of sea-surface temperature and sea-ice conditions) are being assembled to permit increasingly detailed descriptions of the state of Holocene climates; these descriptions will aid in examination of mechanisms of climate change and geosphere-biosphere interactions, and in testing of the accuracy of model simulations.
During the Holocene, perihelion shifted from northern summer at 11 ka (11 thousand years ago) to northern fall at 6 ka to northern winter at 0 ka (Berger, 1978). Thus, the Northern Hemisphere seasonal insolation cycle was enhanced at 11 ka relative to 0 ka, and the month of largest positive insolation anomaly then moved from summer to late summer and into fall around 6 ka. Simultaneously, the Southern Hemisphere experienced a reduced seasonal insolation cycle at 11 ka, compared with 0 ka, and the month of largest negative insolation anomaly then moved into southern spring. The magnitude of these seasonal changes was as much as 25-30 Watts/m2, representing changes of 7-8% in solar radiation at 11 ka. Averaged over the annual cycle, the global and latitudinal-average insolation was unchanged by these shifts in the season of perihelion. In contrast, the slightly increased tilt of the Earth's rotational axis at 11 ka, relative to modern, caused high-latitude insolation to incresc in summer and produced a small annual-average increase of insolation of several Watts/m2 near the poles and a slight decrease of insolation in the tropics; here also, the global average insolation was unchanged. The carbon dioxide concentration of the atmosphere w as about 280 ppm at 11 ka and remained approximately constant at that level until about a.d, 1800, when the increase associated with burning of fossil fuels began (Raynaud et al., 1993), Some remnants of the large Northern Hemisphere ice sheets that had existed at the Last Glacial Maximum (LGM), about 21 ka, remained in North America and Europe at 11 ka (~30% of the LGM ice volume), but almost all this excess ice had melted by 6 ka.
This chapter will review the considerable array of Holocene climate simulations undertaken in the past 20 years. The review illustrates that in spite of progress, many questions remain unanswered about the magnitude, spatial structure, and temporal character of the response of the Earth System to the orbitally caused changes in incoming solar radiation. In some regions, terrestrial biosphere-climate interactions appear to amplify the climate change caused by the orbital forcing. Ocean-atmosphere interactions appear to amplify climate changes in some areas and reduce changcs in other areas. In many regions, the changes in Holocene climate simulated by climate models are not as large as the observed changes. This discrepancy, now documented for Holocene climates, raises the possibility that current climate models may also underestimate future climate change.
Abrupt climate changes in the Holocene are not easily explained by orbital forcing alone. However, positive feedbacks in theatmosphere-biosphere-ocean system may have caused climate to change more rapidly, in response to slow changes in orbital forcing, than would otherwise be expected. With the advent of new climate model configurations permitting efficient century or multicentury simulations, it is now possible to examine whether different phases of the Holocene might have had different variability at interan-nual, decadal, or centennial time scales; these studies may provide useful perspectives as we consider the possible character of decade/century climate variability in the present and future.
Maiepnan, 3ami/imeHHbiii asTopcKi/iM npaBOM
Mitchell et aL, 1988; Liao et ah, 1994; Phillips and Held, 1994). The simulations of Holocene climate made with these additional system components exhibited nonlinear effects of sea-ice and snow interactions. In high northern latitudes, the orbitally forced warmer summers and falls led to a later onset of sea-ice formation. With less sea ice and more open water, compared with modern, the surface albedo was lowered, and absorbed solar radiation was increased - a positive feedback. Especially over the Arctic, this effect produced additional warming in summer and fall, and in winter the ice was thinner and covered a smaller area. Spring melt-back was slightly delayed, but this effect was secondary. As a result of thinner ice and more leads, the flux of heat from ocean to atmosphere was increased, and winter temperatures were considerably warmer in higher northern latitudes than for modern conditions. The nonlinear and somewhat counterintuitive response of sea ice to the altered insolation forcing helped produce simulations that were in better agreement with observational evidence (Klimanov, 1984) than the earlier simulations with prescribed SSTs and sea ice. Atmospheric dynamical processes (horizontal advection and mixing by cyclones/anticyclones) were responsible for mixing some of the enhanced Arctic w armth onto the northern continents. Recent work illustrates an additional complexity of sea-ice response to orbital forcing. Compared with the strongly positive feedback exhibited by static thermodynamic sea-ice models, the addition of sea-ice dynamics to the models causes a small negative feedback that reduces the overall sensitiv ity of ice thickness to orbital forcing (Vavrus, 1999).
