Ber78

Obliquity Last 150 Kyr

Figure 8.L Long-term variations of eccentricity, climatic precession, obliquity, and insolation at 65°N in June (Wm 2) from 2(M) kyr BP to 130 kyr AP (Berger, 1978),

Figure 8.L Long-term variations of eccentricity, climatic precession, obliquity, and insolation at 65°N in June (Wm 2) from 2(M) kyr BP to 130 kyr AP (Berger, 1978), time the amplitude of obliquity is also reduced (obliquity going from 22.6 at 10 kyr AP to slightly less than 241 at 31 kyr AP). Because the daily insolation is mainly a function of precession (Berger et al., 1993a), all these particular characteristics lead to an insolation that will vary little over the next 50 kyr, a feature that is strikingly different from the last interglacial (with an amplitude of 25 Wrn : against 105 for 65UN in June). This particular feature is unique over the past 3 Mvr, and the closest analog occurred 400 kyr ago, at stage 11 (Berger and Loutrc, 1996). This might have an influence on the sensitivity of the climate system to other forcings such as the greenhouse gases because, among the driving forces, one (the insolation) will remain almost constant, giving the others a possibility of playing a more important role.

The discovery in 1976 (Hays etal.) that the main periods of these three elements were preserved in the ice volume signal of ocean cores was a major step toward understanding the glacial-interglacial cycles. With time, the astronomical theory of palcoclimates has been extended and is now related mainly to the response of the climate system to the astronomically driven changes of the latitudinal and seasonal distributions of the energy received from the Sun. In particular, it is worth pointing out that the hypothesis by Milankovitch, requiring a low summer insolation in high northern latitudes to trigger a glacial, is actually coherent with the occurrence of a reduced seasonal contrast and of an enhanced insolation in winter, essentially in lower latitudes. Indeed, according to the astronomical theory, it is possible to demonstrate that the long-term variations of the daily insolation depend mainly on precession (except for the latitudes and days close to the polar night, Berger et al,, 1993a). As a consequence, for a given latitude, there is a phase lag of about 2 kyr (one-twelfth of a precessional cycle) between the insolation of two consecutiv e months (Loutre and Berger, 1995). Therefore, when the high latitudes receive less insolation in summer, it is also the case for the mid- and low latitudes, which, in turn, receive more insolation in winter. This has a direct impact on climate, and mild winters in regions where the evaporation is the largest (the tropical oceans) allow more water vapor to be available at the global scale. At the same time, a larger winter latitudinal gradient strengthens the atmospheric circulation and the water vapor transport to the northern high latitudes, where it falls as snow.

To investigate the complexity of such a response, a series of numerical experiments have been performed with climate models of different complexity (Berger, 1995), in addition to the detailed analysis of many proxy records. Among them, the Louvain-la-Neuve (LLN) 2,5-D climate model (Gallee et aL, 1991) has been used to reconstruct the long-term climatic variations over the Quaternary Ice Age. Sensitivity analyses to the insolation changes calculated by Berger (1978) and to the COi atmospheric concentration reconstructed by Jouzel et aL (1993) have been performed over the last 200 kyr (Gallee et al., 1992; Berger et aL, 1998b). Because the LLN model does not yet have a carbon cycle component, the atmospheric CQj concentration is used as a forcing, although it acts as a feedback in the real world. On long time scales (tens and hundreds of thousands of years), CO2 covaries with the Yostok isotopic temperatures (Yiou et aL, 1991) and other climatic variables (Yiou et aL, 1994). Records show minimum glacial CO2 concentrations around 190 ppmv, glacial-inter glacial transitions accompanied by a rapid increase in CO2 concentrations to a maximum of about 290 ppmv, and a gradual return to low COj values during glaciation. These variations have been attributed to climate-induced changes in the carbon cycle, but they also amplify climate v ariation by the accompanying greenhouse efTcct. However, the Yostok CO2 spectrum is far from reproducing all the astronomical frequencies.

