GKSS Research Centre, German Climate Computation Center, Max-Planck-Institute for Meteorologie
Proc. 11th Symposium on global change studies, AMS, Long Beach, CA. In print.
In the past decade considerable progress has been made in understanding and modeling climate on time scales of years to decades. General Circulation Models (GCMs) can simulate variability on time scales up to decades adequately. GCMs have also been used to run time slice experiments simulating past equilibrium states forced by the chemical composition of the atmosphere, astronomical parameters and boundary conditions. Proxy data have been used to set up the boundary conditions, as well as to validate the output of these models. With the increasing speed of computers, climate modelers can now move on to simulating the transient behavior of the climate system over millennia. At the same time increasingly more proxies of climatic variables, namely tree rings, ice cores, laminated sediments and coral reefs, have been detected, and there has been a growing research effort in reconstructing the temporal evolution of the climate from these proxies.
Our project aims at an improvement in understanding the global climate of the last millennium by reaching a synergy between the proxy data collection and the numerical modeling of the ocean-atmosphere system. Apart from conducting free and externally forced GCM experiments, a newly developed method for proxy Data Assimilation Through Upscaling and Nudging (DATUN) will be a central element of our work. This paper outlines our general concepts, and presents examples and first results.
We attempt to obtain a physically-based, spatial and temporal interpolation of paleoclimatic data, and to identify mechanisms that contributed to the past climatic variability. These tasks can be formulated conveniently by means of a state space model. It consist of equations for the temporal evolution of the unobservable climate state, which are given by a coupled atmosphere-ocean GCM, and of observation equations that link these climate states to observables recorded in proxy records. For many applications the observation equations need to be inverted, leading to so-called transfer functions. In some cases a given proxy value may be associated with multiple climate states.