Eastern Pacific SST Trends in CMIP6 vs. Observations

Eastern Pacific SST Biases and Trends in CMIP6 vs. Observations

Scientific Motivation

The Eastern Tropical Pacific is one of the most important regions in the world in terms of climate variability and teleconnections. What happens here affects the entire globe. Some recent papers have noted how trends in this region are quite distinct in CMIP5 models vs. observations.

Coats, S., & Karnauskas, K. B. ( 2017). Are simulated and observed twentieth century tropical Pacific sea surface temperature trends significant relative to internal variability? Geophysical Research Letters, 44, 9928–9937. https://doi.org/10.1002/2017GL074622


This paper showed that, over the past century, the SST gradient is strengthening in observations, but this strengthening trend is not reproduced in models. In the obs, the eastern tropical pacific is warming more slowly than the western, enhancing the meridional gradient.

In a more recent paper, Seager et al. (including myself and @naomi-henderson), examined the trends across a very wide range of CMIP5 models and reanalysis products.

Seager, R., Cane, M., Henderson, N., Lee, D. E., Abernathey, R., & Zhang, H. (2019). Strengthening tropical Pacific zonal sea surface temperature gradient consistent with rising greenhouse gases. Nature Climate Change, 9(7), 517. https://doi.org/10.1038/s41558-019-0505-x


We argued that this different trends in models was a consequence of the cold bias of their equatorial cold tongues. So understanding and eliminating this bias is a key challenge for climate modeling.

@naomi-henderson has apparently done some digging into CMIP6 and confirmed that the bias is still there.

Proposed Hacking

  • Examine the tropical pacific SST trends in the CMIP6 models
  • Examine biases in SST climatology and thermocline structure compared with obs
  • Look for quantitative relationships between winds, equatorial currents (incl. the undercurrent), and cold tongue bias
  • See if this can explain the spread in SST trends under global warming

Anticipated Data Needs

Monthly-average ocean fields of

  • 3D ocean temperature, salinity, and velocity
  • Surface heat flux and wind stress

For preindustrial, historical, and future warming scenarios. Ideally we would do the full ocean heat budget, but I think this is not feasible with CMIP data.

For observations, we will need some reanalysis and climatologies (i.e. WOA). Kris also mentioned a new high-resolution glider-based climatology of the equatorial region that could be ideal model validation.

Anticipated Software Tools

This is mostly all doable in standard xarray.

Desired Collaborators

People who actually understand tropical dynamics! (Not me unfortunately :laughing:)

1 Like

@rabernat @naomi-henderson, this is very similar to what I wanted to propose as a project. I will add a related discourse proposal with a UW tag; it will focus a bit more on the time evolution of the Pacific SST gradient (and GHG vs. aerosol forcing). We can then consider merging if you are planning to work on this too.

1 Like

How are you planning on handling observations? Is there somewhere specific on the cloud that we should store them? Are they already there or is part of the hacking going to be writing code to scrape them from online databases and load as xarray datasets?

(I’m mostly asking because I am coming up against these questions in my projects too.)

@rabernat, this is a very cool project idea. I would like to work on this, in particular I would like to interface with the oxygen project (cc @yassir.eddebbar @matt-long).
We have recently looked at the Equatorial Undercurrent and its relationship to variability in the Pacific OMZ(aplologies for the paywall), and have found that the ‘shape’ of the EUC, in particular the lower boundary is important for the OMZ. I would propose to try to characterize the EUC in the CMIP6 ensemble (mean, variability, forced response) as part of this project (since its based in Lamont) and also interface with the folks over at @matt-long’s oxygen project? Do you think that is relevant here?

Ocean observations are usually pretty lightweight so can be loaded on demand from OpenDAP or downloaded on the fly. There is no specific tool or data catalog in place which can handle all observations.

More generally, I think it’s best if this project were merged with @rcjwills’s project:

I am realizing that, with my organizer duties, it’s probably not wise for me to also be leading a project myself.