94
Chapter 5: Interpreting High-Resolution Proxy Climate Data
some trees can be extrem<;lly great (e.g. Stahle et al. , 1988; Villalba, 1990;
Lara and Villalba, 1993) and effort is increasingly being directed towards the
development and climatic interpretation of multi-millennial chronologies, perhaps constructed from a combination of living, historical and even sub-fossil
material (e.g. Schweingruber et al. , 1988j Briffa et al. , 1992aj Graybill and
Shiyatov, 1992j Cook et al. , 1992). In such chronologies there must always
be uncertainty regarding thelong-term stability of (non-climate) environmental influences or differing climate sensitivity due to inhomogeneity in the
site characteristics of the sam pIes. Attempts to reconstruct long timescale
climate change (centuries to millennia) using tree-ring data should focus attention on these problems, as weH as on the importance of standardization
and the recognition that the reconstructions may represent extrapolations
beyond the range of the calibration (and verification) data.
N otwithstanding these difficulties, the construction of millennial treering chronologies offers the exciting prospect of reconstructing inter-annual,
decadal, and century timescale variability, and the potential for comparing
century-to-millennial climate inferences with other less weH resolved proxy
evidence such as that provided by palynology and glaciology.
5.8 Conclusions
• When interpreting proxy data it is necessary to appreciate the difference
between empirical and theoretical signal.
• Empirical signal is a statistical property of the data and varies according
to how they are processed. Its expression is dependent on replication.
• Theoretical signal needs to be established with care and can sometimes
be enhanced or masked by chronology processing methods and by other
non-climate forcing.
• Theoretical signal should be demonstrated and quantified, e.g. through
regression. It may be limited to a particular "season".
• Regression skill should be assessed realisticaHy, ideaHy through the use of
independent validation, the assumptions and limitations of which should
be recognised.
• Regardless of regression skill, remember that "standardization" of the
primary data may restrict the timescales represented in the reconstruction.
• Always remember the underlying assumption of uniformitarianism in the
response of data to climate forcing. This may be hard or impossible to
prove.
Chapter 5: Interpreting High-Resolution Proxy Climate Data
some trees can be extrem<;lly great (e.g. Stahle et al. , 1988; Villalba, 1990;
Lara and Villalba, 1993) and effort is increasingly being directed towards the
development and climatic interpretation of multi-millennial chronologies, perhaps constructed from a combination of living, historical and even sub-fossil
material (e.g. Schweingruber et al. , 1988j Briffa et al. , 1992aj Graybill and
Shiyatov, 1992j Cook et al. , 1992). In such chronologies there must always
be uncertainty regarding thelong-term stability of (non-climate) environmental influences or differing climate sensitivity due to inhomogeneity in the
site characteristics of the sam pIes. Attempts to reconstruct long timescale
climate change (centuries to millennia) using tree-ring data should focus attention on these problems, as weH as on the importance of standardization
and the recognition that the reconstructions may represent extrapolations
beyond the range of the calibration (and verification) data.
N otwithstanding these difficulties, the construction of millennial treering chronologies offers the exciting prospect of reconstructing inter-annual,
decadal, and century timescale variability, and the potential for comparing
century-to-millennial climate inferences with other less weH resolved proxy
evidence such as that provided by palynology and glaciology.
5.8 Conclusions
• When interpreting proxy data it is necessary to appreciate the difference
between empirical and theoretical signal.
• Empirical signal is a statistical property of the data and varies according
to how they are processed. Its expression is dependent on replication.
• Theoretical signal needs to be established with care and can sometimes
be enhanced or masked by chronology processing methods and by other
non-climate forcing.
• Theoretical signal should be demonstrated and quantified, e.g. through
regression. It may be limited to a particular "season".
• Regression skill should be assessed realisticaHy, ideaHy through the use of
independent validation, the assumptions and limitations of which should
be recognised.
• Regardless of regression skill, remember that "standardization" of the
primary data may restrict the timescales represented in the reconstruction.
• Always remember the underlying assumption of uniformitarianism in the
response of data to climate forcing. This may be hard or impossible to
prove.