In middle and low latitudes, inserting an interactive soil moisture submodel in place of prescribed soil wetness triggered soil moisture feedbacks that acted to promote moisture recycling (a positive feedback). As a result, northern monsoon regions experienced larger precipitation increases due to orbital forcing than occurred in models with prescribed soil moisture (Ciallimorc and Kutzbach, 1989; Phillips and Held, 1994). Some northern continental interiors became even warmer and drier in the summer because of orbital forcing caused by increased evaporative losses and lowered soil moisture.
The maximum response of surface climate to orbital forcing clearly shifts forward in the annual cycle as perihelion shifts forward in season between 11 ka and 6 ka (Kutzbach et. al., 1998). Thus, the maximum increase in temperature and monsoon precipitation shifts from July at 11 ka to September at 6 ka; the total increase is less at 6 ka than at 11 ka.
10.3.3 Climate Models with Terrestrial Biosphere Interactions
Observational evidence indicates that significant changes of vegetation occurred in high, middle, and low latitudes during the Iiolocene (Wright et al., 1993). The boreal forest shifted northward, replacing tundra; grasslands expanded eastward in North America, replacing forest; and grasslands shifted northward in North Africa, replacing desert. Models have begun to be used to test how these changes in vegetation might influence the climate.
One-way Forcing: Vegetation Sensitivity Experiments. An indication of the potential response of the terrestrial biosphere to orbitally forced climate change has been gained by using the output of climate models - namely, the seasonal cycle of solar radiation, temperature, and precipitation - as input to vegetation models (Harrison ct al.+ 1995; Kutzbach et al., 1998). The simulated biome distributions for 11 ka (early Holocene) arid 6 ka (mid-Holoccnc) indicate (1) in the Arctic: northward expansion of boreal forest (taiga) replacing tundra; (2) in mid-latitude continental interiors: expansion of warm grassland /shrub replacing cool grassland/shrub; and (3) in the northern tropics: expansion of grassland and xerophytic woods/scrub replacing desert in N. Africa, and expansion of warmer and moister biome types in parts ofSE and E. Asia. Although individual models differ from one another, these results were generally consistent among 10 climate models for 6 ka orbital forcing (Harrison et at, 1998).
Having confirmed that changes of climate cause vegetation change, what about changes in the opposite direction? Are the vegetation changes sufficiently large to affect climate? Sensitivity experiments with climate models suggest that this is so. In high latitudes, prescribed replacement of tundra by forest causes lowered springtime albedo because the trees extend above the snow whereas tundra remains snow-covered. This lower albedo of the forested landscape in late winter and early spring, relative to tundra, favors an increase in absorbed solar radiation and therefore additional warmth (Bonan ct ah, 1992). Thus in the I loloccne, if the orbitally forced increase in summer warmth caused forest to replace tundra, then the lowered albedo in the winter half year could cause additional warming - a positive feedback (Foley et ah, 1994; TEMPO Members, 1996). In these forest-replacing-tundra sensitivity experiments, the additional warming due to vegetation change almost equaled the original warming due to orbital change alone. Similarly, if orbttally forced enhancements of summer monsoon precipitation in North Africa caused grasslands to invade desert, then the combination of lowered albedo and increased moisture recycling over the vegetated surface could have caused additional rainfall, a positive feedback (Street-Perrott et at,, 1990; Kutzbach et al., 1996; Brostrom et al., 1998). Changes in soil texture and water-holding capacity could also have promoted increased moisture recycling.
Changes in the size of lakes and wetlands caused by climate change may also produce feedbacks on the atmosphere (Cxtc and Bonan, 1997; Brostrom ct al., 1998; Carrington et al., in press).
Coupled Atmosphere-Biosphere Interaction. Because the results of one-way sensitivity experiments suggested that vegetation changes might exert strong positive feedbacks on climate, various groups have used coupled atmosphere-biosphere models to simulate the two-way interactions between climate and the terrestrial biosphere during the Holocene (Texier et al,, 1997; Claussen and Gayler, 1997; Pollard et al., 1998; Ganopolski et al., 1998; Foley, et al,, in press; Doherty et al,, in press). Almost all these coupled models simulate expansion of northern forests and northern subtropical grasslands during the Holocene. The models differ both in the extent of vegetation change (for example, forest biome shifts of 100 km or 500 km) and in the magnitude of the climate change (for example, sub-Saharan rainfall increases of 10% or 50%). In most studies, the positive feedback of vegetation on the African monsoon still falls short of matching the observed major extension of monsoon rainfall and grasslands into the Sahara during the early and mid-Holocene; the one exception is Ganopolski et al. (1998), who find a very large positive feedback.
respond to orbital forcing with varying feedbacks of biosphere and ocean. New modeling capabilities, when combined with increases in computer resources, should allow detailed study of these complex interactions.
