Please note: You are viewing the unstyled version of this website. Either your browser does not support CSS (cascading style sheets) or it has been disabled. Skip navigation.


   Print  PDF  Change text to small (default) Change text to medium Change text to large


The modern-day global distributions of ocean properties, such as temperature, salinity, and nutrients,
tell the story of a system with distinctive and peculiar features. Small surface regions serve as portals
to the deep interior ocean, where these circulation paths resemble funnels that expand with depth.
Thus, great volumes of seawater can be traced back to a small fraction of the sea surface, and subpolar
and polar regions are most influential [Gebbie and Huybers, 2010, 2011]. Considering that the ocean
accounts for over 90% of the recent addition of heat in the Earth system, it is clear that these processes
ultimately dominate the global heat budget (with similar arguments available for the Earth’s carbon
budget, as well). The memory of the deep ocean to these surface changes lasts over centuries to
millennia, putting our understanding of the ocean’s role in climate dynamics in an uncomfortable
position. The planet’s energy balance is disproportionately affected by a few, poorly-observed polar
sites that set the conditions of the vast interior of the ocean and are locked in for many centuries at
minimum. The current role of the ocean in the greater Earth system therefore requires a consideration
of these long oceanic timescales, and knowledge of the cumulative effect of events that sometimes pre-
date our instrumental records.

To determine what the climate of the 21st Century will look like, projections on the decadal to
centennial scale are necessary, yet we have little information about internal ocean property variations
over the last century. To exacerbate matters, significant climatic changes are known to have occurred
in the past, such as the Little Ice Age a few hundred years ago to the Last Glacial Maximum (LGM)
around 20,000 years ago. While these changes were in the distant past, their climatic effect was large
and the resulting oceanic disequilibration is likely to have shaped the ocean as we know it today. In
my work, I seek to answer such questions as: How are the alternating warm and cool periods of the
past still affecting the ocean today? How does the ability of the ocean to uptake heat vary over time,
and to what extent will the ocean serve as a brake on atmospheric warming in the future? Ultimately,
I seek to contribute to basic climate questions such as, what is the origin of the carbon dioxide in the
pre-industrial atmosphere, and what does that imply about carbon uptake in the anthropogenic era?

I approach paleoceanographic problems from the perspective of a physical oceanographer, with the
belief that explanations of past climate are unlikely to be any simpler than the explanations of modern-
day phenomena. In problems involving tracer transport, I believe that the solution of a modern-day
analog is a prerequisite to understanding many paleoceanographic problems. My goal is to test hy-
potheses about the interpretation of observations, involving the design of a hypothesis-specific numer-
ical model. I usually find that a hypothesis is most cleanly tested through the use of an inverse method
(i.e., data assimilation), where the consistency of a model and observations is quantifiable.

Keck Cave
Enlarge image

(Keck Cave at UC Davis)

Visualizing ocean circulation over the last 30,000 years

This multidisciplinary and multi-institutional NSF-funded project brings together a team of computer scientists, physical oceanographers, paleoceanographers, and computational geophysicists (from WHOI, UC Davis, and UC Santa Barbara) to reconstruct the past ocean circulation from sparse geochemical data collected from fossils in deep sea sedimentary cores. The research takes advantage of the unique analytical resources and interdisciplinary collaborative environment provided by the UC Davis KeckCAVES (W.M. Keck Center for Active Visualization in the Earth Sciences), an immersive interactive visualization technology that helps identify meaningful patterns in complex datasets. As the paleo-data is too sparse to use simple interpolation mapping schemes, we are using knowledge of ocean dynamics together with inverse methods to provide the first global 4-dimensional maps of the properties of the paleo-ocean. This project is part of an overall goal of answering the question, just how different were ocean pathways during the Last Glacial Maximum (20,000 years ago)?

The project brings new visualization tools to WHOI through collaboration with the UC Davis group. Collaborator Oliver Kreylos, recently highlighted by the New York Times, is designing a low cost Virtual Reality system that will be housed at WHOI and will complement the KeckCAVES facility. Also, check out this highlight reel of visualization techniques.

