Looking beyond the mean in tree-ring chronologies

Abstract:

Climate reconstructions based on tree-ring variability at an individual site are typically based on measurements from 20-100 trees. In order to increase the signal-to-noise ratio, reconstructions are typically based on the mean tree-ring growth across all timeseries. However, there is the possibility that individual trees respond somewhat differently to growing conditions and that additional information exists within a given collection of tree-ring growth records beyond the mean. We begin by comparing timeseries of mean detrended tree-ring width variability to timeseries of the standard deviation, and higher moments of tree-ring width across individual tree-ring series for stations in the International Tree-Ring Databank. Strong correlations exist between mean tree-ring growth and the standard deviation of tree-ring growth across records, indicating that variability in tree-ring width in a given year is not normally distributed. Non-normally distributed tree ring responses to temperature, for example, implies that the distribution of tree-ring growth may record information about the distribution of temperature, in addition to the the mean growing season temperature. We evaluate the feasibility of reconstructing temperature distributions, with a focus on extreme high temperatures, by comparing the distribution of daily-resolution instrumental observations with the distribution of tree-ring growth records. Any progress herein has the potentially to generally increase the interpretability, and hence value, of tree ring chronologies.