Palaeogeography, PaUzeoclimatowgy, Pakzeoecowgy (Gwbal and Planetary Change Section), 82 (1990): 79-85 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands
Paleoenvironmental modelling and global change W.R. PELTIER Department of Physics, University of Toronto, Toronto, Ort. M5S lA 7 (Canada) (Received November 22, 1989)
Introduction The reconstruction of past environments using infonnation extracted from the geological record, even in the absence of models of paleoenvironmental change that are able to explain the observed variability, will clearly play an extremely important role in the International Geophere-Biosphere Program. Such data provide the only framework within which we may come to understand what constitutes the natural long time scale variability in the climate system. If we wish to "prove" an anthropogenic effect then we shall have to show that the "signal" at issue (e.g. an increase of mean surface temperature, a decrease of continental ice volume, or an "eustatic" rise of sea level) is not plausibly of internal origin. Only if we can embed our understanding of the climate system in an appropriate (mechanistic) mathematical model, however, will we be able to effect a truly satifactory separation between internal and anthropogenically driven variability and therefore truly understand what the future may have in store as the system responds to the various stimuli to which it is subject. Simply put we need models in order that we may predict the future on the basis of the present and appropriately recent past. On timescales of hours to days only the similarly "fast physics" of the climate system plays any important role in determining intrinsic predictability and therefore only the planetary atmosphere (say) need be explicitly included in the model. As the timescale of interest increases, additional components of the "earth system" Elsevier Science Publishers B.V.
come into play. Thus the oceans enter the system as active" players" when the timescale exceeds a week or so and become progressively more strongly dominant as timescales exceed the annual. On still longer timescales of the order of decades to centuries we begin to see direct effects of the deep ocean circulation and of natural variations of atmospheric CO2 concentration, for example. On still longer timescales of tens of thousands to hundreds of thousands of years (in the Milankovitch band) we discern evidence of the activity of continental ice sheets as these respond to the astronomical forcing in such a way as to generate cycles of glaciation and deglaciation. Solid earth geophysical processes such as glacial isostatic adjustment enter the same timescales. On timescales in excess of a million years we observe the influence of mantle convection as the land-ocean configuration evolves in response to this penultimate of "slow" physical processes. It might be imagined, because the IGBP is to be focussed upon an attempt to understand the evolution of "the earth system on timescales of decades to centuries, that the only models of interest will be those that embody processes operating on timescales of the same order as these. This is clearly not the case. The problem is that the secular variations in the climate system that we might wish to ascribe to anthropogenic influence may in fact be associated with some naturally occurring oscillation operating on a much longer timescale. For example, a secular rise of sea level observed on a tide gauge in a record of 100 years duration could be due to the influence of glacio isostatic adjustment
so acting on a timescale of thousands of years and not to the melting of continental ice due to a CO2 enhanced greenhouse effect. The "slow" physical processes may therefore contribute crucially to the background against which anthropogenically induced global change must be observed. Appropriate models of these slow physical processes, designed to accord with the geological record of paleoenvironmental change, may be employed to filter the observations so as to more clearly reveal the anthropogenic signal. Also, the modelling of the slow physics has much to teach us of the manner in which the various components of the earth system interact on these timescales.
A hierarchy of three paleoenvironmental models with slow physics The field of paleoenvironmental modelling is not one that has an especially voluminous literature associated with it but it is nevertheless substantial and I wish therefore to begin with a disclaimer. I will not attempt to discuss at all the important contributions to this subject that have been made using conventional atmospheric general circulation models since these have no active geology and therefore are liable to be of less interest to this group. Similarly, I will not discuss the chemical oceanographic box models that have been developed in an attempt to understand the carbon cycle (say) in the context of observations of increasing atmospheric CO2 concentration. What I will do is simply to focus upon a number of related models that speak of important issues of paleoenvironmental change during the Pleistocene period of earth history and which are liable to prove themselves useful in the context of IGBP.
