STATUS OF THE PALEOCLIMATE MODELING INTERCOMPARISON PROJECT (PMIP)

S. Joussaume¹ and K. E. Taylor²

¹Laboratoire de Modélisation du Climat et de l´Environnement
D.S.M. / Orme des Merisiers / Bat. 709
C.E. Saclay
91191 Gif-sur-Yvette cedex
also at LODYC, CNRS / Université PM Curie / Orstom
France

²Lawrence Livermore National Laboratory
P. O. Box 808, L-264
Livermore, CA 94551

Table of Contents

Abstract

1. Introduction
2. Background
3. Experiment Design
4. Current Status and Plans

References

Abstract 

Partly inspired by AMIP, the Paleoclimate Modeling Intercomparison Project (PMIP) was initiated in order to coordinate and encourage the systematic study of atmospheric general circulation models (AGCMs) and to assess their ability to simulate large changes of climate such as those that occurred in the distant past. Project goals include identifying common responses of AGCMs to imposed paleoclimate "boundary conditions," understanding the differences in model responses, comparing model results with paleoclimate data, and providing AGCM results for use in helping in the analysis and interpretation of paleoclimate data. PMIP is initially focussing on the mid-Holocene (6,000 years Before Present) and the last glacial maximum (21,000 yr BP) because climatic conditions were remarkably different at those times and because relatively large amounts of paleoclimate data exist for these periods. The major "forcing" factors are also relatively well known at these times. Some of the paleoclimate features simulated by models in previous studies seem consistent with paleoclimatic data, but others do not. One of the goals of PMIP is to determine which results are model-dependent. The PMIP experiments are limited to studying the equilibrium response of the atmosphere (and such surface characteristics as snow cover) to changes in boundary conditions (e.g., insolation, ice-sheet distribution, CO2 concentration, etc.)

PMIP has been endorsed by both IGBP/PAGES and WCRP/WGNE, and more than fifteen modeling groups are participating. Several of these groups have completed one or more of the PMIP simulations. Model output will be archived at the Program for Climate Model Diagnosis and Intercomparison (PCMDI) in a structure similar to the AMIP standard output. A workshop involving a representative of each of the PMIP modeling groups is planned for the Fall of 1995 in which results from the PMIP simulations will be shared and subprojects focussing on specific issues will be formed.

1. Introduction 

For more than two decades now atmospheric general circulation models (GCM's) have been used to study paleoclimates. One purpose of these studies has been to attempt to determine why climates of the past were so different from today. This essentially academic interest in paleoclimates has been augmented recently by a new imperative to evaluate whether climate models can be relied on to predict future climate change. If we can demonstrate that climate models successfully simulate past climatic conditions, then we should have more confidence in their predictions of future climates. This growing interest in paleoclimate simulations has led to the initiation of the Paleoclimate Modeling Intercomparison Project (PMIP). PMIP has been established to coordinate internationally a small set of paleoclimate modeling experiments that should lead to a better understanding of these climates and should uncover both strengths and limitations of current GCM's. In order to evaluate climate models PMIP recognizes the importance of strengthening and widening the collaborations between climate modelers and paleodata experts who are using environmental data from terrestrial and marine sediments and from ice cores to reconstruct paleoclimates. Increased interaction between these two communities should enhance the usefulness of paleoclimate reconstructions and also help in the interpretation of regional paleoclimate data.

The first two PMIP experiments involve simulations of the last glacial maximum 21,000 years ago (21 ka BP) and the mid-Holocene (6 ka BP). Climate modeling groups from Australia, Canada, France, Germany, Japan, Korea, the United Kingdom, and the United States are performing these GCM experiments under identical, imposed "boundary conditions." For the last glacial maximum, the glacial ice cover reconstruction by Peltier (1994) has been adopted and prescribed in all the models. This change along with imposed changes in the carbon dioxide content of the atmosphere, as estimated from fossil air trapped in ice cores retrieved from Greenland and Antarctica, and orbitally determined changes in the insolation pattern, associated with changes in the earth's axial tilt and the season of perihelion, presumably lead to significant changes in the atmospheric circulation and a climate rather different from the present. Similarly, changes in prescribed conditions for the mid-Holocene have been imposed in the other PMIP experiment.

Here we provide a brief history of PMIP, its goals and its structure. We describe the two PMIP experiments and the scientific issues that will be addressed, and we summarize the current status of the project and future plans.

