TY - JOUR
T1 - Specification and estimation of the transfer function in dendroclimatological reconstructions
AU - Auffhammer, Maximilian
AU - Li, Bo
AU - Wright, Brian
AU - Yoo, Seung Jick
N1 - We thank the Giannini Foundation for generous funding of this research. All remaining errors in the mansucript are soleley ours.
Auffhammer thanks the Giannini Foundation for support. Lis research was partially supported by the NSF Grant DMS-1007686. The authors thank the Guest Editor and two anonymous referees for their helpful comments and suggestions.
PY - 2015/3
Y1 - 2015/3
N2 - We identify two issues with the reverse regression approach as implemented in several classic reconstructions of past climate fluctuations from dendroclimatologcical data series. First, instead of estimating the causal relationship between the proxy, which is measured with significant error, as function of climate and formally inverting the relationship, most papers estimate the inverted relationship directly. This leads to biased coefficients and reconstructions with artificially low variance. Second, we show that inversion of the relationship is often done incorrectly when the underlying causal relationship is dynamic in nature. We show analytically as well as using Monte Carlo experiments and actual tree ring data, that the reverse regression method results in biased coefficients, reconstructions with artificially low variance and overly smooth reconstructions. We further demonstrate that correct application of the inverse regression method is preferred. However, if the measurement error in the tree ring index is significant, neither method provides reliable reconstructions.
AB - We identify two issues with the reverse regression approach as implemented in several classic reconstructions of past climate fluctuations from dendroclimatologcical data series. First, instead of estimating the causal relationship between the proxy, which is measured with significant error, as function of climate and formally inverting the relationship, most papers estimate the inverted relationship directly. This leads to biased coefficients and reconstructions with artificially low variance. Second, we show that inversion of the relationship is often done incorrectly when the underlying causal relationship is dynamic in nature. We show analytically as well as using Monte Carlo experiments and actual tree ring data, that the reverse regression method results in biased coefficients, reconstructions with artificially low variance and overly smooth reconstructions. We further demonstrate that correct application of the inverse regression method is preferred. However, if the measurement error in the tree ring index is significant, neither method provides reliable reconstructions.
KW - Climate change
KW - Climate variability
KW - Inversion
KW - Paleoclimatic data
KW - Reconstruction method
UR - http://www.scopus.com/inward/record.url?scp=84939878624&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84939878624&partnerID=8YFLogxK
U2 - 10.1007/s10651-014-0291-6
DO - 10.1007/s10651-014-0291-6
M3 - Article
AN - SCOPUS:84939878624
SN - 1352-8505
VL - 22
SP - 105
EP - 126
JO - Environmental and Ecological Statistics
JF - Environmental and Ecological Statistics
IS - 1
ER -