Significant peaks appear near 143, 59, 31, and 22 kyr and at shorter periods (Yiou et aL, 1991), although peaks near the orbital frequencies arc more clearly visible in the Yostok temperature record at 41.7, 24.1, and 18.3 kyr. The peaks near 100 kyr are strongly shifted to the left in both Yostok records (133.3 kyr for temperature and 143 kyr for CO>), and the peak near 41 kyr has very little significance in the CO> record. Periods close to 60 and 30 kyr (also present in the temperature record) do not appear to be related to the orbital frequencies themselves nor to their combination tones but rather seem to be caused by ice sheet/bedrock interactions. On shorter time scales, the picture is even more complex (Fischer et al., 1999). During all the last three terminations (between marine isotopic stages ¡MIS] 8 to 7, 6 to 5, and 2 to I), the rise in COj concentration lags behind temperature change by 400 to 1000 years. Between 14 and 13 kyr a brief small decline in CO2 lags the Antarctic cold reversal (ACR) in the Antarctic isotope temperatures by ~400 years, but it occurs kyr before the Younger Dry as cooling event (Blunter et aL, 1997). As a consequence, the ACR leads the Younger Dryas by at least 1.8 kyr, and CO? rises steadily during the lounger Dry as. During the Holocene and Eemian inter glacials, atmospheric CO2 concentrations drop by "-10 ppmv after an initial maximum reached, respectively, at 10 kyr BP and 128 kyr BP. This drop might be attributed to a substantial increase in the terrestrial biospheric carbon storage extracting COi from the atmosphere. In the case of the Eem, COi concentration does not show a substantial change in the following 15 kyr, despite a distinct cooling over the Antarctic ice sheet. Not until 6 kyr after the major cooling of MIS 5,4 does a substantial decline in CO? occur, another 5 kyr being required to return to an approximate phase relationship of CO1 with the temperature variations. During the Holoccne, atmospheric CO» concentrations even increase during the last 8 kyr. In contrast, high CO2 concentrations are not sustained during MIS 7, but rather CO? follows the rapid temperature drop into MIS 7,4, reaching a minimum 1 to 2 kyr after the minimum in the isotope temperature.

In the experiments made with the LLN climate model, the insolation changes alone act as a pacemaker for the glacial-inter glacial cycles, but CO2 changes help to better reproduce past climatic changes and, in particular, the air temperature and the southern extent of the ice sheets. Actually we have shown that the albedo and water vapor-temperature feedbacks play a fundamental role in amplifying the astronomical perturbations. Forced by the insolation, albedo, and CO? of the Last Glacial Maximum (LGM)j the model by Berger et aL (1993b) leads to a 4.5 C cooling in the Northern Hemisphere (recall that the equilibrium response of the LLN model for a doubling of the CO2 concentration is a warming of 2 C). Two-thirds of the LGM cooling (3 C) is explained by the astronomical and albedo forcing when allowing for the water vapor feedback, which, by itself, accounts for a cooling of \2 C In a CXValonc forcing experiment the cooling amounts to 1.5 C, 40% of it (0.6 C) being due to the water vapor feedback. These results therefore stress the fundamental role of the albedo and water vapor feedbacks, They also show that the combined effect of a change in the CO2 and in the insolation plus albedo forcings, all of which generate the water vapor feedback, is about equal to the sum of the responses of the climate system w hen one or the other is kept constant. This type of linear response might be due to the fact that the perturbation in all cases remains small (a few degrees) as compared with the basic state of reference (^280 K), although the system itself is described with a set of highly nonlinear equations. This kind of behavior has also been found recently in more-complex climate models (Ramaswamy and Chen, 1997).

Leads and lags have also been analyzed in the response of the 1 ,LN climate model to the insolation and GOj forcings during the Herman interglacial and over the whole last glacial-interglacial cycle (Berger et aL, 1996), If we take the June insolation at 65 N as a guide for the time scale, the maxima of the simulated Northern I lemisphere ice volume -reached at 134 and 109 kyr BP, respectively, for isotopic stages 6 and 5d - lags behind this insolation by 6 kvr The minimum ice volume is reached at 126 kvr BP and lasts mt me

10 kvr, covering the whole period during which insolation is decreasing by 201,o from a maximum of about 550 Wm"* to a minimum of 440 Wm"*. Summer temperature of seawater at the surface (SST), at middle to high latitudes, lags behind insolation by a few thousand years. As a consequence, SST starts to decrease well before (actually

