About 6 months ago, I wrote up a summary of how paleoclimatologists determine the temperature of the past, focusing primarily on oxygen isotopes Paleoclimatology Primer. It was relatively well received, so I figured I would continue the lessons on paleoclimatology. Today we will focus on the problems/opportunities inherent to the oxygen isotope system, and a few other systems that help us get a stronger handle on the climate of the past. I highly recommend you go back and read that post first, as this post builds upon what we learned 6 months ago.
As a warning, almost all of the links I have provided here require a paid subscription (with a few exceptions), but almost all of the links are from major journals (Science, Nature, Geology, etc.) and your local library is likely to have a subscription. The abstracts of each paper are always free though.
While it has been a while since I posted, this post will hopefully be the first in a 6 part series I hope to put up over the next two months or so. The first two posts will principally focus on how scientists create temperature reconstructions in greater detail than the primer I posted earlier. The following three posts will be about the trends we see in the climate system on different time scales. The first of those three posts being the climatic trends over the last ~33 million years (hint: it has gotten steadily colder). The second post will focus specifically on the last million years (hint: it has gotten colder and more variable). The third post will focus on the trends we see just over the last 60,000 years through the last glacial to today. The final of the five posts will talk about the history of sea level change: how we study it, what has happened in the geologic past, what has happened in the instrumental past, and what is happening now. By the end of this series, you should be more than ready to school pretty much any one (who isn’t a climatologist of course) about the Earth’s natural climate cycles.
In the primer on paleoclimatology I told you about oxygen isotopes and what they can tell us when we measure them in the CaCO3 tests of foraminifera. We extract temperature data from the 18O/16O ratios of the foram tests, but the ratios do not purely reflect temperature. They also reflect the 18O/16O ratio of the seawater in which the shells precipitated. The oxygen isotopic ratio of seawater reflects two components: how ‘heavy’ the ocean is with respect to oxygen isotopes caused by sequestration of light oxygen in ice sheets on the continents (which I told you about last time), and the evaporation/precipitation effect (erroneously referred to as the ‘salinity effect’ in many articles/books/papers). All three components (temperature, ice volume, and evap/precip) affect the 18O/16O ratio in the carbonate shells of foraminifera. So how do we extract the portion that is due solely to temperature?
Let’s focus on the easy one first, extracting the ice volume effect from your oxygen isotope record. Ideally, if you can know the average oxygen isotope ratio of the seawater at any given time, than you could simply subtract that value from your data and you would be left with only two variables to account for. But unfortunately (for climate science, but fortunately for those of us who want to, you know, live), the active hydrologic cycling on our planet constantly changes the oceanic composition, so the glacial ocean no longer exists. There are a couple ways to deal with this though.
Unfortunately, there aren’t a lot of options for direct measurement of ancient water (we will discuss those options a little further down). However, we have some means of addressing the problem indirectly. A rather clever attempt at addressing the changes in δ18O due to the ice volume effect was to look at the oxygen isotopic value of benthic foraminifera from places where the water can’t get any colder than it is today (just barely above the freezing point). There are a handful of places in the world’s ocean where the water temperature at the seafloor is between -1 and 0°C. Labeyrie et al. (1987) (Variations in mode of formation and temperature of oceanic deep waters over the past 125,000 years.) looked at the benthic δ18O record from the deep Norwegian Sea where temperatures are currently -1°C. The thinking was that the temperature of the deep water here could not possibly have gotten any colder, so any change in the δ18O record in the direction of heavier isotopes can be purely attributed to changes in the average value of seawater due to changes in ice volume. Labeyrie’s work showed a change of ~1.1‰ during the last glacial from today. Among the problems with this technique is in addition to assuming the water doesn’t get any colder, it also assumes the water doesn’t get any warmer. If it did, they could potentially be underestimating the ice-volume effect.
Another attempt at determining the ice volume effect on the δ18O of seawater during the last glacial utilized the fact that certain modern species of coral have very specific depth habitats (<5 meters from the surface). Richard Fairbanks in 1989 (A 17,000-Year Glacio-Eustatic Sea-Level Record - Influence of Glacial Melting Rates on the Younger Dryas Event and Deep-Ocean Circulation.) used ancient coral deposits to determine the sea level curve coming out of the last glacial cycle. Once he had developed a sea-level curve, he merely used an equation that turned meters of seawater into the oxygen isotope value of seawater based on the mean values of ice in Greenland and Antarctica. He accounted for a seawater change of ~1.2‰ in the oxygen isotopes using this method. It is a crude method, true, but at the time, it was the best the scientific community had. Waelbroeck and others (2002) (Sea-level and deep water temperature changes derived from benthic foraminifera isotopic records) did a similar calculation (sea level to oxygen isotopes) albeit from a much more complicated data set and algorithm. They posited a change of ~0.95‰.
