Sara Sjöstedt-de Luna: Modelling varved lake sediment to detect past environment and climate changes - a functional data analysis approach

A statistical method to analyze historical archives such as annually laminated (varved) lake sediments is suggested, with the aim of reconstructing past environment and climate changes. The seasonal pattern of a yearly profile is expected to contain valuable information about past climate, with similar profiles corresponding to similar climatic conditions. A functional data analysis approach is proposed to sort the varves (years) into a set of k homogenous groups. Penalized splines and k-means cluster analysis are tools used in the classification process which takes into account that the yearly profiles are of different lengths. In particular we analyze a varved lake sediment core from the lake Kassjön (N. Sweden). The sediment record consists of ~6400 varves, each varve around 0.5-1 mm thick. Image analysis was used to generate the observed data of yearly profiles (in terms of grey-scale variation). We present some results from the analysis of the Kassjön data.

Per Arnqvist, Sara Sjöstedt-de Luna, Amitava Mukherjee, Department of Mathematics and Mathematical Statistics, Umeå University, Christian Bigler, Ingemar Renberg; Department of Ecology and Environmental Science, Umeå University.