Searcharxiv⌕ Search

arXiv subjects

Lasse Holmström

Publications and source records attributed to Lasse Holmström.

3 recordsLinked to original sources
arXiv2015

A Bayesian spatiotemporal model for reconstructing climate from multiple pollen records

Holocene (the last 12,000 years) temperature variation, including the transition out of the last Ice Age to a warmer climate, is reconstructed at multiple locations in southern Finland, Sweden and Estonia based on pollen fossil data from lake sediment cores. A novel Bayesian statistical approach is proposed that allows the reconstructed temperature histories to interact through shared environmental response parameters and spatial dependence. The prior distribution for past temperatures is partially based on numerical climate simulation. The features in the reconstructions are consistent with the quantitative climate reconstructions based on more commonly used reconstruction techniques. The results suggest that the novel spatio-temporal approach can provide quantitative reconstructions that are smoother, less uncertain and generally more realistic than the site-specific individual reconstructions.

stat.AP↗
arXiv2013

Finding a consensus on credible features among several paleoclimate reconstructions

We propose a method to merge several paleoclimate time series into one that exhibits a consensus on the features of the individual times series. The paleoclimate time series can be noisy, nonuniformly sampled and the dates at which the paleoclimate is reconstructed can have errors. Bayesian inference is used to model the various sources of uncertainty and smoothing of the posterior distribution of the consensus is used to capture its credible features in different time scales. The technique is demonstrated by analyzing a collection of six Holocene temperature reconstructions from Finnish Lapland based on various biological proxies. Although the paper focuses on paleoclimate time series, the proposed method can be applied in other contexts where one seeks to infer features that are jointly supported by an ensemble of irregularly sampled noisy time series.

stat.AP↗