4.2.3.Model projections Recent model simulations confirm earlier projections that, if atmospheric composition remains fixed at today’s level, over the next century the Earth will warm another 0.5˚C and sea levels rise another 10 cm. Although the committed warming would have reached near-equilibrium by then, sea levels would continue to rise for millennia thereafter as deep oceans and ice sheets respond to the warmer surface temperatures. However, some experts note that, since these simulations do not allow allow for effects of past volcanic eruptions, the warming that we are committed to may be significantly higher. If all offsetting aerosol effects are excluded, the estimated total committed warming relative to pre-industrial conditions could be as much as 2.4˚C. The unrealized but committed warming is also likely to be much larger in polar regions than elsewhere. For example, in the Arctic there could be an additional increase in temperature of 1 to 3˚C. Most models also show that once the system has reached an equilibrium response to current forcings, the added warming would create an El Niño like response in atmospheric and ocean circulation, significantly enhancing precipitation in some regions of the Pacific and Indian Oceans327-328.
A large number of current generation models have also been used to repeat past simulations of the climate system response to projected changes in climate forcings over the next century and beyond. Most such simulations are based on changes in radiative forcings consistent with SRES emission scenarios or equivalent. Results, in general, suggest a 50% probability that global average temperatures will rise another 1.5˚C within the next three to five decades. Warming by 2100 is projected to be in the range of 2 to 4˚C. It also appears unlikely that underestimation of past aerosol offsets have resulted in a similar underestimation of climate sensitivities, and thus of the magnitude of future warming. A number of models suggest the warming will have an El Niño-like pattern, and that Indian Ocean and Pacific temperatures will respond more rapidly than those of the North Atlantic and Southern Ocean. The magnitude of response relative to background noise will be greatest in polar and tropical regions. The warming is also expected to increase the propagation rate of planetary waves in the low-latitude Pacific Ocean, thus affecting interannual climate variability. If emissions progressively decline from IPCC SRES projected levels in 2100 to zero by 2300, surface temperatures could warm by up to another 2˚C by the time climate stabilizes in 3000. Between 15 and 28% of accumulated CO2 emissions between today and 2300 would still be airborne at that time. Meanwhile, ocean thermal expansion alone would have added about 2 meters to global sea levels270,289,329-334.
Figure 6. A climate index based on nine different types of temperature and precipitation extremes is used to show the level of significance of projected climate change over the next century for various land regions of the world. The projected changes are derived from simulations with three GCMs (ECHAM5, HADCM3 and CGCM2) using IPCC SRES A2 and B2 emission scenarios. The higher the value, the more significant the change. Most sensitive regions are primarily in high and low latitude regions (from Bättig, Wild and Imboden, 2007, ref. # 329)
There is also evidence that temperature variability in high latitudes will likely decrease, largely due to the attenuating effect of reduced duration of the ice cover season. Other projected changes include reduced temperature variability over the equatorial Pacific and small increases in variability in mid to high latitude summers. The latter implies risk of more frequent and longer lasting heat waves335.
All models project a decrease in convective mass fluxes, enhanced horizontal moisture transport, decreased sensible heat transport in the extratropics and enhanced evaporation-precipitation patterns. As the troposphere warms, the tropopause rises and the middle atmosphere between 30-70 km above the surface cools. This increases ozone and water vapour concentrations in the middle atmosphere. The rise in the tropopause and other secondary factors would cause the zonal jet streams, transient kinetic energy and momentum mass to strengthen and shift poleward. Subtropical static instability also expands the dry zone of Hadley circulation system. The response of the Arctic Oscillation appears to be complex and non-linear, but generally increases sea level pressures over the Gulf of Alaska and lowers them over eastern Canada and the North Atlantic303,336-340.
Average global precipitation is projected to increase by between 1 and 3% for each degree of surface warming. Some experts suggest that this may be a significant underestimation. Models also indicate that the precipitation response to rising temperatures will be weaker in warm climates than in cold regions. In the extra-tropics, the integrated water column is projected to change in a manner consistent with the constant relative humidity equation. Snow seasons become shorter everywhere, and the total amount of snow fall is expected to decrease in mid-latitudes. However, snow fall is projected to increase in the high latitudes of the Northern Hemisphere. Changing circulation patterns, including a weakening of both the Walker and Hadley circulation systems, will also weaken poleward transport of moisture and precipitation and will reduce the intensity of the Asian monsoons299,341-342.
