Modeling paleoclimates

Data describe, models explain. Both are required to document and understand the past variations of Earth’s climate, and to help address the present problem of assessing climate change that may result from human activities. Models (for the most part conceptual as opposed to numerical) have long been applied for understanding climate variations during the Quaternary. Indeed, over a century ago, in a set of papers that contributed to the foundation of scientific method in the geosciences (multiple working hypotheses), T.C. Chamberlin (1897, 1899) provided a comprehensive conceptual model for explaining long-term climatic changes that is remarkably modern in some of its elements. What is regarded as a “climate model” today is generally a computerized numerical representation of the physical processes involved in the climate system, but conceptual models still play an important role in paleoclimate research. Whenever any kind of paleoclimatic data is interpreted, either quantitatively or qualitatively, some kind of model is invoked. Paleoclimatic data (of the kinds reviewed in this volume) and climate models play a complimentary role in understanding climate change. The data record how climate has changed, but data alone cannot provide an unambiguous explanation of why a particular climate state occurred or changed. This situation arises because most climatic variations recorded geologically have multiple, hierarchial causes (e.g. there is more than one way to create drought in a region) and because environmental subsystems display generally nonlinear responses to climate. Consequently, multiple cause-and-effect pathways can produce the same response in a paleoclimatic indicator. This indeterminacy of the “climate signal” is mitigated somewhat by considering networks of paleoclimatic data and by examining multiple indicators at individual sites, but such “multi-proxy mapping” cannot in itself eliminate the indeterminacy. Models based on physical principles (or widely accepted empirical representations of those physical principles) do have the potential to provide mechanistic explanations of past climatic variations, provided they are known to work, are applied in an appropriately designed experiment, and (perhaps most importantly) explicitly account for all of the components of the climate system that are involved in a particular climate change. Although comprehensive models of the climate system and its individual components (the atmosphere, oceans, biosphere, hydrosphere, and cryosphere) are evolving rapidly, the development of a comprehensive model that can simulate the temporal and spatial variations of climate on both global and local scales, using as input only the records of the external controls of climate (i.e. an “Earth-system model”), is perhaps a decade away. The indeterminacy of the data and the present limitations of the models thus dictate a synergistic approach for understanding climate variations that relies on integrating paleodata with paleoclimate model simulations. In this chapter we review the process of climate-system modeling and present a taxonomy of the models recently applied in the study of Quaternary climate change and variation. We also briefly trace the development of climate modeling since the 1965 INQUA volume and its companions were published. A synopsis of climate-modeling results for North America is provided, and we conclude with a discussion of some the emerging issues in the application of models for understanding climatic variations.