Gebbie, G.,C.D. Peterson, L.E. Lisiecki, and H.J. Spero, "Global-mean marine d13C and its uncertainty in a glacial state estimate," submitted.

Streletz, G.J., G. Gebbie, B. Hamann, and O. Kreylos, “Interpolating Sparse Scattered Data Using Flow Information,” J. Computational Sci., 2016.

Kronenberger, M., C. Weber, G. Gebbie, O. Kreylos, L.H. Kellogg, L.E. Lisiecki, C. Peterson, H.J. Spero, B. Hamann, and H. Hagen, "A Novel Distance Measure for Ocean Reconstruction from Sparse Observations Demonstrated on the Atlantic," 2015.

Gebbie, G., "How much did Glacial North Atlantic Water shoal?" Paleoceanography, 29, doi:10.1002/2013PA002557. Also see auxiliary material.

Gebbie, G., "Tracer transport timescales and the observed Atlantic-Pacific lag in the timing of the last Termination," Paleoceanography, 27, PA3225, doi:10.1029/2011PA002273, 2012.

Gebbie, G., and P. Huybers, “How is the ocean filled?,” Geophys. Res. Lett., 38, L06604, doi:10.1029/2011GL046769, 2011.

Gebbie, G., and P. Huybers, “Total matrix intercomparison: a method for determining the geometry of water-mass pathways,” J. Phys. Oceanogr., 40 (8), doi:10.1175/2010JPO4272.1, 1710– 1728, 2010.

Deep ocean tracers, dynamics, and mixing

If ocean tracers are the most complete modern-day dataset and also the main source of proxy information for the pre-instrumental ocean, how well can the global large-scale ocean circulation be determined? In a series of papers [Gebbie & Huybers, 2010, 2011, 2012], I show that the property measurements of the World Ocean Circulation Experiment (WOCE) during the 1990s are sufficient to invert for a baseline ocean circulation that has higher resolution than any previous box inversion. My resulting circulation is consistent with several million observations of temperature, salinity, and nutrients, quantifying the information about the time-averaged and large-scale ocean circulation contained in the data. An underutilized form of information is the property trail that a conservative tracer leaves in its path, which permits the large-scale pathways and the relative effects of advection and diffusion to be inferred. The consequences are far-reaching; while previous investigators have been able to decompose the world ocean into a handful of traditional water masses such as North Atlantic Deep Water (NADW) and Antarctic Bottom Water (AABW), over 10,000 distinct surface sources are distinguished by this method. One specific implication is that the oft-cited equipartition of NADW and AABW in the deep North Pacific is untenable as it violates tracer conservation laws, such as the conservation of salt. Instead, the North Pacific has less than 20% NADW by volume, potentially making the North Atlantic a less powerful lever arm for climate change mechanisms.

Despite a good understanding of the large-scale velocity field from temperature and salinity distributions, general circulation models and radiocarbon observations disagree by a factor of 2 for the age, or time elapsed since contact with the surface, of the deep Pacific water. To reconcile this discrepancy, I developed a novel method to invert for the 3D global circulation by adding rate constraints from radiocarbon to the pathways previously diagnosed from the conservative tracers. The result is a direct estimate of the circulation fluxes due to both large-scale and small-scale processes, unlike previous methods that arrive at transit time distributions that indirectly describe the circulation. In my inferred circulation, the discrepancy in deep Pacific ventilation is reconciled by accounting for the mixing of a multitude of sea surface water sources. Due to the mixing, the true age of deep Pacific water is skewed toward the larger estimates: almost 1,500 years in some locations. In current research with S.R. Jayne, I am leveraging the opportunities opened by this global circulation estimate to map ocean mixing rates in a way that synthesizes direct observations of mixing, hydrographic observations, ventilation rates, and turbulent energy sources. With this advance, I am quantifying the net amount of interior ocean upwelling and downwelling, a key metric with implications for the closure and independence of the bottom and deep global overturning cells.

