Response of tropical Pacific SSTs to climate forcing in CMIP6
The gradient of sea-surface temperatures (SSTs) between the western and eastern equatorial Pacific, with warmer temperatures in the western Warm Pool than in the eastern Cold Tongue, is a key asymmetry in the climate system. It influences where the atmosphere exhibits deep convection, how the Earth system radiates energy to space, and the nature of global teleconnection patterns. Past generations of CMIP models have generally shown a long-term El-Niño-like weakening of the Pacific SST gradient with global warming, whereas recent observations have shown a La-Niña-like strengthening. For example, Coats & Karnauskas (2017, https://doi.org/10.1002/2017GL074622) compared CMIP5 models to observations over the period 1900-2013 and found them to disagree on the sign of change.
This discrepancy is thought to arise from either (i) model biases in the eastern equatorial Pacific and long-term forced response (Seager et al. 2019, https://doi.org/10.1038/s41558-019-0505-x), (ii) a transient response to anthropogenic forcing that will eventually reverse sign (Kohyama et al. 2017, https://doi.org/10.1175/JCLI-D-16-0441.1), or (iii) natural climate variability. Investigation of this problem has been limited by the need for many ensemble members to distinguish the forced climate response from natural variability. The goal of this project is to take advantage of recently developed statistical methods that can identify the forced response with fewer ensemble members to compare the time evolution of tropical Pacific SSTs across CMIP6 models.
As an example (shown below), the average Pacific SST gradient response over 20 ensemble members in the CESM large ensemble still shows substantial variability associated with El Niño, which is not agreed upon between subsets of the full 40-member ensemble. However, a new signal-to-noise maximizing pattern filtering method is able to detect the forced response that is agreed upon across ensemble members, and to do so with as few as 3-5 ensemble members (Wills et al., in prep. – can share a draft on request). This enables investigation of the time evolving response of tropical Pacific SSTs to CMIP6 forcing scenarios within individual models (many of which run at least 3-5 ensemble members). This will help to identify the potential roles of transient oceanic responses and/or aerosol forcing in recent Pacific SST gradient trends and assess remaining biases in the long-term forced response. Related statistical methods can estimate the forced response in a single realization and can therefore be used to compare to observations.
Implement signal-to-noise maximizing pattern filtering in xarray and apply it to SSTs in CMIP6 historical and future scenarios. Examine the evolution of the equatorial Pacific SST gradient in particular.
Compare across CMIP6 models and scenarios. How do SST gradient changes vary by season? How does SST variability change with changes in mean state? Investigate associated changes in surface fluxes and thermocline depth. Investigate associated teleconnections (e.g., impact on global sea-level pressure or precipitation anomalies).
If this approach is successful for the Pacific SST gradient, it can also be applied to look at the time evolution of SST anomalies in other key regions such as the North Atlantic warming hole or the Southern Ocean.
Anticipated Data Needs
Monthly-average fields from CMIP6 historical and future warming scenarios:
• Sea-surface temperature
• Surface heat flux and wind stress
• 3D ocean data (temperature, salinity, wind) could also be nice to look at too
A secondary need would be for DAMIP aerosol-only or greenhouse-gas-only simulations, but not very many of these are available yet.
Anticipated Software Tools
Standard Python / xarray
Any climate scientists who are interested, especially experts in xarray.
Note that this is very closely related to the “Eastern Pacific SST Biases and Trends in CMIP6 vs. Observations” project proposed by @rabernat, and we will likely work together with or merge with anyone working on that project.