11 kyr before) the ice sheets start to grow on the continents (as, for example, at 116 kyr BP). In the 50-55 N band, the simulated SST then starts to rise again at that time and is therefore in antiphase with the continental ice volume. Cooling rises at the end of the melting phase of the ice sheets, a result in agreement with the reconstruction by Cortijo et al, (1994), who showed an abrupt cooling in the high latitude ocean only a few thousand years after the start of the last interglacial; and warming occurs at the beginning of their growing phase, in agreement with the hypothesis of Ruddiman and Mclntyre (1979), who claimed that the ice sheets were growing under warm SST conditions. At higher latitudes (70—75 N), the behavior of the zonal mean temperature at the surface is more complex, reflecting not only the direct influence of insolation but also the influence of sea ice, seawater, and the ice sheets.

To further investigate the importance ofCX>2 changes, in addition to the insolation, an atmospheric CO? concentration decreasing linearly from 320 ppmv at 3 Myr BP (late Pliocene) to 200 ppmv at the LGM was also used to force the model. Under such conditions, the model simulates the entrance into glaciation around 2.75 Myr BP (Li et al., 1998a), the late Pliocene-early Pleistocene 41 kyr cycle, the emergence of the 100 kyr cycle around 900 kyr BP (Berger et al., 1999), and the glacial-interglacial cycles of the last 600 kyr (Li et al., 1998b). The hypothesis was put forward that during the Late Pliocene (in an ice-free, warm world) ice sheets can develop only during times of sufficiently low summer insolation. This occurs during large eccentricity times when climatic precession and obliquity combine to obtain such low values, leading to the 41 kyr period between 3 and 1 Myr Sensitivity analyses have also demonstrated that to prevent ice sheets from developing, CO2 must be relatively large between 4 and 3 Myr BP (more than 450 ppmv in our model), in particular around 3.8 Myr BP when all the astronomical conditions are most favorable. But, at the same time, CO? concentration must decrease (less than 370 ppmv in our model) after 3 Myr BP to allow the ice sheets, triggered by the insolation forcing, to grow sufficiently at around 2.7 Myr BP. On the other hand, it is interesting to point out that the 100-kyr signal appears in the climatic record at around 1 Myr BP, when the 100-kyr component of eccentricity starts to fade. Very strong between 2 and 1 Myr BP, it weakens progressively between 1 Myr BP and now, and it finally disappears over the next 500 kyr for the benefit of the 400-kyr cycle, in relation to the-2-Myr period of term number 6 in the expansion of eccentricity.

In a glacial world, ice sheets persist most of the time except when insolation is very high in polar latitudes, again requiring large eccentricity but leading this time to an interglacial and finally to the 100-kyr period of the last 1 Myr. Using a CO? concentration reconstructed over the last 600 kyr from a regression based on SPECMAP (Li et al., 1998b), it has been shown that stage 11 and stage 1 require a high CO? to reach the interglacial level. The insolation profile and modeling results at both stages tend to show that stage 11 might be a better analog for our future climate than the last Eemian interglacial. Such a CX>2 reconstruction was used because no other CO? record (Vostok in particular, Petit et al, 1999) was yet available over such a long period at the time the experiment was performed, It must be understood, however, that this is only a temporary surrogate because the SPECMAP <518O record is not verv well correlated with the CO2 record over the period for which both data were existing (Li et al, 1998b). Using the calculated insolation and a few scenarios for CO2, the climate of the next 130 kyr has been simulated (Loutre and Berger, 2000), showing that our interglacial w ill most probably last particularly long (50 kyr). This conclusion is reinforced if we take into account the possible intensification of the greenhouse effect as a result of human activities over the corning centuries.

Given the complexity of the interactions between the different components of the climate system, a major challenge remains; to understand how the astronomical forcing is amplified by different processes that control the response of the climate system. In this chapter, we focus on the role of the atmospheric GO2 concentration changes, sea level changes, and the vegetation-climate feedback over the last glacial-interglacial cycle.

8.2 CO2 and Insolation Thresholds

Experiments made with the LLN NH climate model sought to explain its response to the astronomical and CO; forcings. These experiments show, in particular, that the sensitivity of the Northern Hemisphere ice volume to CO? is not constant through time.