Past Ocean Circulation

The Atlantic meridional overturning circulation has been invoked to trigger, amplify, and enable
large climatic changes such as the waxing and waning of the great ice sheets during the Last Ice
Age. One of the most robust changes in the past ocean was seen in deep-sea stable carbon-isotope
ratios about 20,000 years ago during the Last Glacial Maximum (LGM). These observations have
been interpreted to reflect the shoaling of North Atlantic Deep Water and the vertical shift of the
deep Atlantic overturning cell, but transforming paleoceanographic data into physical oceanographic
quantities is a major obstacle. To test the hypothesis that the LGM circulation was rather different from
today, I used a compilation of paleo-data to constrain an inverse problem analogous to the already-
solved modern-day problem. Rather than a circulation change, the majority of the carbon-isotope
signal can be explained by changes in the Southern Ocean source value [Gebbie, 2014]. Unfortunately,
the true value of the southern source is highly uncertain due to the lack of observations there, as well
as possible post-depositional contamination. A best estimate is a shoaling of the deep water mass
interface of a few hundred meters rather that the oft-cited 1000 meters, as seen in the high-resolution
glacial water-mass atlas that I produced. By attempting to quantify the information in the paleo-data,
my research has shown that the conventional understanding of the LGM ocean may need to be revised.
In addition, this research suggests that the millennial-scale events of the last deglaciation, rather than
the LGM, were times where the circulation was more radically different from the modern-day.

My paleoceanographic research has leveraged a multidisciplinary and multi-institutional NSF project
with computer scientists and paleoceanographers. I have utilized the unique analytical resources
available at the UC Davis KeckCAVES (W.M. Keck Foundation Center for Active Visualization in the
Earth Sciences) to develop a suite of visualization and analytical tools to explore fundamental climatic
questions about changes in past global ocean circulation. The KeckCAVES is a tool that exploits the
human capacity to visually identify meaningful patterns in complex datasets and to interact with those
data through pattern recognition and change detection algorithms. I have developed a visualization
room at WHOI that uses the KeckCAVES software for multiple purposes: paleoceanographic research,
outreach projects including WHOI journalist fellows, and graduate education. As part of the larger
visualization project, I have taken on a leadership role in research that links computer science and
oceanography [Gebbie et al. 2015], especially methods to visualize ocean data in a way that is con-
sistent with ocean physics. I have advised two graduate students, Dan Amrhein and Greg Streletz,
and Postdoctoral Investigator, Tom Chalk, in this interdisciplinary project. In current research with R.
Ferrari and M. Jansen, I am combining visualization tools and inverse methods to test whether the
glacial bottom overturning cell was disconnected from the rest of global circulation, leading to an
isolated abyss that hoarded much of the carbon that is in the atmosphere today.

Connecting Past and Present Ocean Circulations

Given the long equilibration timescales, the ocean is never completely in equilibrium and is properly
regarded as transiently responding to surface conditions. This viewpoint is useful when considering a
puzzling observation of the ocean’s past; the observed midpoint of the last deglaciation occurred 4,000
years earlier in the deep Atlantic than the deep Pacific. Such a lag is greater than that expected by a
response to a uniform surface perturbation, and thus has been interpreted as a sign of a circulation
slowdown. Using the baseline circulation (discussed above) held fixed in time, I hypothesized that
the transient response to surface changes may explain the conundrum. To address the hypothesis, I
developed a fully-transient four-dimensional inverse method to reconstruct the deep ocean over the
last 20,000 years and to seamlessly fit the modern-day observations as well. I found that the observed
lag in the sediment core records does not necessarily point to drastic shutdowns in circulation rates, as
out-of-phase surface forcings in the Northern and Southern Hemispheres also can fit the data [Gebbie
2012]. The recovered deglacial time history of oceanic surface properties has striking similarities
to independent North Atlantic records not included in the inversion. The project emphasizes that
paleoceanographic surface changes share many similarities to those of the modern-day, as the entire
sea surface rarely changes as a uniform whole but instead has significant regional variability.