1. Glacial isostatic adjustment and relative sea level change Above I commented on the problem of discerning anthropogenic influence in modern tide gauge records of secular sea level change against the background associated with glacial isostatic adjustment following the last deglaciation event
of the present ice age that ended about 7000 years ago. This is an important problem because the tide gauge records show unambiguous (if perhaps not truly globally representative) evidence that "eustatic" sea level is presently rising at a rate that exceeds 1 mm yr-l. This has been attributed to the expected greenhouse warming having caused both a thermal expansion of the oceans and a melting of the small ice sheets and glaciers of the world (e.g. Gomitz et a!., 1982; Meier, 1984). The problem with this interpretation has been that the tide gauges do not sample the ocean basins homogeneously and therefore, given that the observed rate of secular rise is a strong function of geographic location, there has been little confidence expressed (Barnett, 1983) in the meaningfulness of the observation itself. Recent analyses have shown however (Peltier, 1988; Peltier and Tushingham, 1989) that when the gravitationally self consistent global model of postglacial sea level change (e.g. Peltier, 1982) is employed to filter the tide gauge data so as to eliminate the influence of glacial isostasy, then the filtered data clearly reveal a signal that is much more spatially uniform. Furthermore, the globally averaged rate of sea level rise revealed by the filtered data is increased by about 30% when such contamination due to ongoing isostatic adjustment is removed. Of course the geophysical model employed to filter the tide gauge data is believable only to the extent that it can be shown to accord with the geological record of postglacial sea level change that for the most part consists of 14C controlled sea level histories in the time window from 0-18 kyr B.P. Figures 1 and 2 of this paper respectively show a sequence of fits of the sea level model to a sequence of rsl records and the main output of an EOF (Empirical Othogonal Function) analysis of the PSMSL (Permanent Service for Mean Sea Levei, Bidston, U.K.) data base of secular sea level trends with the effect of isostatic adjustment removed. This standard spatial eigenvector-temporal eigenvector presentation reveals a spatially homogenous signal of present day sea level rise of magnitUde near 2.4 mm yr-l, that might be ascribed to the "greenhouse" effect. Clearly, understanding the geophysics/geology
PALEOENVIRONMENTAL MODELLING AND GLOBAL CHANGE
MAKTAK FIORD BAF.
8 CHURCHILL MAN.
-- ....... .... ..., ,
Time (ky BP)
.... , , ,
.... ...... _----
~-r--------r-------~------~ 15 Time (kyBP)
BOSTONMA. g-r------------------------~ ,
" ,, ,, ,,
8 RIMOUSKI QUE.
,, ,, ,, ,
Fig. 1. Radiocarbon-dated relative sea level histories at six sites on the North American continent. The solid crosses denote observed paleo-shorelines, the solid curves denote the predictions from a high resolution version of the global isostatic adjustment model, while the dashed curves are results from a previously constructed low resolution model.
is crucial in this case to extracting the modern climate signal. 2. The mid-Pleistocene climate transition A fundamental problem in paleoclimatology is clearly the origin of the 10 6 year cycle in the variation of continental ice volume that has dominated the global climate record throughout the past million years of earth history. Multi-
million year records of this variability are provided by ~180 data from deep sea sedimentary cores. Figure 3 shows a somewhat "atypical" record, that from ODP site 677 in.the Panama Basin (N.J. Shackleton, pers. comm., 1988). The time scale for this record has been deduced using a somewhat novel technique developed in my group at Toronto that involves "tuning" to achieve phase coherence with the astronomical input in the obliquity and eccentricity-preces-
U G) > C
, ,, ,,, "'i
" " " " ",
,,,. ,, ,,
I I I I I.. 01' \•' \ ...
