Monday, April 8, 2013

Quelccaya icecore and regional temperatures

Prof Lonnie Thompson and colleagues have a new paper in Science Express called "Annually Resolved Ice Core Records of Tropical Climate Variability Over the Past 1800 Years". With annual resolution it becomes worthwhile to compare with measured surface temperatures. I did this with TempLS, using GHCN/ERSST and restricting to an area 1000 km around the location of the Quelccaya site in Peru. The Quelccaya data is here.

This is done partly in response to a thread at Climate Audit, where there was speculation about the possible relationship of the d18O proxy to surface temperature. The standard TempLS analysis is below the step.

The analysis runs from 1900 to 2009. Since Thompson uses the "thermal year" starting for SH in July, I have followed that convention. So what is marked as 2009 is actually Jul 2009-Jun 2010.

Here is a map of all the GHCN/ERSST stations within the radius. ERSST "stations" are the 5x5 lat/lon grid points of the data. The map includes all stations that have reported from 1900-2013, many for only short times. The Quelccaya site is shown with a big blue dot.

Next is a plot of the numbers of stations reporting in each year (for at least 8 months)

Finally, here is the plot of surface temperatures (black) vs Quelccaya d18O. I normalised the d180 data by matching the mean and sd over base 1961-90 to the temperature measure.

Note that the scaling is not to be relied on as I did it by matching. For the first fifty years (with rather small station numbers), there are indications of correlation, but some sranger things recently. The big dip in 1999 in d18O does not seem related to this measure of local temperature. However land temperatures were actually cold, but the 1000 km radius picks up a lot of ocean, which was warm.


  1. I don't think you can assume that ma is a local temperature proxy. Think about how the material for the proxy is deposited: By precipitation of course.

    Certainly it makes no sense to include stations that are "downwind" of the ice core location. (Given the low-frequency nature of the proxy, you could probably just average all of the stations in that latitudinal band.)

  2. Carrick,
    Yes, I don't assume that. That was why I wanted to use a regional temperature; some people were just looking at nearby individual stations.

    It's hard to know which way the wind blows. One often thinks from the west, which I guess is your basis for the latitude bands. But the winds from the Pacific come over coastal desert. I'm no expert here, and this site is very high, but there is the possibility that the primary precipitation source is from the Amazon side.

    I did look at 1500 km too, it was much the same.

    Incidebtally, further spread gets dominated by Pacific SST, for better or worse.

  3. According to climatology, primary precipitation is from Amazon side.

  4. At this latitude, the dominant wind direction is probably from the east, though I admit I haven't looked at the data. 1000-km radius is pretty localized for climate, especially for a phenomenon that is probably integrating over a much larger area than that.

    I say "latitudinal band" because as I understand it there is a lot of temporal smearing in the nature of the proxy and in how it's measured. (So you might as well enjoy the spatial smoothing associated with a larger area of integration.)