Statistical properties of tree ring data

A statistical analysis is made of the sequences of annual tree ring widths and indices. The expected value of growth during any year is shown to be proportional to the standard deviation of the growth, so that the coefficient of variation is a measure of the sensitivity of the growth of a tree. Tree ring data were found to be non-randomly distributed in time. The large values of serial correlation indicated that the non-randomness cannot be attributed entirely to climatic factors. Correlogram and power spectrum analyses applied to a sequence of tree ring indices for a pinyon pine showed that the data were generated by an autoregressive process. The statistical parameter measuring the sensitivity or complacency of growth is used to derive a growth function.