Forward modeling of paleoclimate proxies

The object of paleoclimatology is to infer of climate changed in the past from a variety of geological archives. Typically, inferences are made statistically i.e. in an inverse way. A promising (but difficult) approach is to look at things the other way, and deterministically model physical, geochemical and biological processes giving rise to paleoclimate proxies. This approach is called "forward modeling".

Modeling errors in ordinate
It is well-known that many processes influence the formation of a geological object. Most of them are unrelated to climate and some can therefore be treated stochastically from a climatic point of view. Further, each tropical proxy class (coral, speleothem, sediment, tree-ring, ice core) has its own mechanisms of genesis that can be deterministically modeled to some extent. The ultimate goal of the approach is to transform “pseudoproxy” studies [e.g. Mann & Rutherford, GRL, 2002], which are currently the main testing ground for multiproxy climate reconstructions, into a closer analog for real proxy networks.

Modeling errors in abscissa
In addition, climate proxies pose a unique challenge to timeseries analysis. The vast literature on the topic, mostly derived from electrical engineering and financial mathematics, typically assumes that the time axis is exactly known, a crucial assumption that is violated in most paleoclimate studies. This underlines the importance of developing quantitative methods that adequately treat this “uncertainty in abscissa”.

This (incipient) work is at the interface between biogeochemistry and climatology.