10.3.5 Time-Dependent Simulations of Abrupt Climate Change
Studies of Holocene climates with three-dimensional climate models have employed relatively short "snapshot11 simulations (10-50 years' duration) because of computational limitations, in contrast, time-dependent simulations of coupled climate systems with statistical-dynamical models have been run for hundreds of thousands of years and have provided major insights about the climate response to orbital forcing (and other forcing) at glacial/interglacial time scales (Berger et al., 1992, 1998; Berger, 1998). Recently, a statistical-dynamical atmospheric model coupled to a terrestrial biosphere model and simplified ocean basin mode) was used for a 9,000-year simulation of Holocene climate (Ciaussen et aL, 1999). This simulation displayed an abrupt climate response to the gradually changing orbital forcing. The abrupt change at about 5.5 ka was traced to nonlinear vegetation feedback in the northern monsoon system as perihelion shifted from northern summer (9 ka) into northern fall, that is, beyond the monsoon maximum of the early I lolocene. At about 5.5 ka, the vegetated Sahara, which had been formed bv the orbitally forced monsoon enhancement of the earlv Holocene, reverted
to desert, accompanied by an abrupt increase in surface albedo and an equally abrupt decrease in precipitation. Lake-level evidence from North Africa indicates an abrupt decrease in moisture at about this time. Mcltwater pulses from the last stages of déglaciation could also have been a factor in abrupt changes of the thermo-haline circulation and climate in the early Holocene, such as at 8 ka (Street- Per rot t and Perrott, 1990). With increasingly efficient models and more computer resources, time-dependent simulations of Holocene climate will become more common, and opportunities for studies of abrupt climate change will expand.
10.3.6 Decade/Century Variability as a Function of Mean Climate State
A question of great current interest is whether recent changes in the character of ENSO are related to changes in the mean climate caused by greenhouse warming. If the mean climate changes, will variability change? Changes of the character of El Niño are reported in simulations of climate with increased atmospheric concentrations of CO2 (Knutson and Manabe, 1994; Meehl and Washington, 1996). Decade/century variability may also have changed in the Holocene, compared with present, because the external forcing and mean climate were different at 11 ka, 6 ka, and 0 ka (for example). High-resolution records (corals, laminated lake sediments, layered ice cores, tree rings) offer the possibility to observe such changes in variability.
Several studies have reported coupled-model simulations of Holocene climate v ariability (Hewitt and Mitchell, 1996; Otto-Bliesner, 1999; Clement et aL, 1999). ENSO teleconnection patterns may have been different at 6 ka compared with present (Otto-HI icsner, 1999). Clement et aL (personal communication) show thai the changed seasonal insolation cycle of the early Holocene might have suppressed El Nino events. Simulations with a coupled atmosphere-ocean model for 11 ka and 0 ka (Liu et a)., 2000) showed that ENSO occurred in both time periods, and with similar spatial patterns, but the overall variability of eastern Pacific SST was reduced by 10-20% at 11 ka compared with modern. One possible explanation for the early Holocene reduction in ENSO variability is related to the enhanced Asian summer monsoon, which is forced by the enhanced summertime insolation at 11 ka, The strengthened monsoon helps to produce a stronger Walker cell over the equatorial Pacific and therefore increased equatorial easterlies, enhanced upwclling, and a colder equatorial ocean. The cli ma to-logical bias toward colder conditions in the central and eastern equatorial Pacific could in turn favor less intense ENSO events (Liu et al, 2000). These first simulations of reduced interannual variability in the early Holocene appear to be somewhat consistent with preliminary paleoclimatic evidence that strong ENSO events developed only after 7 ka-6 ka (Sandweiss et al., 1996; Rod bell et al., 1999).
Observational records of Holocene climates provide a wealth of information about changes in the mean climate, abrupt changes, and changes in variability. A hierarchy of models is being used to study and help us understand the complex atmosphere-ocean-biosphere interactions that have occurred in response to orbital forcing. The results of these studies will help to inform us about the relative stability of Earth's climate and biosphere and its sensitivity to changes in external forcing, whether truly external (such as orbital) or human-related. Significant progress has been made in understanding how orbital forcing influences climate. However, this review serves to illustrate that climate models have so far been unable to simulate the full magnitude and spatial and temporal structure of Holocene climate change. Therefore, many challenges remain - both to improve the observational record and to improve the simulations and our understanding of the operative mechanisms.
ACKNOWLEDG M E N' T
The research on paleodimates ai the University of W isconsin-Madison was funded in targe part by grants from the National Science Foundation (NSF) Climate Dynamics Program and Earth
System History Program, the Department of Energy, and modeling and computational resources from the NSF-sponsored National txnicr for Climatic Research,
Was this article helpful?