Humans have been recording the planet's average temperature since the late 1800s but that's not nearly enough to understand where the current climate is heading, absent a geological timeline context. Luckily, scientists have all sorts of proxies they can use to gauge past temperature; things like tree rings, fossilized shells or ice cores. One recent study published in Nature used, for the first time, sediment cores drilled from the ocean's floor to extend the climate's historical temperature reading to a continuous two-million-year-old timeline. It's the longest reliable reading of the planet's average temperatures so far.
Carolyn Snyder was responsible for patching up all the sediment core data as part of her PhD dissertation at Stanford University. There were 59 cores in total which were used to calculate 20,000 individual ocean temperature data points. That's about a data point every 100 years over two million years. This means the temporal resolution is pretty bad and the study couldn't capture sudden fluctuations in temperature -- like the one we're experiencing now -- but it's enough to record major changes in the climate like ice ages which need thousands of years to unfold.
Even so, Synder's readings are still very useful and seem to match with previous re-constructions. Ice cores, for instance, have been extensively studied and these can gauge past temperatures up to 800,000 years ago. When compared with the sediment readings, Synder got a correlation of 0.72, where 1 would have been a 'perfect match'. But ice cores aren't exactly the most reliable data points since these record temperatures at the poles and do not necessarily reflect average global fluctuations. Snyder's correlation might be even better.
What the sediment cores tell us is that for the first 800,000 years or so, the planet gradually cooled. By 1,2 million years ago, this trend stopped and the climate entered a transitional period marked by more frequent bouts of warming and cooling. During this shift, known as the mid-Pleistocene transition, glacial cycles happened every 40,000 years instead of every 100,000 years as recorded previously.
It's not clear why this transition happened -- or if it happened in at all. Maybe a change in external causes, like those in orbital forcing or the sun's activity was responsible, or maybe the poor temporal resolution missed it. More sediment cores might help answer this.
In the second part of the paper, Snyder makes some bold assessments, though. What she did was to couple carbon dioxide with the temperature readings she gathered over the past two million years to determine the climate's sensitivity to greenhouse gasses. This measure tells us how much the planet will warm per doubling of carbon dioxide. Using this history, Snyder forecasted a 9°C warming per doubling -- that's almost three times more than the best IPCC estimate of 3°C per doubling.
Should we be very worried, then? A lot more than we already are? Not so fast, said Michael Mann, a paleoclimate expert at Penn State.
“The estimate of earth system sensitivity is so much higher than the prevailing estimates that one has to consider it somewhat of an outlier, and treat it with an appropriate level of skepticism,” Mann said.
“I regard the study as provocative and interesting, but the quantitative findings must be viewed rather skeptically until the analysis has been thoroughly vetted by the scientific community.”
Richard Alley, another scientist from Penn State, was more direct. He said the 9°C value Synder reached is the absolute upper limit for the planet's climate sensitivity. Reality is a tad different since this sensitivity isn't only dependent on greenhouse gasses.
"This assumes that all of the temperature change over the ice-age cycles arose from the greenhouse-gas change," Alley told Ars. "But, we have high confidence that the ice ages were driven by features of Earth’s orbit and that the temperature would have changed (just not as much) if the greenhouse-gas forcing had not changed."
While the second part of Snyder's paper is controversial, the two-million-year-record is highly valuable. Now, we need to extend the timeline and improve that temporal resolution. The more date we have, the better our climate models become, and with them our ability to forecast climate change.