A commenter at Real Climate asked the following interesting question about the Urals/Yamal area reported in Briffa et al 2013:
So, for this region, what data do you wish you had, that could be plausibly gotten (i.e, not like 1000-year-old measured temperature records), that you don’t have? And why?
Jim Bouldin of RC described the question as an “excellent” one and on this point, RC and I are in agreement. It’s one that I have some ideas on.
On the surface, the Polar Urals are an almost ideal location for temperature-sensitive dendro. They are high latitude and high altitude. Within a distance of a kilometer or so, one can go from closed forest to alpine tundra. Subfossil trees are plentiful. Despite this plenty, Briffa et al 2013 use a very small dataset of subfossil trees, the limited coverage being exacerbated by their exclusion of 21 of 73 trees as being “root collars”. Due to their very small sample, CRU ends up with only two trees in part of the 11th century – a critical period in the modern-medieval comparison. There is an obvious need for more data.
In addition, there is an urgent need for accurate metadata, especially on the altitude of indovidual samples. Altitude is an important factor in local tree growth (as one can see by the sharp altitudinal gradient to the treeline), but the CRU metadata does not include the altitude of individual trees. Briffa et al 2013 does not even provide accurate information on the location of the sites used in its Polar Urals chronology. Altitude changes of samples are an important potential inhomogeneity that can only be addressed through accurate metadata, an issue apparently neglected in Briffa et al 2013 which splices living tree data from sites whose altitude appears to be materially lower than the reported altitude of the subfossil samples.
One relatively easy way of improving the situation would be to do extremely detailed map and measurements along the lines of what a geologist would do at a geological outcrop. The program described below may seem very detailed by dendro standards, but I’m convinced that it could be done at a reasonable cost and would result in an exemplary dendro data set.
The first step would be to mark out an altitudinal transect going from the alpine tundra to the closed forest. From Google Earth, it looks to me like this would be less than a km (say 800 m) A transect of width of 20 meters would be a workable width, giving a transect area of about 1.6 hectares, a practical area for detailed mapping. Geologists would mark the boundaries of the transect, perhaps with small cairns of stones at key points.
The next step would be to do a high-resolution map (say 1 cm map to 1 meter) on which the location of EVERY dead and living tree in the transect would be recorded. Information on each tree (height, etc…) should be recorded. Even if there are 5000 living trees and 1200 dead trees, this is a practical number. Geologists do this wort of detailed mapping all the time.
Next, the mappers could cut a cross-section from every dead tree and (say) 15% of the living trees, measure ring widths of the cores back at their lab and crossdate the cores using standard dendro techniques. Some proportion of the cores would not be crossdatable (especially those with relatively few rings), but suppose that 80% were crossdatable.
One would then end up with a data base of more than 800 crossdated subfossil cores and more than 500 crossdated living trees, each with exact coordinates and altitude. From our experience at Almagre, we know that a single dendro can collect tens of cores per working day and that the measurement of ring widths is automated and can be done relatively inexpensively. If “root collars” are the form in which subfossil data is available for older stunmps (e.g. medieval), then it would be prudent to sample living trees and standing dead trees at root collar as well as chest level, to facilitate analysis, rather than simply rejecting root collar data. While bending over may be a slight imposition on field dendros, geologists regularly bend over to examine rocks and I’m sure that dendros could be found to take cores at root collar as well as chest level. If not, I’d suggest that some field geologists be trained to do root collar cores.
I realize that this is an optimistic proposal. CRU dendros prefer to stay in East Anglia. Indeed, to my knowledge, no CRU dendro has ever even been to Yamal or Polar Urals. But there are capable Russian dendros who might be encouraged to take on this sort of program. From this sort of database, one could do real analysis, without having to listen to tiresome CRU whinging about poor replication. If one were lucky, one might even be able to accurately measure changes in treeline through the medieval period and Little Ice Age and modern warming, thereby obtaining a proxy that would be convincing to all parties to the debate. If the transect were properly marked, dendros could return to it in the future and definitively measure changes.
Optimistic or not, the collection of a dataset of over 1000-1500 crossdated cores from one (or two) well-located altitudinal transects at Polar Urals is one answer to the RC question.