To test the Hays et al. (1976) hypothesis that the orbital forcing acts as a pacemaker of the ice ages, experiments were made in which insolation was allowed to change, but CO2 was kept constant to either 210, 250, or 290 ppmv (Berger et al., 1998a). As C02 varied around 230 ppmv most of the time during the last 200 kyr, these concentrations correspond to an average for, respectively, glacial, intermediate, and interglacial times (the present-day CO2 concentration is already 60% above the average value of the last glacial-interglacial cycle). These simulations show (Figure 8.2) that the

Northern hemisphere ice volume (10h knVj

Figure 8.2. Long-term variations of the Northern Hemisphere ice volume simulated by the Norihcrn Hemisphere 2-D LLN model (Gal lee et al., 1991) in response to the astronomical forcing (Berber, 1978) and a constant CO2 (2It) ppmv solid ]ineT 250 ppmv small dashed line, 2911 ppmv dashed line) (Bergeret al., 19*>Sb>.

Figure 8.2. Long-term variations of the Northern Hemisphere ice volume simulated by the Norihcrn Hemisphere 2-D LLN model (Gal lee et al., 1991) in response to the astronomical forcing (Berber, 1978) and a constant CO2 (2It) ppmv solid ]ineT 250 ppmv small dashed line, 2911 ppmv dashed line) (Bergeret al., 19*>Sb>.

modeled ice volume variations are comparable to the geological reconstructions only when the CO2 is low (210 ppmv), but, more importantly, that the response of the climate system is far from being linear in C02. For a CO? of 210 ppmv, the glacial maxima are reached at 181, 136, and 20 kyr BP, with about 35 x 106 km* of continental ice in the Northern Hemisphere. Secondary maxima of about 25 x I06 km * occur at 109, 90, and 61 kvr BP,

Analysis of the range of ice volumes simulated under these three constant CO¿ concentrations show that the sensitivity of the simulated Northern Hemisphere ice sheets volume to COj is far from being constant in time. It is larger at 136, 90, and 20 kyr BP, w hen the range of ice volume is much broader. These times correspond to secondary minima in the insolation curve, with a moderate value situated between 462 and 472 Wm"2 at 65°N in June. For the other three ice maxima (181, 109, and 61 kyr BP), the insolation reaches its deepest minimum (—4443 Wm 2) 7 to 11 kyr before ice sheets form under the three CO2 concentrations. At these times eccentricity is large, solstice occurs close to aphelion, and obliquity is low. A tentative conclusion may therefore be that the sensitivity of the climate system to CO2 is larger (i.e., a broader range of ice volume occurs in response to a given range of CO?) for intermediate v alues of the insolation minima. For the other ice maxima, the insolation minima are deep enough (lower than 460 Wm-2) to drive the system into a glacial stage for the three CO2 values.

The amplitude of the ice volume change for the three experiments is very different, but the timing of the build-up of the ice sheets is very similar. For the 181, 109, and 61 kyr BP ice maxima, the 250 ppmv curve is situated halfway between those of 210 and 290 ppmv. This is not the case for the other maxima. At 136 and 90 kyr BP, the 250 ppmv simulation is definitely closer to the 290 ppmv one. In contrast, at 20 kyr BP the 250 ppmv simulated ice maximum is tied very closely to the 210 ppmv one, This different sensitivity at the Last Glacial Maximum compared with what happens around 90 and 136 kyr BP is actually related to the state of the climate system before these ice maxima; the F.arth is in a glacial mode during isotopic stages 4 to 2, whereas the climate is interstadial before the 90 and 136 kyr BP maxima. I he very large difference in the response of the model for the 250 ppmv CO2 level also seems to indicate the existence of critical CO? concentrations (which are time-dependent) around which the climate system may be responding either as a high or a low CO? level, At the Last Glacial Maximum, even a rather high CO2 does not prevent the Earth from being largely glaciated. This indicates that the threshold value may be as large as 270 ppmv, with only a CO¿ concentration lower than this value leading to a glacial. On the contrary, the situations at 136 and 90 kyr BP require a rather small CO? for the Earth to enter into glaciation, and the threshold might be as low as 230 ppmv. The similar behavior of the response of the model under a 250 and a 290 ppmv forcing during all interglacials and interstadials (except isotopic stage 3) confirms that its sensitivity to CO? is different in a cold-glaciated Earth than in a warm, ice-free Earth. This implies that during the Holocene, an ice age can be initiated, in our model, only with a rather low CO2 concentration (below 250 ppmv at least).

