Mike West
Spring 1993
This paper discusses some issues arising in the study of patterns of historical climate change based on geochemical measurements in deep lake sediment cores. The specific climatological focus is on patterns of variability, and possible cycles, in climatic indicators during the past two or three millenia. The statistical issues raised relate to various aspects of statistical calibration of raw data records, data quality and uncertainty propagation through data processing stages, and problems of time series analysis to assess patterns of variation in the geological records over time. Novel time series methods are introduced to address the questions of cyclicality in the data records and to allow for various sources of uncertainty in dating of the records. Dynamic models with cyclical auto-regressive components, and various aspects of their simulation based analyses, are discussed, and some exploratory analyses of currently available palaeoclimatological data are summarised. Other issues discussed include radiocarbon calibration, uncertainties in data timing, missing data, model combination, and the future prospects in the current and related pal\ae oclimatological contexts.
Keywords: Bayesian Spectral Analysis; Climate Change; Cyclical Time Series; Dynamic Linear Models; MCMC; Radiocarbon Dating; Random Observation Times
The manuscript is available in pdf format
The referred paper was published (with discussion) in the 1996 volume Bayesian Statistics 5, (eds. J.O. Berger, J.M. Bernardo, A.P. Dawid and A.F.M. Smith), Oxford University Press, 461-486.
Research partially supported by the NSF under grants DMS-9024793, DMS-9305699 and DMS-9311071.
The author won the international competition for the 1994 Mitchell Prize for ``the Bayesian analysis of a substantive and concrete problem'' based on the work reported in this paper.