80
Chapter 5: Interpreting High-Resolution Proxy Climate Data
Most studies have been concentrated in temperate to mid-Iatitude regions
where there is a wealth of species which exhibit clear annual rings. Valuable
information has, however, also been obtained from higher latitudes and from
certain subtropical species.
In harsher, more marginal, regions, often near the edges of the ecological range of tree species, the seasonal limitation on growth may become so
strong that radial growth may not occur around the entire tree. This will
produce partially "missing" rings which can only be detected by comparison
with sampies from other parts of the tree or other trees. Similarly, severe and
abrupt weather conditions can sometimes cause cell growth to slow down or
even stop during the growing season, resulting in what appear to be "double"
rings in particular years. Again, these are detected by the comparison of the
ring series from many trees. This meticulous "cross dating" of replicate sampIes is an extremely important aspect of dendrochronology and distinguishes
it from many other so-called high-resolution proxy studies. It is important
to appreciate the importance of this point. Other palaeoclimate data sources
are capable of providing annual data e.g. lake varves (Petterson et al., 1993),
ice cores (Thompson, 1990) and corals (Dunbar and Cole, 1993), but tree-ring
data are dated, absolutely, to the year, by reference to a continuous, replicated, chronology, stretching unbroken to modern times. The comparison
of temporally overlapping ring-width series enables mean chronologies, thousands of years long, to be assembled from many individual-tree series that
may typically have only 200-300 rings (e.g. Baillie 1982). Merely counting
back rings (or layers in ice cores or <;orals) from the present is not sufficient
to ensure that rings have not been overlooked or are not missing from the
sampie series. Only crossdating ensures the absolute timescale.
5.4 Chronology Confidence
5.4.1 Chronology Signal
The replication of data inherent in constructing an average chronology for a
si te or region is another axiom of dendrochronology. It is only by comparing replicate series that one can judge the strength of common variability
(and hence forcing) within any proxy series, and it is only by averaging data
series that non-common variability, "noise", can be reduced or eliminated.
The variability in common between contemporaneous data series represents
an empirical signal - a purely statistical measure of common growth forcing
among a group of trees. This can be measured using Analysis of Variance
techniques or simply as the mean correlation coefficient (RBAR) of all replicate comparisons (in fact it is possible to calculate separate quantities representing the strength of common growth forcing within and between trees
at a site, but, for simplicity, we will consider only the one between-tree value
here. For further details see Fritts, 1976; Wigley et al., 1984; Briffa and
Chapter 5: Interpreting High-Resolution Proxy Climate Data
Most studies have been concentrated in temperate to mid-Iatitude regions
where there is a wealth of species which exhibit clear annual rings. Valuable
information has, however, also been obtained from higher latitudes and from
certain subtropical species.
In harsher, more marginal, regions, often near the edges of the ecological range of tree species, the seasonal limitation on growth may become so
strong that radial growth may not occur around the entire tree. This will
produce partially "missing" rings which can only be detected by comparison
with sampies from other parts of the tree or other trees. Similarly, severe and
abrupt weather conditions can sometimes cause cell growth to slow down or
even stop during the growing season, resulting in what appear to be "double"
rings in particular years. Again, these are detected by the comparison of the
ring series from many trees. This meticulous "cross dating" of replicate sampIes is an extremely important aspect of dendrochronology and distinguishes
it from many other so-called high-resolution proxy studies. It is important
to appreciate the importance of this point. Other palaeoclimate data sources
are capable of providing annual data e.g. lake varves (Petterson et al., 1993),
ice cores (Thompson, 1990) and corals (Dunbar and Cole, 1993), but tree-ring
data are dated, absolutely, to the year, by reference to a continuous, replicated, chronology, stretching unbroken to modern times. The comparison
of temporally overlapping ring-width series enables mean chronologies, thousands of years long, to be assembled from many individual-tree series that
may typically have only 200-300 rings (e.g. Baillie 1982). Merely counting
back rings (or layers in ice cores or <;orals) from the present is not sufficient
to ensure that rings have not been overlooked or are not missing from the
sampie series. Only crossdating ensures the absolute timescale.
5.4 Chronology Confidence
5.4.1 Chronology Signal
The replication of data inherent in constructing an average chronology for a
si te or region is another axiom of dendrochronology. It is only by comparing replicate series that one can judge the strength of common variability
(and hence forcing) within any proxy series, and it is only by averaging data
series that non-common variability, "noise", can be reduced or eliminated.
The variability in common between contemporaneous data series represents
an empirical signal - a purely statistical measure of common growth forcing
among a group of trees. This can be measured using Analysis of Variance
techniques or simply as the mean correlation coefficient (RBAR) of all replicate comparisons (in fact it is possible to calculate separate quantities representing the strength of common growth forcing within and between trees
at a site, but, for simplicity, we will consider only the one between-tree value
here. For further details see Fritts, 1976; Wigley et al., 1984; Briffa and