2. Background 

The PMIP effort developed out of a NATO Advanced Research Workshop, convened in 1991, which brought together climate modeling groups and scientists who collect and analyze paleoclimate data. A fruitful interaction and exchange of information between these two groups led to a commitment to initiate a more cooperative and coordinated effort to compare model simulations with paleoclimate data. The workshop participants agreed to focus initially on two specific periods in the past, the last glacial maximum and the mid-Holocene.

The report from this workshop summarized the plans for PMIP and enumerated the purposes of the project:

• to compare model simulations with paleoclimate data to identify and understand the consistencies and inconsistencies.
• to understand how different representations of physical processes in models can lead to differences in simulated climates. The intercomparison of models used to simulate paleoclimate conditions complements other GCM intercomparison projects (notably, AMIP) in which many of these same models are being run with other boundary conditions.
• to identify regions of the globe where intensified efforts to refine paleoclimate reconstructions might be most useful in evaluating climate models. Areas where PMIP experiments yield differences in simulated climate can be targeted by paleodata experts with the goal of resolving these differences.
• to better understand the range of the uncertainty of current estimates of climate sensitivity. GCM parameterizations are developed and validated using present-day observations. Current data cannot, however, ensure that these parameterizations will respond realistically as the climate undergoes major changes that lie well outside the range of variability contained in conventional climatic records of the past century; this leads to uncertainty in model estimates of climate sensitivity. Past climates offer a unique opportunity to test models in this respect, at least when sufficiently quantitative and globally distributed paleodata are available and when the causes of the changes are deterministic and well quantified.
• to use climate model output to help in the interpretation of paleoenvironmental data.

In the first stage of PMIP the boundary conditions to be used in both the 6 ka BP and 21 ka BP experiments were defined. The insolation patterns were standardized following Berger (1978). Carbon dioxide concentrations were prescribed according to ice core data (Lorius et al., 1985). As a result of a workshop sponsored by NSF through COHMAP (EOS, 1994), it was decided that PMIP should adopt Peltier's (1994) reconstruction of the glacial ice thickness and extent. The data were then archived at the National Geophysical Data Center (NGDC) in Boulder, Colorado and have been electronically distributed to the PMIP modeling groups.

From the beginning, PMIP's progress has been marked by remarkable cooperation among the modeling groups involved. Gradually, an organizational structure developed (see figure 1) in which we accepted responsibility for coordinating the project, and an advisory committee was formed to provide guidance. A Data-Model Intercomparison Subcommittee was also formed to facilitate communication between climate modelers and the much larger community of scientists involved in collecting and analyzing paleoclimate data. It was agreed that NGDC should serve as the archive for paleoclimate data of interest to PMIP modelers and would also distribute the PMIP boundary condition data sets (including reconstructions of glacial ice and sea surface temperatures). It was also decided that the Program for Climate Model Diagnosis and Intercomparison (PCMDI) should serve as the initial archive of PMIP model output.

 
Figure 1 - PMIP Organizational Structure

The first step to encourage stronger involvement of the paleodata community in PMIP and to foster stronger interactions between data experts and modelers was taken in the autumn of 1993 when a NATO Advanced Research Workshop was convened in Aussois, France. The primary aim of the workshop was to develop a strategy for coordinating the collection, analysis, and synthesis of paleoclimate data to expand its usefulness to projects like PMIP. The goals of the workshop were generally met and an increased collaborative spirit was kindled between the scientists who primarily collect and analyze paleoclimate data and scientists engaged in climate modeling. The formation of the PMIP Data-Model Intercomparison Subcommittee was based on a recommendation originated at this workshop.

By the end of 1993 PMIP had been endorsed by both IGBP/PAGES (International Geosphere-Biosphere Program / Past Global Changes) and by the Working Group on Numerical Experimentation (WGNE) which is a part of the World Climate Research Programme (WCRP). These connections ensure coordination of PMIP with related international programs.

The number of modeling groups participating in PMIP has grown to about eighteen. Some groups are carrying out all the PMIP simulations, while others will limit their participation to only one of the two PMIP time periods. Participants are contributing a standard set of model output to the PMIP archive (at PCMDI).

