Progress in reconstructing climate in recent millennia

What makes science different from politics?

That’s not the start of a joke, but it is a good jumping off point for a discussion of the latest publication on paleo-reconstructions of the last couple of millennia. As has been relatively widely reported, Mike Mann and colleagues (including Ray Bradley and Malcolm Hughes) have a new paper out in PNAS with an update of their previous work. And this is where the question posed above comes in: the difference is that with time scientists can actually make progress on problems, they don’t just get stuck in an endless back and forth of the same talking points.

We discussed what would be required in an update of these millennial reconstructions a few months back and the main principles remain true now. You need proxies that are a) well-dated, b) have some fidelity to a climate variable of interest, c) have been calibrated to those variable(s), d) that are then composited together somehow, and e) that the composite has been validated against the instrumental record.

The number of well-dated proxies used in the latest paper is significantly greater than what was available a decade ago: 1209 back to 1800; 460 back to 1600; 59 back to 1000 AD; 36 back to 500 AD and 19 back to 1 BC (all data and code is available here). This is compared with 400 or so in MBH99, of which only 14 went back to 1000 AD. The increase in data availability is a pretty remarkable testament to the increased attention that the paleo-community has started to pay to the recent past – in part, no doubt, because of the higher profile this kind of reconstruction has achieved. The individual data-gatherers involved should be applauded by all.

The increase in proxy records allows a whole bunch of new things to be done. First off, the importance of tree rings can be tested more robustly. With the original MBH98 proxies, there was only enough other data to go back to 1760 if you left out the tree rings. The match was pretty good over multi-decadal periods, but the interannual variability was much larger without tree-rings. Now though, the Northern hemisphere land temperature reconstructions without tree rings can go back to 1500 AD or 1000 AD depending on which of two methodologies are used. For the NH land and ocean target, it’s even possible to get a coherent non-tree ring reconstruction back to 700 AD! As before, there are some differences (notably in the 17th Century where the tree rings indicate colder temperatures), but the recent warming is anomalous regardless.

Secondly, you can screen records and pick targets more finely: do you want only records that match local temperatures? Done. You want to get a handle on global and southern hemisphere means as well as the northern hemisphere? Done. Other screens could easily be implemented.

The two methodologies used themselves span the range of different approaches that people have used. ‘Composite and scale’ (CPS) is perhaps the simplest method – it is basically an average of all the temperature proxies scaled to the target time series. The other method is denoted ‘Error in variables’ (EIV) in this paper, but is really a simplified application of the RegEM climate field reconstruction method used in a couple of more recent papers. It is essentially a fancy multiple regression to the target time series that can incorporate non-local proxies as well. The point of using two methods is to demonstrate what is, and what is not, robust, and to give an idea of what the structural uncertainty in these estimates is – something not easily calculated using standard statistics. That uncertainty is clearly larger as you go back in time, and larger still for the southern hemisphere.

Other improvements over previous work are that more proxy data sets go past 1980, and so calibration up to 1995 is possible. That allows more of the recent trends to feed into the calibration and highlights the so-called divergence problem in some (but not all) recent tree-ring records. That divergence is significantly lessened without tree-rings or using the EIV method.

Figure: Spaghetti plot of the new reconstructions over a) 1800 and b) 1000 years

along with selected older ones for comparison.

Page 1 of 2 | Next page