Talk by Peter Thejll, DMI – Niels Bohr Institute - University of Copenhagen

Forward this page to a friend Resize Print kalender-ikon Bookmark and Share

Niels Bohr Institute > Calendar > 2011 > Talk by Peter Thejll, DMI

Talk by Peter Thejll, DMI

Detecting non-linearity in climate proxies

To understand past climate reconstruction methods are often applied to proxies collected from a range of locations: tree-rings, varved sediments, ice-cores, coral-rings and so on. Such reconstruction methods can be based on a wide range of statistical and mathematical methods, and can yield e.g. reconstructed local or global-mean temperatures, among several possibilities. Common among the work in this field is often the assumption that the climate and the climate proxies are linearly related. Without knowing the biological and geophysical processes that lead to a climate imprint on various proxies it may be helpful to have tools for testing whether the linearity assumption is warranted.

We present such a tool here - generalized additive modelling (GAM), and show its application to a set of 30 NH temperature proxies, all with annual resolution and of length at least 1000 years. A simple reconstruction of NH mean T is shown, and discussed. The GAM method is quite general and can be applied to different types of reconstruction methods. R software has been published for the GAM method.