3. Experiment Design 

The PMIP experiments are designed to determine the model predicted equilibrium climate that is consistent with certain imposed changes in boundary conditions characteristic of the periods under study. The term "boundary conditions" is used here to indicate various prescribed conditions (e.g., orbital parameters determining the insolation pattern, atmospheric composition, glacial ice distribution and in some PMIP simulations, sea surface temperatures), which are considered to be external to the components of the climate system considered by typical GCM's. By considering equilibrium climatic states ("climate snap-shots"), we limit the kinds of issues that can be addressed concerning the evolution of climate from one state to another. On the other hand, we greatly reduce the computational expense of doing these experiments, and we also simplify the task of intercomparing model results by isolating the atmospheric and near surface climate from the deep ocean and cryospheric components of the climate system. Table 1 summarizes the boundary conditions prescribed for the PMIP simulations.

In the mid-Holocene experiment, the most important change in boundary conditions is the change in insolation pattern, due primarily to the precession of the equinox. For simplicity and to help isolate the effects of the change in insolation pattern, we assume that the sea surface temperatures (SST's), which are prescribed in this experiment, are identical to today's SST's. The CO2 concentration is also assumed to be lower during the mid-Holocene (280 ppm, instead of the near present-day value of 345 ppm). All other conditions are assumed to be identical with the present climate.

As outlined above, the mid-Holocene climate experiment changes are forced primarily by a different seasonal distribution of insolation. The Northern Hemisphere extratropical latitudes received larger incoming solar radiation during the summer season than at present, allowing for a warming of the continents. The terrestrial data coverage for this time period is relatively complete and often accurately dated (e.g., COHMAP, 1988; Wright et al., 1993; and other data archived by the NOAA Paleoclimate Program at NGDC). These data will permit a relatively detailed comparison with the models' simulated climatic response over the continents to the accurately known changes in the insolation pattern. In particular, the 6 ka BP experiment should be useful in evaluating ground hydrology parameterizations which strongly influence continental climates. We also intend to examine and characterize the response of the monsoon circulation to the change in insolation pattern, following the pioneering work by Kutzbach (1981) and later results from Mitchell et al. (1988).

For the other PMIP time-period, the last glacial maximum (21 ka BP), two experiments are underway. In one experiment SST's have been prescribed, while in the other they are being computed by the models (which include at the very least a "slab" or mixed layer ocean). In both experiments the massive ice sheets covering North America and Scandanavia have been prescribed according to a recent reconstruction by Peltier (1994). In the experiment with prescribed SST's, the CLIMAP (1981) reconstruction of the seasonally varying temperature distribution has also been imposed. In the experiment with computed SST's, we assume the horizontal transport of heat by the oceans is the same as for the current climate. In both experiments the CO2 concentration is prescribed to be 200 ppm and the insolation pattern reflects the 21 ka BP orbital configuration.

The last glacial maximum climate is characterized by large changes in the surface boundary conditions (ice sheet extent and height, sea surface temperatures (SSTs), albedo, sea-level) and atmospheric carbon dioxide concentration, but only minor changes in the insolation pattern. This period is important for understanding how ice sheets and lowered CO2 levels influence climate. Data for both boundary conditions and model verification are relatively abundant for this period. Over the oceans, SSTs have been reconstructed (e.g., CLIMAP) and over the continents this period has also been extensively studied (e.g, COHMAP). Among the climatic features of interest in this experiment are the simulated changes in the Northern Hemisphere jet stream location and associated changes in the storm tracks (see, for example, the initial experiments by Gates, 1976; and many subsequent studies, including Valdes and Hall, 1994; Manabe and Broccoli, 1985; Rind, 1987; Kutzbach and Guetter, 1986; Joussaume, 1993). For the case in which the SSTs are not prescribed, comparisons will be made between the model-computed SSTs and the CLIMAP SSTs. Of particular interest is the question of whether the tropical temperature changes simulated by the model are consistent with the paleoclimate reconstructions.

The PMIP models will in most cases be very similar to those that are used for doubled CO2 experiments and in the Atmospheric Model Intercomparison Project (AMIP), which is evaluating models (with sea surface temperatures specified) by comparing their simulation of today's climate with the observations for the 10-year period from 1979 through 1988. This will make it possible to obtain a more complete understanding of model differences and how they originate.