The combined effects of an enhanced hydrological cycle and changes in atmospheric circulation patterns will alter regional precipitation patterns and water resources in complex ways. In general, precipitation and runoff are projected to increase significantly in high latitudes, southeast Asia and central Africa. Mean precipitation deposited by the Asian monsoon is projected to increase by 8% by 2100, with some Middle East and central Asian regions receiving as much as 24% more than today. Winter snow fall is projected to increase over north-eastern Eurasia and decrease over western Europe. This would also affect mid-latitude circulation patterns over Eurasia, and may be a factor in the projected changes in the Asian monsoons. Increased storms associated with the poleward shift of storm tracks, like that over the North Pacific Ocean, can also cause significant local increases in downwind precipitation. Substantial decreases in precipitation are expected in the Mediterranean, south Africa, south North America and central America343-345.
Figure 7. Projected changes in annual mean precipitation runoff by 2100, in mm/day. Projections are based on a weighted average of results from 19 different model experiments, using radiative forcing derived from the IPCC SRES A1B emission scenario (Nohara et al., 2006, ref. #344).
The primary factor affecting the stability of the North Atlantic thermohaline circulation system appears to be changes in surface freshwater influx from precipitation and related river runoff. When the freshwater flux increases, it reduces the density of waters from the Greenland-Iceland-Norwegian Seas flowing into North Atlantic, and hence the rate of deep water formation in the region. When it decreases, the reverse happens. Fresh water fluxes in sub-polar waters and air sea heat flux changes are secondary factors. Surface salinity is likely to increase for much of the rest of the North Atlantic. Greenland ice sheet meltwater is only expected to have an important influence on these changes under the most extreme melt rate scenarios. Recent simulations indicate that, while enhanced greenhouse gas forcing tends to increase such high-latitude freshwater input into the North Atlantic, increasing aerosol concentrations have the reverse effect. These two opposing forcings may have largely offset each other during the past half century, leaving the North Atlantic thermohaline circulation system relatively unaffected. However, all model results appear to indicate that, as forcing due to increased greenhouse gas concentrations progressively dominates other forcings in future decades, fresh water flux from increased rainfall in higher latitudes will cause the ocean circulation system to slow down. The system may shut down completely if the freshwater addition exceeds one Sv (106 cubic meters per second) . Model outputs disagree on the rate and magnitude of the slowdown over the next few centuries, and on its reversibility in subsequent centuries. Most suggest that the weakening of the ocean overturning will be temporary provided future greenhouse gas emissions follow the lower SRES scenarios. However, some advanced models that include positive carbon cycle feedbacks indicate a collapse of the Atlantic overturning is likely under high SRES emission scenarios, and possible for the intermediate scenarios. Some experts suggest that if concentrations reach 4xCO2, the circulation collapse could last for millennia. The impacts of a collapsed thermohaline circulation would interact with those due to the other physical climate system responses to enhanced greenhouse gases in complex ways, including significant decreases in temperatures in western Europe and North America. A weaker or arrested Atlantic overturning would have major impacts on atmospheric circulation and precipitation patterns, and would thus indirectly affect Pacific Ocean circulation. These changes could intensify ENSO behavior346-357.
Several studies have also tried to predict the natural short term variability of the North Atlantic circulation in order to estimate how it might modify the effects of enhanced greenhouse gas forcing over the next decade. One such study, using a simplistic method, suggests that the recent increase in Atlantic overturning rates is in the process of reversing, bringing rates back to their normal state. This would cause a cooling effect over the North Atlantic and Europe that would offset much of the effects of greenhouse gas warming for the next decade or so. In contrast, another also predicts that natural variability is currently offsetting anthropogenic warming influences, but that this offset will reverse soon and then begin to add to the anthropogenic effects, with a net increase in global temperatures between 2004 and 2014 of 0.3˚C. It projects that half of the years between 2009 in 2014 will be hotter than the current record year of 1998358-361.
Many climate models are now much more realistic in their simulations of current ENSO events. However, the model simulations still disagree on how future climate change will affect the mean ENSO state. Some suggest these events will become weaker while others project stronger events. This disagreement arises because the two key factors that affect ENSO behavior – SST response to thermocline and wind variability, and cloud and wind feedbacks – are projected to have large but offsetting responses to warmer climates that largely cancel out. There does appear to be some agreement, however, on the pattern of effects that future ENSOs will have on North American weather. Because of weakened and shifted pressure patterns associated with ENSO events, the warming anomaly over high latitudes that they cause will likely be weaker than today, and the cool anomaly over the southern regions will intensify362-366.
Results from an earth system model used to successfully simulate glaciation and deglaciation processes of the past suggest that adding the effect of GHG forcing over the next few centuries to that of natural solar forcing will likely delay the onset of the next global glacial period by 50,000 years or more277.
