Investigating the dynamic behavior of the total Earth System remains a grand challenge for the scientific community. It is motivated by our limited knowledge about the consequences of largescale perturbations of the Earth System by human activities, such as fossilfuel combustion or the fragmentation of terrestrial vegetation cover. Will the system be resilient with respect to such disturbances, or will it be driven towards qualitatively new modes of planetary operation?
This question cannot be answered, however, without prior analysis of how the unperturbed Earth System behaves and evolves in the absence of human influence. Such an analysis should, for example, provide answers for questions concerning the amplification of Milankovitch forcing to glaciation episodes or the mechanisms behind the Dansgaard-Oeschger oscillations. The key is to be able to resolve the ciritical drivers, triggers and responses within the system such as orbital variations, ocean and atmospheric circulation, land surface cover, and marine primary productivity, among others. While present individual atmospheric, marine, and terrestrial models are sensitive to variations in these, the numerical formulations are too complex (slow) to run fully coupled complete models for extended model times on present machines. It is thus helpful to use reduced-form models in order to run fully coupled earth system models. We call these Earth System Models of Intermediate Complexity (EMIC), and convened a workshop to bring together some of the key investigators internationally who are beginning to develop techniques for such model coupling and application.
Earth System analysis primarily relies on a hierarchy of simulation models. Depending on the nature of questions asked and the pertinent time scales, there are, on the one extreme, zero-dimensional tutorial or conceptual models like those in the "Daisyworld" family. On the other extreme, three-dimensional comprehensive models, e.g. coupling atmospheric and oceanic circulation with explicit geography and high spatio-temporal resolution, are employed. The former, simple heuristic models can be perceived as hypothesis-generating machines that allow to explore major geosphere-biosphere feedbacks, like insolation-albedo-vegetation dynamics. The latter, comprehensive models are capable of zooming into brief time slices of Earth history, keeping the slow systems components virtually constant.
Hence, there is a need for "Earth-System Models of Intermediate Complexity" (EMIC's) which address two fundamental requirements: first, they must be simple enough to permit numerical integration over many millennia and, second, they must be complex enough to yield a reasonably realistic picture of the Earth System by the inclusion of more interactions than are possible using the comprehensive models.
The workshop aimed at bringing together scientists in order to review the present range of EMIC's. We did conduct a 'model competition', but rather an exchange of ideas about the various approaches and their basic differences. The key theme of the meeting was: "How far can we reduce a comprehensive Earth-System model without losing the crucial feedbacks that govern the overall systems dynamics?" Conversely, "How robust (complex) must subsytem models be before coupling and relaxation of prescribed boundary conditions increases modelled reliability of coupled system responses rather than decreasing it?" The answer depends on the spatio-temporal scales under consideration. The workshop focused on two- and three-dimensional global models spanning time windows of several thousand years and longer.
A workshop report is being written, and a GAIM newsletter article will appear in the Summer issue.