Improving Coral Paleoclimate Reconstructions

Since we have only been observing the tropical Pacific in detail for a few decades, drawing conclusions on the drivers of long-term variability based on instrumental data can be problematic. Applying paleoclimate reconstructions can help with the situation: for this part of the world, corals are one of the best 'proxies' for past variability. Corals grow rapidly, meaning they can detect variations on timescales of a few months (i.e. you can look at individual El Nino/La Nina events), they are highly sensitive to large-scale climate variability, and they grow in ideal locations like the central equatorial Pacific.

The isotopic ratio of oxygen in the aragonitic skeletons of these corals is what is commonly measured: this ratio is sensitive to both temperature and the isotopic ratio of oxygen in the surrounding seawater. Both of those things, in turn, are quite sensitive to large-scale changes in the ocean, and my work focuses on applying an improved understanding of the details of these linkages, using a combination of field observations, paleoclimate information, and ocean model simulations to make our reconstructions more accurate.

This work is funded by a combination of my NSF Ocean Sciences Postdoctoral Fellowship, and an NSF RAPID award which enabled the field component of the project.

The Isotope-Enabled Regional Ocean Modeling System

As wonderful as the existing database of coral paleoclimate records is, there are some fundamental obstacles to deploying them with sufficient accuracy to enable quantitative evaluation of climate models. A major issue is the lack of physical understanding of linkages between large-scale circulation changes and changes local to the reefs themselves: in other words, we don't have a very accurate idea of how much coral isotopic ratios at a particular island should vary given an El Nino event of a particular strength. And that is exactly what is required in order to understand where climate models are going wrong (or right!) with their representation of ENSO dynamics.

Part of the problem is that there are very few actual observations of isotopic ratios in seawater: this is a difficult quantity to measure, and although people are trying to fix the data gaps, what data are available are extremely sparse. So as part of my NSF fellowship, I developed a new tool which is capable of giving us a full, time-varying, three-dimensional spatial field of isotopes which are consistent with real 20th century climate variability. This is the isotope-enabled Regional Ocean Modeling System (isoROMS; Stevenson et al. 2015). isoROMS uses all of the same physics as "regular" ROMS, but includes the capacity to read in precipitative fluxes of water isotopes, and calculate the appropriate evaporative fluxes leaving the ocean (see figure below). isoROMS is also a step beyond isotope-enabled general circulation models for this particular problem, since it enables the simulation of observed ENSO events at any arbitrary spatial scale without the mean-state biases that exist in coupled climate models.

The initial validation of isoROMS is now complete, and has been published in Paleoceanography. I am currently finalizing a new simulation which goes further back in time (1948-2009), and am developing new 'conversion' methods between coral isotopes and metrics of ENSO variability which will be submitted for publication in the next few months.

Field Observations at the Line Islands During the 2015-16 El Nino

In support of my modeling efforts with isoROMS, my collaborators and I obtained funding through an NSF RAPID award to perform observations over the 2014-16 period at two prime locations for ENSO reconstruction: Palmyra Atoll and Christmas Island. We have created time series of temperature, salinity, current velocities, seawater and precipitation isotopic ratios for both locations at this point, and are actively working on using these data to better constrain the modeling results.

Check back soon for more details on our fieldwork!

Past Work: Uncertainty Quantification in Coral Records

I initially became interested in the coral reconstruction problem while working on ENSO changes in simulations with the CESM and other coupled climate models. Natural variability is typically so large that you need several centuries' worth of data to get robust ENSO statistics (Stevenson et al. 2010), so the next logical step is to try and extend the record back in time.

Previous work on the accuracy of ENSO variance from "pseudoproxy" estimates based on coral records (Stevenson et al. 2013) indicate that the errors are prohibitively large - over 50% in some cases! This was the inspiration for the development of isoROMS, above.