A number of modeling studies have also explored the effect of anticipated climate forcings on regional climates. Multiple model projections for the Antarctic/Southern Ocean region, when weighted on basis of model performance in predicting current climates, suggest an average regional warming over the next century of 0.34˚C per decade, an increase in total precipitation (particularly over eastern continental regions), increased intensity of circumpolar westerlies, and a total decrease in ice cover of 33%. Stronger winds are projected to change ocean circulation. The largest sea ice reductions are projected for the Weddell and Amundsen-Bellingshausen Seas. Stronger southward flowing currents in the Ross Sea may delay the rate of sea ice reduction in the region. For the Arctic, AR4 global model simulations project winter season warming of up to 14˚C by 2100, a rapid sea ice decline resulting in virtually ice free summer conditions by 2100 under high and medium SRES emission scenarios, and a freshening of Arctic Ocean waters because of increased river runoff and decreased ice formation. This freshening, however, is unlikely to directly affect the North Atlantic thermohaline circulation system because of offsetting effects of decreased ice melt in the North Atlantic. Studies using an improved Canadian RCM nested within the Canadian coupled climate model, while also projecting a dramatic increase in Arctic temperatures, imply only modest precipitation changes in most areas of the Arctic by 2050367-374.
Model simulations also show varying results for response of the climate over the Amazon to future radiative forcing. Some simulations suggest increased rainfall over the region, others a decrease. A key reason for the uncertainty in the region`s precipitation response is the differences in model projections of ENSO behavior and of magnitudes of warming in the southern Atlantic, both of which affect regional moisture fluxes. The role of aerosol forcing in altering north-south Atlantic Ocean temperature gradients and on climate patterns in the Amazon region relative to that of greenhouse gas forcings is also a factor that is changing with time. Resolving these differences in Amazon precipitation projections is particularly important because of the related implications for the role of the Amazon forests in a strong carbon cycle-climate feedback231,375.
5.0 TRENDS 5.1 Paleoclimates 5.1.1 Pre-Holocene Arctic ocean sediment records confirm that atmospheric CO2 concentrations, together with other factors such as stratospheric ice clouds and ocean heat-salinity feedbacks, have played an important role in modulating the Earth`s climate during the past 55 million years. New studies suggest average Arctic Ocean temperatures were about 23 to 24°C at the beginning of this period, coincident with high CO2 concentrations and a much more active global hydrological cycle. About 10 million years later, CO2 concentrations had dropped substantially and ocean temperatures decreased to mean values of about 10°C. This coincided with the onset of Arctic sea ice cover and Antarctic glaciations8,376.
By the early Pliocene, some 3-5 million years ago, net radiative forcing of the Earth`s climate appears to have been very similar to that of today. However, polar regions of the Northern Hemisphere were still significantly warmer, most of the continents remained ice free and sea levels were some 25m above current levels. One of the reasons for the difference in climates from that of today despite similarities in greenhouse gas forcing may be tied to ocean circulation patterns that were in a persistent El Niño like mode, with a deep thermocline that kept warm high latitude ocean waters near the surface. The progressive cooling of global climate over subsequent millennia eventually appears to have triggered the onset of deep water formation at high latitudes, bringing much colder waters to the surface and dramatically cooling polar regions377.
Ice cores extracted from polar ice sheets, as well as other proxies from lower latitudes, have helped provide a much more detailed picture of the changes in global climates of the past one million years, and the processes involved. In Antarctica, for example, new deep ice cores now provide a comprehensive record of regional temperatures, atmospheric composition, dust loading and other climate factors for the past 880,000 years. These records show that a dominant 100 thousand year (ky) cycle between glacial and interglacial periods has persisted throughout the period, although warming during the interglacials in the early part of the record was weaker. Some experts suggest this weaker warming during early interglacials may imply a lower frequency cycle of 400-500 ky length that modulates the behavior of the 100 ky cycle. Superimposed on the 100 ky cycle, in turn, are shorter cycles of 42 ky and 23 ky associated with the Earth`s orbital precession and obliquity. The data also indicate that local temperatures and atmospheric CO2 concentrations follow changes in O2/N2 ratios (a proxy for summer solar insolation) by several millennia. This is consistent with the dominant theory that these cycles are related to changes in the Earth`s orbit and hence to patterns of solar insolation. Dust concentrations in the Antarctic cores are also about 25 times greater during peak glacial periods than during interglacials. Similar evidence also appears in ocean sediment cores of the equatorial Pacific. High dust loading enhances nutrient supply to oceans, thus increasing removal of CO2 from the atmosphere through biological processes. This, together with the reduced amount of solar radiation reaching the surface due to a less transparent atmosphere, would have helped enhance the glacial cooling. Isotopic and model data both suggest that this resulted in Antarctic temperatures during the last glacial maximum some 13°C colder than the current interglacial5,378-381.
