By Aimée


Imagine someone tampered with a thermostat in a building and now it’s faulty. It cannot maintain the set temperatures and allows the heating (or cooling) from outside to influence inside temperature. The culprit turned himself in, but the problem now is to understand how much more (or less) the temperature in the building will change each day from its normal temperature. Things are of course made complicated because the temperature that was set by the thermostat already varied from day to day, season after season.  The thermostat can’t be fixed, although the temperature variation problem can be fixed by changing the humidity in the building, opening (or closing) windows and the like. So, a year later, how does one quantify the current temperature variations that have been superimposed onto the original ones?

This random analogy is (somewhat) similar to what forensic climate scientists (a.k.a. paleoclimatologists, paleoceanographers) are faced with.  The latter half of the 20th century has been the warmest recorded in the past millennium. But it’s not just the temperature. Lots of other things are unprecedented: frequency of cyclones, length of droughts, wind patterns, strengths of ocean currents et cetera. We know that human-induced (anthropogenic) activities have exacerbated some of these changes but in order to refine future climate predictions we need to understand the natural variability that underlies these atypical changes.

Back to the thermostat scenario. You might say the best way to quantify deviation of temperature from previous values is by doing a simple subtraction: obtain thermostat records prior to its malfunction and compare them with the anomalous ones from the last year. I’d say, “great idea”! Problem is the only records that are available when the thermostat was working well, are for the last 6 months. Prior to that no records were kept. In order to really understand how the temperature varied from season to season, you’d need at least 5 years of records. On top of that different parts of the building react differently. Rooms that have North-facing windows will be warmer than those with South-facing ones because they get direct heating from the sun all year round. The ones with East-facing windows will behave opposite to West-facing windows at different times of day. The best hope we’ve got is to look around for “evidence” left in the building on how past temperatures have varied in the past 6 years. And the more rooms we can get “evidence” from, the better the understanding of how temperature are likely to affect the building as a whole in the future.

In climate-speak, this “evidence” is in form of archives. They faithfully record climate changes. Now you may wonder, where do forensic climate scientists draw their “6-year mark”? Good question. They answer is quite subjective – it depends on the environmental parameter (proxy) under investigation, but the archive usually sets the terms and conditions. The Holocene* has proved to be an excellent period of time to use to understand background present-day natural variability. The Holocene, last 11,700 (or 11,500) years, is the time period where “boundary conditions” are most like they are today. For example sea-ice extent, ocean circulation patterns, sea-level, topography – all have remained virtually similar to what we see today. Therefore understanding how climate acted during this period will give us an idea of how the climate should behave now, sans anthropogenic forcing.

These archives range from tree rings, to lake and marine sediments, to speleotherms, to corals (my favourite thing). So far they’ve told us a lot about the earth’s climate on various timescales. For example, (fossil) corals in the pacific are the ultimate archives for reconstructing El Nino (ENSO) patterns. Information on ENSO is important as it helps drive other aspects of global climate change. A paper released earlier this year in Science neatly shows how looking at the past can help us fine-tune current climate models. Associate Professor Kim Cobb and her team looked at fossil corals from the Line Islands  and showed that the large ENSO variations seen of recent are not unprecedented. These large variations occurred previously in the Holocene. Climate models assume that these variations are mainly externally driven by orbital cycles, variation in solar radiation, volcanism et cetera. Whereas Cobb et al., demonstrate that the ENSO variance does not show much correlation with external forcing – there is a lot of internal (natural) variability that needs to be well characterised. That said, the agreement between models and proxy records is quite astounding and it makes these models reliable. The problem is mainly with spatial variability (different rooms of the building acting in different ways to the same thing). More data is needed and there’s always more that needs to be discovered, refined and re-defined. That’s what makes being a forensic climate scientist so much fun! I propose a new TV show. Enough with the crime shows already!

 *When it comes to deciphering climate records, the Holocene period does not usually include the period after the industrial revolution when we messed up the isotopic composition of the atmosphere by burning fossil fuels. Then atomic bomb testing in the 1950s really didn’t help the natural climate cycle at all. So the last two hundred years are usually called “modern” or some have used this fancy word – Anthropocene (although the start of the Anthropocene is a highly contentious issue)!