Modeling of Past Climates—Perspectives: Beginnings, Present, Future

New ideas for modeling of past climates go hand in hand with new observations, with advances in climate models and earth system models, and with advances in computer capacity and speed. Important first steps in quantitative climate modeling using energy balance models were underway in the early 20th century. Dynamical climate models were first used to study past climates in the 1970s and 1980s, with a focus first on the atmosphere, and then on coupled models of atmosphere and upper ocean. I had the privilege of first working with Warren Washington in 1981 as we used a version of CCM0 to conduct a simulation of the climate of 9000 years ago. I have appreciated his early and continuing support and appreciation for the importance of studies of past climate. Currently, coupled climate model simulations include atmosphere, global ocean, vegetation, cryosphere and carbon cycle components. This rapid development in modeling potential has been greatly facilitated by the rapid increase in computational power. Equally important has been the rapid development of more diverse, accurate and worldwide observations of past environments.

The topics of early and current research on modeling of past climates come from a diverse range of ideas about the mechanisms that might force fundamental changes in climate: changes in greenhouse gases, changes in insolation caused by orbital changes, changes in land-sea distribution, changes in orography, and changes in ocean gateways. Certain fundamental principles of modeling and analysis have been important in the past, are important now, and most likely will continue to be important. These principles are enumerated by way of examples.

Looking toward the future, new observations, improved models and even faster computers are to be expected. There will also be new challenges: making intermodel comparisons and model/data comparisons, including non-linear feedback mechanisms properly, studying the response of climate to combinations of forcing, and simulating variability and abrupt change accurately. Another challenge will be to simulate the environmental records directly (rather than providing only conventional climate variables as output) so that simulated environmental records may be compared directly to observed environmental records. What's past is prologue.