4. Current Status and Plans 

Most of the 18 modeling groups participating in PMIP are currently carrying out their simulations in preparation for a PMIP Workshop planned for October 1995. The workshop will provide the first opportunity for the PMIP modeling groups to report on their results and to begin to explore model differences. The workshop will serve to initiate the model intercomparison within PMIP. Results from the PMIP simulations will be shared in order to identify common responses, discover where disagreement among models is large, and identify in individual models any unusual or extreme responses that might warrant further investigation. We plan to begin comparing model simulations with available paleoclimate data and identify areas of consistency and disagreement. The Data-Model Intercomparison Subcommittee will also participate in the workshop, and we hope to determine if there are particularly critical data needs from the modelers' perspective and also learn what the paleoclimate data community needs from modelers. We expect to identify issues of common interest and form subprojects for further study. Finally, participants will share methodologies and approaches for analysis and diagnosis of paleoclimate simulations and propose (or demonstrate) strategies for intercomparison of model results and comparison of results to observations.
A second workshop is tentatively planned for the spring of 1996 which would focus on model-data comparisons. In addition, results from ongoing subprojects would be presented and ideas for new subprojects would be discussed.
Acknowledgment. We thank J.E. Kutzbach and J.F.B. Mitchell for their helpful comments on the text. This work was performed under the auspices of the Department of Energy Environmental Sciences Division at the Lawrence Livermore National Laboratory under contract W-7405-ENG-48 and the Commissariat l'Energie Atomique with support from EC under contract EV5V-CT94-0457.

References 

Berger, A., 1978: Long-term variations of daily insolation and Quaternary climatic changes. J. Atmos. Sci., 35, 2362-2367.
CLIMAP, 1981: Seasonal reconstructions of the earth's surface at the Last Glacial Maximum. Geological Society of America, Map Chart Series MC-36, Boulder, Colorado.
COHMAP, 1988: Climatic changes of the last 18,000 years: observations and model simulations. Science, 241, 1043-1052.
EOS, 1994: Reconstructing the last Glacial and deglacial ice sheets, EOS, 75, 82-84.
Gates, W.L., 1976: The numerical simulation of ice-age climate with a global general circulation model. J. Atmos. Sci., 33, 1844-1873.
Joussaume, S., 1993: Paleoclimatic tracers: an investigation using an atmospheric general circulation model under ice age conditions, part 1: desert dust. J. Geophys. Res., 98, 2767-2805.
Kutzbach, J.E., 1981: Monsoon climate of the early Holocene: climatic experiment using Earth's orbital parameters for 9000 years ago. Science, 214, 59-61.
Kutzbach, J.E., and P.J. Guetter, 1986: The influence of changing orbital parameters and surface boundary conditions on climate simulations for the past 18,000 years. J. Atmos. Sci., 43, 1726-1759.
Lorius, C., J. Jouzel, C. Ritz, L. Merlivat, N.I. Barkov, Y.S. Korotkevitch, and V.M Kotlyakov, 1985: A 150 000 years climatic record from Antarctica. Nature, 316, 591-596.
Mitchell, J.F.B., N.S. Grahame, and K.J. Needham, 1988: Climate simulations for 9000 years before present: seasonal variations and effects of the Laurentide ice sheet. J. Geophys. Res., 93, 8283-8303.
Manabe, S. and A.J. Broccoli, 1985: The influence of continental ice sheets on the climate of an ice age. J. Geophys. Res., 90, 2167-2190.
Peltier, W.R., 1994, Ice age paleotopography. Science, 265, 195-201.
Rind, D., 1987: Components of the ice age circulation. J. Geophys. Res., 92, 4241-4281.
Valdes, P.J. and N.M.J. Hall, 1994: Mid-latitude depressions during the last ice-age. In Long-Term Climate Variations - Data and Modelling. Edited by J.-C. Duplessey and M.-T. Spyridakis, NATO ASI Series I, vol 22, 510-532.
Wright, H.E., Jr., J.E. Kutzbach, T. Webb III, W.F. Ruddiman, F.A. Street-Perrott, and P.J. Bartlein, eds., 1993: Global Climates Since the Last Glacial Maximum. University of Minnesota Press, Minneapolis, MN.

Advisory Committee

L. Gates, J. Kutzbach, S. Manabe, J. Mitchell, W. Prell, C. Prentice, D. Rind, A. Street-Perrot

Coordinators

S. Joussaume
K. Taylor

Paleodata Participants

Climate Modelers

NGDC
R. Webb
PCMDI
K.Taylor

Data-Model Intercomparison Subcommittee

P. Bartlein, J. Guiot, S. Harrison