Of course, the ideal way to figure out the oxygen isotope ratio of an ancient ocean is to find some of the ancient water from that ocean sequestered somewhere safe where it could not be altered over time. When dealing with water values of the distant past (think millions of years), one place you can look is in evaporite deposits (Hardy 1996; Lowenstein et al. 2001). Evaporites are highly crystalline minerals that form when a water body evaporates. Think salt. When salt crystals (or gypsum or any of the other evaporites) are forming the crystals grow together in rather random directions. Sometimes, if you are lucky, the crystals grow together in such a way that they leave an open space in between them. This space frequently contains some water from the ocean basin in which they formed. This water is actually a geologically isolated sample of an ancient water body (thanks geology!) There are caveats though.
If a body of water is evaporating, it probably is not in direct contact with the open ocean, and therefore is unlikely to reflect the true open ocean signature of the ice volume effect. Therefore the error bars are fairly large. In addition, aside from the difficulty in getting samples, dating when the crystals formed, and the age of the water contained within is notoriously difficult to do thanks to the ductile nature of salt and its tendency to flow under pressure. In addition, if you are interested in any of the other ions within the water sample (magnesium, boron, etc.), they will have been concentrated in the evaporation process, so you don’t have any other isotope systems to help you extract other data. However, in truly ancient paleoclimate systems, this data may be all you have.
The last glacial cycle (~100,000 years) is special in that there is another place where we can look for geologically isolated water samples that is not impacted by the problems we have in halite inclusions. This technique, illustrated beautifully by Daniel Schrag and others (1996) (Pore Fluid Constraints on the Temperature and Oxygen Isotopic Composition of the Glacial Ocean) focused on the pore water within a sediment column drilled in the tropical Atlantic. Essentially, water at the bottom of the ocean can advect downward into the sediment column and become isolated from the ocean. Schrag performed oxygen isotopic analysis on the pore water fluids throughout the sediment column. After calculating the diffusion coefficient of the sediments at that site based on a number of parameters, including temperature and porosity, they managed to establish a model of fluid flow within the sediment column. This allowed them to discount upward advection/diffusion of water in the sediment column and identify the pore waters that were sequestered since the last glacial. The value they assigned to the δ18O of seawater during the last glacial was ~1.0 ‰ below today.
Most of the paleoclimate community today uses ~1.1‰ when attempting to subtract the ice volume effect from their oxygen isotope curves for the last glacial. I know the values I have given you thus far mean nothing without their context; I might as well be measuring in hamsters. But I gave them to you for a reason. The results from the different methods range from 0.95 – 1.2‰. That is a spread of 0.25‰. Now if I told you that 1°C is approximately equal to 0.22‰ in the δ18O record (we will focus more on this fact next time), the reason for all the different techniques becomes clear. And before you ask, what is the point of arguing over a single degree, remember that the lower limit of human caused climate change we are facing by the end of this century is projected to be approximately 1.1°C on average. One degree of temperature change for our planet as a whole is a very major deal. In fact, the IPCC says that:
There is medium confidence that approximately 20 to 30% of plant and animal species assessed so far are likely to be at increased risk of extinction if increases in global average temperature exceed 1.5 to 2.5°C over 1980-1999 levels. Confidence has increased that a 1 to 2°C increase in global mean temperature above 1990 levels (about 1.5 to 2.5°C above pre-industrial) poses significant risks to many unique and threatened systems including many biodiversity hotspots.
As you can probably imagine, given all of this work just to figure out the change in the δ18O record of seawater during the last glacial maximum, trying to establish a comprehensive curve of seawater δ18O through time is remarkably difficult. It may even prove to be to complicated to accomplish. There have been a few attempts, such as Fairbanks and Waelbroeck’s referenced above. But the level of detail that may be needed to address short term events may prove prohibitively difficult. Ultimately, as scientists, we have to make educated guesses, and present our data in a range rather than discreet points to account for the uncertainty.
Next time, we will go over how to strip out the evaporation/precipitation part of the δ18O signal, and wrap up what we know about paleotemperatures from a few other proxies.
Update: I updated the diary to make clear that the 1°C comparison between the potential error in the reconstructions and human induced climate change is a comparison to the low end of the projections over the next 90 years. The IPCC states that the temperature increase by 2100 will be between 1.1 and 6.5°C based on a range of societal and governmental responses to the threat of climate change. I am not stating that 1°C is the most likely result, as the most likely result is probably higher. I simply wanted to show how catastrophic the low range would be, because it is comparable to the error we were trying to work out in my field (which I think is pretty much taken care of at this point). The point of this diary is not to address what makes the different IPCC reconstructions give different results (which might make for a good diary in the future, but I would need to joint write it to fill in some of the gaps in my own understanding of the physics). The point was simply to make clear how the paleotemperature record is constructed. I hope I accomplished that goal.
I also corrected a handful of spelling errors (he -> the, fro ->for, Paleontology -> Paleoclimatology). You know, the common mistakes that spell check misses.