Evidence from other regions of the world indicates that these slow cycles from glacial to interglacial conditions were global in nature. Data from the bottom of Greenland ice cores as well as from marine sediments in adjacent Atlantic waters indicate that much of southern Greenland was vegetated during the last interglacial, providing a habitat for a variety of insects. It also appears to have been covered with boreal forest during the long, warm interglacial 400 ky before present. Arctic coast lines also indicate sea levels during these interglacials some 4-6m higher than today, with Greenland and Antarctic ice sheet contributions believed to be of similar magnitude. However, there is also biotic DNA evidence from ice core bottom sediments that remnants of the ice sheet have been present continuously in southern Greenland for at least the last four glacial cycles382-385.
Simulations with improved climate models that now include dust and vegetation feedbacks indicate that, on a global scale, the temperature range from peak glacial to peak interglacial periods was between 3 and 7°C. This is more than one degree greater that past estimates based on model simulations without these feedbacks. These simulations also suggest that changes in climate during the Last Glacial Maximum 21,000 yeas ago were significantly amplified in polar regions386-388.
The quality of the ice core records are also good enough to reveal climate fluctuations on much shorter time scales of millennia or less. Particularly prominent are the abrupt hemispheric scale temperature fluctuations during the last glacial period, referred to as Dansgaard-Oeschger (D-O) events and Heinrich events, which appear to be linked to periods of sudden cessation of the North Atlantic Deep Water (NADW) formation. Careful comparison of the Greenland ice core record with a new higher resolution Antarctic record indicates a strong inverse relationship between Antarctic warm events and the cold phase of D-O events in the Northern Hemisphere. The intensities of Antarctic warm events appear to be linked to durations of the cold phase of D-O events. Both the Greenland and Antarctic records also show strong CO2-temperature correlations during these anomalies. Experts suggest this likely occurs because the slowdown in the NADW formation associated with the D-O events causes a reduced CO2 uptake into oceans, and thus a rise in atmospheric CO2 concentrations. The circulation slowdown also causes the Southern Ocean to warm and become stratified. The higher CO2 causes a global warming that is not enough to offset the effects of the D-O event in the north but increases temperatures in the south. Eventually the warming Antarctic causes greater sea ice melt and once again reduces ocean stratification. Atmospheric CO2 decreased again after the D-O event was over. Changes in Antarctic temperatures during the D-O events lagged the CO2 changes, consistent with a see-saw effect. However, investigators acknowledge that these processes may be different during interglacials, and therefore may not be a good proxy for future climates9,10,387,389-392.
The nature of these millennial scale oscillations is, in many ways, complex. Temperature changes tended to be greatest in winter in high latitudes and the North Atlantic, but during spring elsewhere. Methane concentrations rose about a half century after the onset of changes in temperature, consistent with a delayed land response. There are also indications that the effect of the slowdown of NADW formation extends through the Southern Ocean into the Pacific Ocean, causing regional collapse of ecosystems there. Following the most recent such abrupt D-O event (the Younger Dryas), some of the affected species did not return for 5000 years. Other data sources also imply these events affected east Asian monsoonal activity and Antarctic sea ice cover through complex and non-linear processes, and that climates were less stable during their transition stages97,393-398.
While there is general agreement that freshwater pulses into the North Atlantic may be a key factor in starting these abrupt anomalies, there is still considerable speculation as to what triggers the freshwater pulses. Some experts suggest that the estimated 1470 year return period of D-O events is evidence that this may be linked to a sub-harmonic response to solar forcing combined with system noise. However, others argue that Greenland ice core data do not support this hypothesis399-401.
Land and marine proxy data indicate that the Laurentide ice sheet covering central and northern Canada experienced its first rapid melt period between 11 and 8.5 thousand years before present (kybp), after the end of the Younger Dryas event. During this early phase of the Holocene, shifts in the currents of the North Atlantic also caused a regional peak in sea surface temperatures off coastal North America of about 6°C, relative than today. By comparison, surface water temperatures along Canada’s eastern shores experienced little change. During this period, the North Atlantic experienced a minimum sea ice cover, and the Canadian high Arctic was about 3°C warmer than today, causing major changes in regional ecosystems and adjacent ocean circulation. This period appears to have ended coincident with the sudden drainage of Lake Agassiz ~8.4 kybp and related changes in the North Atlantic. The second phase of rapid Laurentian ice sheet melt occurred between 7.6 and 6.8 kybp. During the next few millennia, Baffin Island in Canada’s north became ice free and July temperatures across North America reached their Holocene maximum. The late Holocene experienced a decrease in Northern Hemispheric temperatures (until the 20th century). Large scale and lengthy droughts occurred with regularity across mid North America during this final stage of the Holocene, with some evidence of linkage to sunspot behavior402-408.
Meanwhile, in the Antarctic region, large ice shelves surrounding the ice sheet also experienced major retreats during the early and mid Holocene. There are indications that these were linked to changes in Southern Ocean circulation patterns, with ice shelves affected by both top and bottom melting. One exception was the large Larson B ice shelf, which has been stable throughout the Holocene, until its recent collapse in 2002409.