NOAA/WDS Paleoclimatology - Southern British Columbia 400 Year Summer Temperature Reconstructions

STUDY NOTES: Summer (May-June-July-August) temperature reconstructions for southern British Columbia, based on tree-ring width (RW), maximum density (MXD), and Blue Intensity (BI) measurements from 7 sites.
ABSTRACT SUPPLIED BY ORIGINATOR: Maximum latewood density (MXD) is a strong proxy of summer temperatures. Despite this, there is a paucity of long MXD chronologies in the Northern Hemisphere, which limits large-scale tree-ring-based reconstructions of past temperature which are dominated by ring-width (RW) data - a weaker temperature proxy at inter-annual time-scales. This paucity likely results from the relative expense of measuring MXD and the lack of laboratories with the facilities to measure it. Herein, we test the ability of a relatively new, less expensive, tree-ring parameter, Blue Intensity (BI), to act as a surrogate parameter for MXD. BI was measured on Engelmann spruce samples from British Columbia where MXD had previously been measured to allow direct comparison between the two parameters. Signal strength analyses indicate that 8 MXD series were needed to acquire a robust mean chronology while BI needed 14. Utilising different detrending methods and parameter choices (RW + MXD vs RW + BI), a suite of reconstruction variants was developed. The explained variance from the regression modelling (1901-1995) of May-August maximum temperatures ranged from 52% to 55%. Validation tests over the earlier 1870-1900 period could not statistically distinguish between the different variants, although spectral analysis identified more lower frequency information extant in the MXD-based reconstructions - although this result was sensitive to the detrending method used. Ultimately, despite the MXD-based reconstruction explaining slightly more of the climatic variance, statistically robust reconstructions of past summer temperatures were also derived using BI. These results suggest that there is great potential in utilising BI for dendroclimatology in place of MXD data. However, more experimentation is needed to understand (1) how well BI can capture centennial and lower frequency information and (2) what biases may result from wood discolouration, either from species showing a distinct heartwood/sapwood boundary or from partly decayed sub-fossil samples.