," , ,, " ,
'I , I , I'
d ~ 1 'I
, , ' \,
--r.---!t--t.J.~~---', 1,,' ",, "~
," , I
, ", , ~ ,
, " ,,, ",
I , , ,
, , ' '
~ "I '
" ", " , ," " ' "
..., " (
Fig. 2. The upper and lower plates respectively show the first (solid curve) and second (dashed curve) spatial and temporal eigenvectors delivered by an empirical orthogonal function (EOF) analysis of secular sea level trends measured on tide gauges at the 40 sites whose names appear above the top plate. Prior to analysis each tide gauge record was filtered so as to remove the secular sea level change that would be occuring at each site if the only active process was glacial ieostatic adjustment. This analysis demonstrates that there exists a strong residual signal that is global in scale and that consists of an rsl rise of magnitude 2.4 mm yr-l.
sion bands. This high resolution record clearly shows a sharp transition near 106 yr B.P., prior to which the 105 yr cycle is absent but following
which it is the dominant source of variance in this paleoclimatic record. A nwnber of simple models of the 10 5 year cycle have been proposed
PALEOENVIRONMENTAL MODEWNG AND GLOBAL CHANGE
that appear to explain the origin of this oscillation, all of which involve coupling the physics of glacial isostatic adjustment to the physics of ice sheet flow and accumulation in a 1-D time deODP677 -
- ~ - ~ -
pendent approximation (Oerlemanns, 1980; Peltier, 1982; Pollard, 1982) but none of which are capable of explaining the observed transition. Recent redesign of the model of Peltier (1982) PANAMA BASIN
I I eI
POWER SPECTRA 0-1 _.- 1 -
1.99 Wyr bp
I I co
Fig. 3. The upper plate shows B18 0 VB time (in thousands of years before present) from ODP Site 677 in the Panama Basin (data from N.J. Shackleton, pers. comm.) with timescale developed using a new orbital tuning technique. Clearly evident is the mid-Pleistocene climate transition that occured near 10 6 yr B.P. Prior to the transition the series shows no evidence of any 105 yr cycle while after the transition has occured the record is dominated by the 10 5 yr cycle. The lower plate shows power spectra for these two 106 yr segments of the record illustrating this point. The four prominent spectral peaks at frequencies higher than 5 2 cycles/l0 years are those due to the variation of orbital obliquity (41 kyr period), and the triplet of lines in the eccentricity-precession band with periods of 19 kyr, 21 kyr, and 23 kyr.
(see also Hyde and Peltier, 1985, 1987) so as to enable it to operate with much higher resolution, has demonstrated that the above mentioned two processes of themselves are in fact insufficient to support an oscillation of the observed amplitude. When the model is forced with the Milankovitch input, although a weak relaxation oscillation of 10 5 year period is induced by the system nonlinearity, no reasonable fit to the data is achievable. However, if an additional feedback loop is added in the form of an additional forcing on the snowline that is
proportional to the time rate of ice volume change, then a strong 10 5 year synthetic oscillation is delivered by the model (an example of one such model output is shown in Fig. 4). I would suggest that this feedback is associated with the sympathetic oscillation of CO2 that is known to accompany the 10 5 yr cycle (e.g. Lorius et al., 1985; Shackleton and Pisias, 1985). However, it could also be associated with a reorganization of the ocean circulation (W. Broecker, pers. comm., 1988). Clearly a great deal of further work remains to be done in this extremely
10.0 +-+---l--+--~--t--If----i--1---+---I---+-+--I--+ 2.S
8.0- , 6.0-
.s ·.s 2.0
10.0 -j-L--1t---I~-L-+----+---+---I---+--I---t---I---+-"'::--J--+---t- -2..5 B.O i
.0 -t-'t...--I--+--i--+--+--+---+---I----+---+----+---+----+--+.2.S ·700.0 -600.0 -SOO.O -4!Xl.0 -300.0 .200.0 -100.0 .0 TIME(KYEARS) Fig. 4. Comparison of the SPECMAP /l 18 0 record (dashed curves) with the predictions of a simple 1-0, time dependent paleo climate models of increasing complexity (a-d). The models differ only in the manner in which they approximate the isostatic adjustment process. In (a) this process is absent whereas in (d) it is treated most accurately. Inspection shows that the incorporation of this process alone into the climate model does not a1low it to fully reconcile the strength of the 10 5 yr line in the 8 18 0 data.
