How long will the Holocene last

The Holocene is a particularly long episode of stable climatic conditions compared with the three previous interglacial periods (Sirocko et al. 2007). Its stability certainly favored the establishment of modern civilizations. What explains the stability of the Holocene, and how long will it be?

Modeling the long-term evolution of climate requires taking into account changes in the spatial and seasonal distribution of incoming shortwave radiation (insolation) at the top of the atmosphere (Figure 4.4). The latter is fully determined by three parameters: eccentricity, obliquity, and climatic precession (Berger 1978). Eccentricity is a geometric measure of the stretching of the Earth's orbit. It varies between about 0.002 and 0.040 on periods of 400 000 and 100 000 years. It is presently small (0.016) and a minimum will be reached in 27 000 years.

Figure 4.4 Month-latitude distribution of incoming shortwave radiation (insolation) received from the Sun at the top of the atmosphere between 9000 years BP and 6000 years AP (after present). A mean distribution assuming no eccentricity and a mean obliquity of 23°20' was subtracted in order to highlight the effects of precession and obliquity changes. Precession redistributes heat across the seasons (positive anomaly around July 9000 years BP and positive anomaly around January at present). The decrease in obliquity during the Holocene contributes to reduce summer insolation in both hemispheres.

Insolation anomaly (W m 2) 9 kyr BP 6 kyr BP

3 kyr BP

Insolation anomaly (W m 2) 9 kyr BP 6 kyr BP

3 kyr BP

J F MAM J J ASON D Present

J F MAM J J ASON D Present

6 kyr AP

6 kyr AP

15 20 25 30

15 20 25 30

Obliquity is the angle between the Equator and the orbital plane (ecliptic). It varies between 22 and 25° with a period of 40 000 years. Obliquity has decreased by 0.8° during the past 9000 years, and it will continue to decrease over the next 10 000 years. As a consequence, the distribution of insolation is being slightly modified with (i) less insolation available to the summer hemisphere, with highest differences at high latitudes, and (ii) a decrease in annual mean insolation polewards of 43°N and 43°S, symmetric about the Equator, which compensates for a corresponding increase equatorwards of 43°. Changes in annual mean insolation are small (1 to 2 W m-2) but provisional calculations confirmed by sensitivity experiments with a comprehensive climate model reveal that the corresponding thermal forcing easily explains sea-surface temperature changes by 0.5 to 1°C (Liu et al. 2002; Loutre et al. 2004). Precession is the cycling of the angle formed by the position of the Earth on 21 March, the Sun, and the point of the orbit closest to the Sun (perihelion). The cycle takes about 21 000 years. Perihelion was reached in July 11 000 years ago. It then drifted from July to later in the year. It occurs in January today. Given that insolation decreases with the square of the Earth-Sun distance, precession during the Holocene caused a decrease in June insolation and a corresponding increase in January. Precession does not alter annual mean insolation at any latitude. The seasonal redistributing action of precession makes it an efficient modulator of seasonal weather systems, such as tropical monsoons (Kutzbach and Otto-Bliesner 1982; Harrison et al. 2003; Braconnot et al. 2004). These effects naturally become less important when eccentricity decreases as at present.

All comprehensive model experiments so far have demonstrated that orbital forcing induces significant and measurable changes in temperature, precipitation, and atmospheric circulation. These changes constitute the "fast" climate response to orbital forcing, which drives - and may be altered by - the slow components of the climate system (ice sheets, deep-ocean dynamics, ocean biogeochemistry, vegetation) over several millennia.

Conceptual models provide a convenient theoretical framework to study free and forced interactions between the slow components of the climate system. A particularly significant development is Saltzman's model of the Late Cenozoic Ice Ages (Saltzman and Maasch 1990). This three-equation model with nine parameters represents the interactions between the carbon cycle, ice sheets, and ocean circulation. Probabilistic inference with this model (Hargreaves and Annan 2002) indicates an immediate end to the Holocene, with ice volume reaching a maximum in around 60 kyr (assuming no anthropogenic perturbation). But is this prediction correct? For example, it is noted that the Saltzman-Maasch model fails to reproduce the steadily increasing trend in CO2 concentration during marine isotope stage (MIS) 11 (after termination V, 400 000 years ago) (Raynaud et al. 2005; Siegenthaler et al. 2005). Therefore, some stabilizing mechanisms may have been ignored in this model. We therefore turn to another conceptual model presented by Paillard (2001).

Paillard's conceptual model features three possible climate regimes (glacial, mild glacial, interglacial) to which the system is successively attracted depending on insolation and ice volume. Contrary to Saltzman's, Paillard's model succeeds in predicting the correct length for MIS 11 (two precession cycles). Paillard's model can then be used to predict the length of the Holocene. The result is ambiguous. The prediction can either be a short Holocene (glacial inception already begun) or a very long one (glacial inception in 50 kyr) depending on the model parameters. Yet the tested parameter values seem equally reasonable: the model provides a satisfactory fit to data of the past 800 kyr (Imbrie et al. 1984) in both cases. Only the observation that we are presently not undergoing a glaciation allows us to reject the first solution. In other words, the length of the Holocene would not have been predictable 9000 years ago with Paillard's model.

