2.5 Climate model evaluation and improvement
A major intersection of PAGES and CLIVAR objectives falls in the area of climate modeling. Whereas PAGES focuses heavily on the development and use of paleoenvironmental reconstructions, and CLIVAR emphasizes the development and implementation of a hierarchy of climate models, both programs share the ultimate goals of climate dynamics: to arrive at a theoretical synthesis of climate change over all scales of variability, and to use this theory to achieve maximum predictive skill. At present, only parts of these ambitious goals have been reached. On the interannual timescale, a synthesis between process models and observations for ENSO has been found, and a considerable skill of prediction has been achieved (Cane et al., 1986, 1994, 1995; Barnett et al., 1993). However, serious gaps exist: three-dimensional models are not yet capable of simulating all of the observed dynamics of ENSO. At the other end of the timescale spectrum, a theory of glacial cycles has been formulated that successfully predicts the periodic components observed in the paleoclimatic record (Berger, 1992; Imbrie et al., 1992, 1993). However, quantitative modeling has not been satisfactory in representing this glacial theory, because uncertainties exist regarding the sensitivity of the climate system (e.g., 100 kyr cycle), and high-resolution archives exhibit much richer dynamics than predicted by the theory.

For the timescales that are of interest to both PAGES and CLIVAR, namely interannual to centennial, an effort is needed to first standardize the interface between climate models and paleoclimate data. Advanced statistical methods in the spatial, temporal, and frequency domains must be employed to extract patterns of change and variability from the observed data. Likewise, model data have to be analyzed using the same tools.		 Second, even more urgent is the establishment of a comprehensive theory for interannual-, decadal-, and century-scale climate variability. This will require a better understanding of atmosphere-upper ocean interaction, the role of sea ice, and the dynamics of thermohaline circulation.

The PAGES/CLIVAR working group on model evaluation and improvement expanded on model-oriented foci of the other working groups, and identified several research foci in which a paleoclimatic perspective would contribute significantly to the goals of CLIVAR. These include:
  • Simulation of "extreme" climate conditions

Climate models that are developed for use in CLIVAR should be able to reproduce climatic conditions significantly unlike those of the past 150 years, and in doing so delineate the limits of the range of variability present in the climate system. For atmospheric general circulation models (AGCMs), the international Paleoclimate Modelling Intercomparison Project (PMIP) is already involving over 18 3-D modeling groups in simulating the quasi steady-state climate of two late Quaternary periods (Table 3):

6 ka:
A period by which the Northern Hemisphere ice sheets were gone, but in which insolation was different from today. This is a test of the sensitivity of AGCMs to orbital changes in radiation. Reconstructions of temperature, hydrology, vegetation, and other parameters indicate that they were significantly different from today (Fig. 11).
21 ka:
(Last Glacial Maximum - LGM): A period characterized by radiation similar to today, but with decreased atmospheric trace-gas concentrations and a large modification of the atmospheric flow due to Northern Hemisphere ice sheets. Simulation of this period tests the sensitivity of the dynamics of these models. LGM conditions around the globe were substantially different from those of today.