SCUBIDO: transforming multivariate proxy data to quantitative climate reconstructions

Palaeoclimate archives record climate variability on a range of timescales from seasonal changes in the climate system to multi millennial variability. Decision makers working on climate mitigation and adaptation typically work on timescales of one year to a few decades ahead only. This means that in order for palaeoclimate reconstructions to be most relevant for policy decisions, the temporal resolution of reconstructions also need to be at an annual to decadal scale. However, traditional approaches to quantitatively reconstruct climate are often expensive and require a lot of archive material and therefore typically have a multidecadal to centennial resolution.

In this presentation we present a new approach to reconstruct climate on high temporal resolutions (annual to decadal) using micro–X-ray Fluorescence (μXRF) data from lake sediments. We have built a statistical model based on Bayesian inference which transforms the semi-quantitative mXRF data into an annually resolved quantitative climate timeseries with quantified uncertainties. Within this study we have synthesised the mathematical details of this approach into a user-friendly R package (SCUBIDO: Simulating Climate Using Bayesian Inference with Proxy Data Observations) which simplifies the modelling process.

We present the output of SCUBIDO from two annually laminated (varved) lake records, Diss Mere in the UK and Nautajärvi in Finland and reconstruct temperature through the Holocene. The results from these reconstructions show long-term climate amelioration and demonstrates clear decadal and multidecadal climate variability, of which is unable to be observed in previously published reconstructions of a lower temporal resolution. Finally in this presentation we will focus on the last two thousand years, a period of time frequently discussed in climate conversations and show the evolution of climate through time.