If you build a temperature record going back in time to judge modern warming against, how certain can you be of your answer? That’s a big question for scientists making such records from effects temperatures have had on the natural world. And figuring out if today’s heat is unique is too great a challenge for the methods scientists normally use to calculate uncertainty, according to Harvard University’s Martin Tingley.
But Martin and Peter Huybers have shown the precise chances that northern areas of the world are warmer than any time in rebuilt records reaching back to the year 1400. They have worked out that there’s less than one chance in 20 that 2005, 2007, 2010 and 2011’s northern summers weren’t the warmest in that time. They also find that summer 2010 has a 99% chance of being the warmest western Russia has seen. There have already been lots of claims made over the unusualness of recent warmth, Martin pointed out, but his and Peter’s are the most robust yet. “We put these estimates on a much sounder statistical footing,” he told me.
Saying one year’s summer is uniquely warm across a long period is difficult for subtle reasons that Martin explained through his height. “I’m a tall guy, 6 foot 4 inches,” he said. “I’ve never met you, but I’m going to bet I’m taller than you. What’s the intuition behind my bet? We have a sense of the distribution of heights. I’m aware I fall pretty far out on the tail, so the chances are if I meet an average person they don’t fall further out than I do. What if I’m in a room with 1,000 people I’ve never met before? Am I still likely to be the tallest in the room? Probably not.”
It would be relatively easy – though time-consuming – to find out if Martin was definitely the tallest in such a busy room with a ruler. But the natural proxy measures used to build historic temperature records are like trying to work out heights from footprints. “We don’t have an exact measurement for each year in the past,” Martin underlined. “We have an estimate, with a distribution of possible values associated with it. Suddenly the uncertainty information that one needs to be sure that a year is warmer than all other years is much different.”
Over the years scientists have gathered proxy data from many different places. And so Martin and Peter used them to make a temperature map of all the land above 45° latitude – effectively meaning everywhere north of Italy and Wyoming. To help sleuth down unique temperatures, they brought 125 different sets of tree ring, ice and lake sediment layer thickness proxies together with modern measured temperatures.
Then, to chip away at the proxies’ uncertainties, Martin and Peter modelled their temperature readouts using a tool called BARCAST that they had previously developed. BARCAST produces an ensemble of thousands of possible histories, based on the links between actual temperatures and direct and proxy measurements at every location. Like Martin’s sense of his height, the scientists built the links from their prior knowledge. However they change the strength of the links slightly between different histories, and allow measurement data to vary slightly to simulate ‘noise’ that can happen in these records. They also assume that places that are close together are similar in their climate for a given year, while one year’s climate is similar to the next at each place. Taking an average across all the histories gives a much more accurate, less uncertain, temperature estimate.
“Instead of getting a best guess of what happened in the past along with an uncertainty interval, it allows us to come up with possible climate histories,” Martin explained. “Each history is consistent with the data and our modelling assumptions. We can use the ensemble to start answering interesting questions, such as which year was warmest with what probability. If I have many realisations of the climate, and at a location 2010 was warmest in 95% of my ensemble members, then I have 95% probability 2010 was the warmest.”
As well as showing that recent summers are almost certain to have been hotter than any since 1400, Peter and Martin also looked at why. Or more specifically, they looked at whether we are now seeing a broader range of temperatures as well as a higher average. “We show that recent warm events, including the summer of 2010 in Russia, seem to be consistent with a shift in the average alone,” Martin said. “Yes, temperatures are getting warmer, but at least at the summer average timescales that we consider in the paper there’s no increased variability.”
Because of the proxies’ limitations whole-summer average temperature values are the best that Martin and his fellow Harvard researchers can do, he underlined. “But as a society, we’re interested in week long heat-waves and very hot individual days,” he added. “So the research group here is moving towards thinking about extremes at shorter timescales, to see if the relationship where the average alone increases with the temperatures holds. We can’t yet say anything about week-long heatwaves, but that’s where the future of this work will be.”
Tingley, M., & Huybers, P. (2013). Recent temperature extremes at high northern latitudes unprecedented in the past 600 years Nature, 496 (7444), 201-205 DOI: 10.1038/nature11969