There is presently no comprehensive model or even an EMIC (see above) capable of representing the interactions between the slow components of the climate system satisfactorily enough to predict the evolution of ice volume and greenhouse gas concentrations over several glacial-interglacial cycles. There are a few EMICs, however, that are able to simulate the evolution of the atmosphere-ocean-ice-sheet system on those time-scales: the LLN-2D model (Gallee et al. 1991, 1992), CLIMBER-2 (Petoukhov et al. 2000; Calov et al. 2005), and the Toronto climate-ice-sheet model (Tarasov and Peltier 1999). The LLN-2D model is particularly suitable to study the evolution of ice volume in response to hypothetical CO2 scenarios and orbital forcing because it successfully reproduces several past glacial-interglacial cycles assuming that the evolution of greenhouse gas concentrations is correctly prescribed (Loutre and Berger 2003).

Both CLIMBER and LLN-2D consistently show no glacial inception during the Holocene when observed CO2 concentrations are prescribed (Claussen et al. 2005; Loutre et al. 2007). Other CO2 scenarios were then tested. Neither LLN-2D nor CLIMBER predicted a modern glacial inception as long as CO2 remained above 240 ppmv during the Holocene. The LLN-2D further shows that even if CO2 concentration decreased in the future down to glacial levels, glaciation will not occur before 50 000 years (Loutre et al. 2007).

The quasi-absence of precessional forcing due to the weak eccentricity may therefore explain the exceptional stability of the Holocene. This statement, however, calls for a few clarifications. First, sensitivity experiments with the LLN-2D model suggest that there is actually a window, roughly between -5000 and +5000 years from now, during which a low enough CO2 concentration (below 240 ppmv) induces a glacial inception (Figure 4.5, red line). This is consistent with sensitivity experiments with a more comprehensive model calibrated on the present-day climate showing accumulation of perennial snow for 240 ppmv CO2 and 450 ppbv CH4 (Ruddiman et al. 2005) (the methane feedback is implicitly taken into account in the LLN-2D via the model calibration on previous glacial-interglacial cycles). These results confirm Paillard's prediction that the present orbital configuration may be compatible with a glacial inception (Paillard 2001), but they also show that the inception scenario is no longer possible given the present CO2 concentration.

Second, the LLN-2D predicts a three-precession cycle duration for MIS 11 when forced by the Vostok CO2 concentrations (Petit et al. 1999), but sensitivity experiments with slightly lower CO2 concentrations result in a short MIS 11 (Loutre and Berger 2003). Paillard (2001) also found that small parameter changes may make

Next glacial inception (LLN-2D model)

m to

Figure 4.5 Prediction ofthe next glacial inception with the LLN-2D climate-ice-sheet model. The LLN-2D model is a model of intermediate complexity representing the interactions between the atmosphere, the ocean mixed-layer, and ice sheets. The CO2 concentration needs to be prescribed. This model correctly reproduces several glacial-interglacial cycles. Here, it is shown that the next glacial inception is not expected to occur before 50 000 years from present under all possible scenarios of CO2 concentration (green, blue, black) except in one scenario (red) where CO2 decreases to 210 ppmv as early as 7000 years BP. (Data are from Loutre et al. 2007.)

20 40 60 80 100 120 Time (kyr)

(Future)

20 40 60 80 100 120 Time (kyr)

(Future)

MIS 11 either short or long in his model. We note that MIS 11 eccentricity was almost as small as today (0.019 versus 0.017) with the implication that in the late Quaternary background climate small values of eccentricity are times when small climate perturbations may decide whether the climate system is sent to a long or a short interglacial. In an extreme formulation of this working hypothesis, the decisive perturbation may be so small that it is unpredictable. Another possible consequence is that glacial inception may crucially depend on the phase of obliquity when eccentricity is small (Masson-Delmotte et al. 2007).

These latter two remarks are relevant to a consideration of Ruddiman's recent proposal that the absence of an ongoing glaciation is due to anthropogenic factors (Ruddiman 2003). Millennia of moderate greenhouse gas emissions due to forest clearance and rice plantations before industrialization would have been large enough to cause a significant deviation in the natural evolution of the slow components of climate (see also Oldfield, this volume). According to Ruddiman, the natural evolution was a glacial inception, with CO2 and CH4 concentrations decreasing to 240 ppmv and 450 ppbv, respectively (they were 280 ppmv and 710 ppbv respectively before the industrialization). Ruddiman argues in later articles (Ruddiman 2005, 2006, 2007) that the major part of these anomalies is actually a feedback to a small perturbation: the ocean did not absorb as much CO2 as it would otherwise have because it warmed up. This explains the lack of a marked isotopic signature in the Holocene ice record of CO2 (noted by Joos et al. (2004) and Broecker and Stocker (2006)). We therefore arrive at an important conclusion (expressed by Crucifix and Berger 2006): the early anthropogenic theory implies - if it is correct - that there was a bifurcation point during the past 6000 years during which the climate system hesitated between opting for a glacial inception or staying interglacial. The anthropogenic perturbation gave it the necessary kick to opt for a long interglacial.

In light of the modeling works discussed above, this hypothesis cannot easily be proved or disproved. Only refined models properly calibrated by means of past data assimilation will allow us to determine if anthropogenic emissions over the past six millennia significantly increased the probability of a long interglacial. This statement should not be confused with another one: it is now granted that we are in this long interglacial and that glacial inception is no longer expected for about 50 000 years. Actually, Archer and Ganopolski (2005) estimate that 25 percent of present anthropogenic emissions will remain in the atmosphere for thousands of years and about 7 percent will remain beyond 100 000 years. They conclude that the CO2 concentration threshold below which a glacial inception occurs may not be reached before 500 000 years.

0 0

Responses

  • DUENNA
    How long Will the holocene interglacial last?
    2 years ago

Post a comment