Caspar M. Ammann is a Scientist II in the Climate and Global Dynamics Division of NCAR studying past and present climate changes. The primary focus of the research is on natural climate variability and change over the past centuries and millennia. The goal is to understand which part of the variability was forced by external forcing -- such as explosive volcanism, solar activity variations, or human changes to the landsurface and atmosphere -- and what variations were simply due to internal variations of the climate system. Solid understanding of variability and more systematic cause-and-effect between forcing and regional climate are crucial for making regional predictions.

Regional Climate Change: Seamless link from Past to Present and to the Future

Global scale climate change of the past few decades has been detected and attributed to anthropogenic forcing. Current climate sysetm models are capable of reproducing most of the general features, but how well are they ready to project or predict regional climate over the coming decades and beyond? Two fundamental issues are limiting this big next step in climate change research:

• First, regional climate generally exhibits much more variability than on the global-scale, yet records are often too short to confidently identify the source of the variations. Much longer records are required.

• Second, emphasis in climate model development has mostly been put on reproducing the mean annual cycle, but the next challenge is to faithfully reproduce the range of variability, and most importantly, the systematic response at the regional scale to the external forcing. A regional response is not only driven by direct effects of the change in forcing but it is just as much affected by changes in circulation and large-scale modes of variability.

Therefore, to identify what part of regional climate variability is driven by external forcing, records need to be extended into the past beyond the instrumental period so that regional changes can be put into, and be understood, in a seamless perspective from past to present. Based on such understanding one can approach future predictions with much more solid expectations.

Process-based combination of Data and Models

To sucessfully identify forced changes in the past, knowledge on the forcings as well as the actual climate response are of course of central importance. Therefore activities for improving our understanding of last millennium climate have focused on many aspects along the chain of key elements:

Past external forcing: novel ways of reconstructing and characterizing external forcing. For example the largest collection of ice core records is used to reconstruct volcanic forcing, and new statistical methods are employed to provide the forcing in a novel probabilistic way. Using this formulation, scenarios of possible future forcing can be drawn that are statistically sound.

Climate Reconstructions: Using climate model output, various methods of climate reconstructions can be tested and critical assumptions and steps can be identified. New methods are indevelopment that are better suited to guard against earlier problems and that are capable of dealing with the diversity of data that can be brought to bare to make sense of past climate and its spatio-temporal aspects.

Modeling: Climate Model studies of response to external forcing: transient climate model simulations with individual or a full suite of external forcings are used to test the models ability to reproduce the spatial and temporal aspects of response in global, hemispheric, regional and local climate that have been identified in real world records. The model environment then allows for a solid investigation of the actual dynamical processes that are responsible to generate the response over different spatial and temporal scales.

Towards Data Assimilation: Finally, the crown jewel of paleoclimatology is the integration of dynamical processes provided by modeling frameworks with the variability and response structures seen in the proxy data. Although individual proxies are inherently noisy, when put into a large scale context, a more consistent and dynamically sound picture can recognized. Independent records can be used to test our understanding of changes and their driving processes in a quantitative fashion for a seamless description of climate from the past to the present, and thus provide a solid foundation for studying the future.