COMPARISONS OF PALEOENVIRONMENTAL OBSERVATIONS AND PALEOCLIMATIC SIMULATIONS: PRINCIPAL RESULTS AND STRATEGIES FOR THE NEXT ITERATION

The records of past climatic variations preserved by paleoclimatic indicators (or “proxy data”) document the behavior of the climate system over time and space, and also provide the only means for testing climate models under configurations of the climate system different from those at present. The first iterations of data-model comparisons, represented by PMIP and earlier less comprehensive projects, demonstrated convincingly that (a) the general approach of simulating regional variations of climate using global models is a sound one, (b) the general trends of climate since the last glacial maximum are well accounted for by a relatively simple conceptual model that invokes changes in insolation, ice volume, and atmospheric composition as the primary drivers, and (c) regional discordances between simulations and observations are most likely due to the absences of feedback and interactions among various components of the climate system not well represented in those earlier models.

In the next iteration of data-model comparison, in particular PMIP Phase II in which fully coupled ocean-atmosphere-vegetation models will be used, the focus will shift from evaluating the ability of models to simulate the mean state of climate at key times, to their ability to simulate variability changes as well as abrupt changes. The optimal evaluation of such models would involve global networks of well-dated, multi-dimensional, multi-proxy records, of sufficient spatial density to resolve regional patterns, and temporal frequency to diagnose interdecadal and interannual modes of climatic variability. However, such networks do not exist, nor are they likely to anytime soon. Consequently, we envision an alternative strategy with two elements: (1) the use of multiannual-to-decadal resolution records from lake and marine sediments and ice cores, which are available over growing networks (while employing also annually resolved records where (and when) available to characterize interannual variability), (2) the reanalysis of the lower temporal resolution but global networks of paleoecological and hydrological data in light of observations that such records necessarily contain the influence of climate variability, which may be retrieved through the application of suitable dynamic models of vegetation and hydrology.