8.3 Sea Level and Vegetation Changes

At the geological time scale, the amplitude of the mean sea level variations might be relatively large. This has a direct impact on the size of the continents by allowing or not allowing the continental platforms to emerge. In particular, the reconstruction of the configuration of the Earths surface at the Last Glacial Maximum by CLIMAP (1976) shows the emergence of the Sunda Isles and of large areas from the China Sea, a bridge between Australia and New Guinea and between Siberia and Alaska, and the disappearance of the North Adriatica, of the Channels, and of the North Sea.

Vegetation and land surface cover have also changed significantly over time. At the LGM (Prentice et al., 2000), the most obvious features are the equatorward regression of forest types in North America and Eurasia and a compression and fragmentation of the forest zones in these regions. I hesc vegetation changes provide evidence for drier conditions than present across large areas of the mid-latitudes. The boreal evergreen forest (taiga) occupied a far smaller area than today, and the temperate deciduous forest was apparently almost nonexistent. Refugia for this temperate forest and the tropical rain forest hiomes may have existed offshore at LGM, but their characteristic taxa also persisted as components of other biomcs. Tropical moist forests in Af rica were reduced.

At the peak of the Holoccne (6 kyr ago), in the northern circumpolar region, taiga extended poleward at the expense of tundra, indicating greater than present growing-season warmth (Prentice, 1998). This shift appears relatively slight (200-300 km) when viewed on global maps, and furthermore it is not symmetrical around the pole. In the northern mid-latitudes of Eurasia, the forest belts shifted poleward. In many cases, these shifts implicate warmer winter conditions even though the orbital forcing alone would tend to produce colder winters (the hiome paradox). In the circum-Mediterranean region, temperate deciduous forests encroached southward and xero-phytic w ood land/scrub were absent, suggesting a moister climate t han today. ISut the largest changes are seen in the monsoon regions, mainly in northern Africa, where the Sahara desert was drastically reduced and the Sahelian vegetation belts shifted systematically northward. The basic mechanisms of this monsoon amplification are caused by the early to mid-Holocene astronomical forcing, but further positive feedback mechanisms, involving changes in land surface and/or sea surface, must be invoked to account for the magnitude of the hiome shifts in Africa.

These vegetation changes are important for climate because the surface albedo is related not only to the snow and ice fields hut also to the nature of vegetation that covers the land surfaces. It can indeed be anticipated that a vegetation-albedo-climate coupling should exist, because changing global climate produces modifications in large-scale precipitation and temperature patterns, which, in turn, modify the vegetation of land surfaces and hence change the surface albedo, thus producing further climatic change, Charney et al. (1977) were among the first to investigate such an effect of albedo change on precipitation in semiarid regions. Otterman et al. (1984) then showed the importance to climate sensitiv ity of a lower effective snow albedo on forests and shifts in the tree line, using a mean annual model. Indeed, snow on forests has a considerably smaller effective snow-cover fraction than snow of the same depth on tundra, leading

Tabic 8.2a. Vegetation-Climate Feedbacks During the Last Glacial-Interglacial and in Glacial Times.

Authors

Model

Results

Gallee et al., 1992 Berger et aL, 1992, 1993c

Harvey, 1988, 1989

Last Glacial-Intcrglacial Cycle

• LLN NH 2.5-D climate Taiga-tundra-snow albedo feedback atmosphere, ocean, ice enhances significantly the response sheets, sea-ice, land to the orbital forcing.

surface

U 5 kyr BP

* Energy balance model Vegetation feedbacks at high and low latitudes contribute significantly to the temperature response to the orbital change. P+T+ —► warm summer —► less sea ice

—► warm winter —► taiga northward P—T—cold biome southward tundra increase —► snow albedo feedback —* glaciation

Harrison et al., 1995

Gallimore-Kut/bach,

0 0

Post a comment