Instead of assuming that the ocean was in equilibrium prior to the industrial era, as is often done
in 20th Century simulations by the Intergovernmental Panel on Climate Change, I have forced an
ocean circulation model by surface temperature reconstructions that go back over the 2,000 years of
the Common Era [Gebbie et al. 2017]. A net cooling of the deep Pacific is expected on account of
the long adjustment timescales and entry into the Little Ice Age. The model simulates a deep Pacific
where waters that last encountered the surface during the Medieval Warm Period are being replaced
by those of the Little Ice Age, leading to a cooling. Such ongoing cooling is confirmed by analysis
of deep Pacific temperature changes since the 1870s, as recorded by the HMS Challenger cruise. In
fact, the deep Pacific cooling is about twice as large in the Challenger data than the model simulation,
indicating either that there are remaining depth-dependent errors in the data, or that the cooling of
the Little Ice Age was enhanced in the North Atlantic region.

Coupled Atmosphere-Ocean Dynamics

My research has also addressed whether the El Ni ̃no-Southern Oscillation is driven by external atmospheric noise or whether it varies due to internal dynamical feedbacks. First, I developed a model to simulate bursts of strong westerly wind along the equator (i.e., Westerly Wind Bursts) that was added to a coupled general circulation model. Then, I showed that a wind-sea surface temperature feedback is sufficient to spontaneously produce interannual variability. My work went beyond previous papers to show that this effect is large enough to be important in the realistic MOM4 hybrid coupled model [Gebbie et al. 2007]. My later paper [Gebbie et al. 2009b] provided the in-depth statistical analysis of how well westerly wind bursts can be predicted from sea surface temperature. The correlation between western Pacific warm pool extent and wind burst probability is robust, and has proven strong enough to improve El Ni ̃no forecast skill through the ocean’s memory [Gebbie et al. 2009a]. Forecast skill is also improved in a fully coupled general circulation model [Lopez et al. 2013], where my westerly wind burst model was adapted to co-exist with a free-running atmospheric model. Future steps include development of a forecast initialization scheme and making predictions for upcoming seasons.

A roadblock to understanding atmosphere-ocean interactions is that there is no reanalysis (i.e.,data assimilation) method that is dynamically consistent with both the equations of the atmosphere and the ocean. Without dynamical consistency, it is difficult to identify the physical mechanisms that matter. Summer Student Fellow Tsung-Lin Hsieh and I have adapted a dynamically-consistent reanalysis method and tested it with the chaotic pendulum, a system analogous to an eddy-resolving ocean model or an atmospheric model. We provide evidence that nonlinearity is not the fundamental obstacle in producing a reanalysis, but rather it is the “controllability,” or the ability to drive the model into an arbitrary state [Gebbie and Hsieh, 2017]. Ocean models that depend upon uncertain surface forcings are likely controllable, even if they have high-resolution and are highly-nonlinear. This research gives promise in the ability to make a coupled ocean-atmosphere reanalysis if the equations are formulated following our pendulum example. I am excited to test these results with the ECCO (Estimating the Climate and Circulation of the Ocean) re-analysis system [Wunsch & Heimbach, 2009], as I believe that eddy-resolving ocean reanalyses and coupled ocean-atmosphere reanalyses can be achieved.

Enlarge image

(Photo by Marinka)

Ice sheet-ocean interactions
As part of the WHOI Arctic Research Initiative, we are tracking the signal of icesheet-ocean interaction in the large-scale water-mass properties of the global ocean with the goal of determining how much the melt of icesheets is accelerating. Preliminary results have diagnosed and quantified the influence of Ronne-Filchner Ice Shelf Water in the formation of Antarctic Bottom Water in an unusual but effective way, using the relatively well-observed seawater signatures of freshwater, temperature, nutrients, and dissolved oxygen to back out what has happened on Antarctica. This method will be applied to the accelerating ice melt occurring on Greenland, as well.

Last updated: December 11, 2017

whoi logo

Copyright ©2007 Woods Hole Oceanographic Institution, All Rights Reserved, Privacy Policy.
Problems or questions about the site, please contact