PALEOENVIRONMENTAL MODELLING AND GLOBAL CHANGE
interesting area. I would strongly suggest it as a focus for lUGS Global Change activity.
3. Geographically realistic models of continental glaciation and deglaciation The existing models of the 105 yr cycle discussed above are of course rather unsatisfactory in a number of rather important physical respects. For example, having no geography whatever, they cannot be employed to address the question as to why the main Pleistocene ice masses occupied the positions they did. In order to address this and related issues we clearly require a more realistic framework for our analysis. If we are interested in timescales of the order of the spacing between successive interglacials then the model of choice will clearly not be a standard atmospheric-oceanographic GCM! Even operating in quasi-static mode its application would require enormous computational resources that are simply unavailable. I would suggest that a great deal may yet be learned, however, by coupling a simple 2-D quasi-linear energy balance model through the Milankovitch forcing to the ice physics and earth physics that appear to be the fundamental ingredients in the 10 5 yr cycle. North et al. (1980) have clearly shown such models to be capable of correctly predicting the most likely sites of continental glaciation for the orbital conditions that obtained at the last glacial maximum. They have been equally successful in predicting the timing of the glaciation of the Antarctic continent to have coincided with the initial isolation of the continent by the circum-polar current after its split from Pangea. There are a number of technical problems that will have to be solved before such models can be successfully employed to mediate the way in which the Milankovitch forcing drives ice physical and thus earth physical processes, but once these are resolved such models will clearly provide a useful means of establishing more direct contact between continental geomorphological observations on the one hand, and oceanographic observations on the other.
A role for paleoenvironmental modelling The few examples provided above I hope will suffice to make the point that a number of paleoenvironmental models are already in hand which promise to provide us with useful means of "codifying" our understanding of the way the earth system operates on long time scales in its pristine state. Such models might be developed to the extent they could be employed to make useful forecasts of future states of the system and to assess the impact of anthropogenic influence upon these future states. This is clearly a fundamental goal of the IGBP initiative. Perhaps the lUGS might consider doing what it can to stimulate the further development of activity in this area of paleaoenvironmental modelling.
References Barnett, T.P., 1983. Recent changes in sea level and their possible causes. Clim. Change, 5(1): 15-38. Gornitz, V., Lebedef, S. and Hansen, J., 1982. Global sea level trend in the past century. Science, 215: 1611. Hyde, W.T. and Peltier, W.R., 1985. Sensitivity experiments with a model of the Ice Age cycle: The response to hannonic forcing. J. Atmos. Sci., 42: 2170-2188. Hyde, W.T. and Peltier, W.R., 1987. Sensitivity experiments with a model of the Ice Age cycle: The response to Milankovitch forcing. J. Atmos. Sci., 44: 1351-1374. Lorius, C., Jouzel, J., Ritz, C., Merlivat, L., Barkov, N.-I., Korotkevich, Y.8. and Kotlyakov,/V.M., 1985. A 150,000 year climatic record from Antare~ic Ice. Nature, 316: 591-596.',_ Meier, M.F., 1984. Contribution of small)l~ to global sea level. Science, 226: 1418. North, G.R., Mengel, J.G. and Short, D.A., 1983. Simple energy balance model resolving the seasons of the continents: Application to the astronomical theory of the Ice Ages. J. Geophys. Res., 88: 6576-6586. Oerlemanns, J., 1982. Model experiments on the 100,000 year glacial cycle. Nature, 287: 430-432. Peltier, W.R., 1982. Dynamics of the Ice Age Earth. Ad. Geophys., 24: 1-146. Peltier, W.R., 1988. Global sea level and Earth rotation. Science, 240: 895-901. Peltier, W.R. and Tushingham, A.M., 1989. Global sea level rise and the greenhouse effect-might they be connected? Science, 244: 806-810. Shackleton, N.J. and Pisias, N.G., 1985. Atmospheric carbon dioxide orbital forcing, and climate, In: E.T. Sundquist and W.8. Broecker (Editors), The Carbon Cycle and Atmospheric CO2 : Natural Variations Archean to Present. AGU Monogr, 32: 303-318.