected larger drifts with ERA-5 without these adjustments. But the magnitude of the drift isn’t really any larger.

Likewise, the consistency between the circulation metrics is also remarkable. The resolution has a much bigger impact. I guess its saying that the bulk wind stresses between the two forcing products are consistent, and model physics is important.

I’m hoping @willrhobbs will do a more detailed analysis of sea ice. I’m not sure I know enough to offer any interpretations. However, it is interesting to see that 025deg_era5_iaf run does capture some of this year’s very low Antarctic sea ice, although it may not be as extreme as in observations.

Re: RYF circulation metrics. Unfortunately I think I deleted the ty_trans_rho diagnostics from these (I think to save space in e14, it was a while ago). However, there are a few metrics plotted in this old notebook -

One last thing: The original reason I was interested in these simulations was to see if they improved on JRA-55’s really weak tropical instability wave activity in the equatorial Pacific. There are suggestions from some papers that the very weak North Equatorial Counter Current in OMIP-2 forced simulations is because of some of the adjustments made to integrate scatterometry data in JRA55-do (but not in ERA-5). However, in the RYF 1/4-degree simulations TIWs are only slightly stronger in ERA-5, and still way too weak (see last plots in I need to revisit this in the IAF simulations.

1 Like

@Aidan I’m bringing a poster on (Antarctic) sea ice comparison to the COSIMA workshop between ERA5, the IAF and obs (1/4 degree model). Not sure how far I’ll get with it, but ideally I’ll get as far as the dynamic/thermodynamic tendency terms,.

1 Like