The quality of multiple proxy data and the statistical techniques used in recent studies to reconstruct climates of the Earth’s surface over the past two millennia have improved considerably. These improvements have helped to address criticisms of past attempts at such reconstruction. Results of these studies replicate the instrumental period of the record well, and continue to support past conclusions that the temperatures of the extratropical region of the Northern Hemisphere during the past half century are very likely unprecedented in at least the past 1300 years, and possibly longer. Results for tropical and Southern Hemispheric regions are less certain, primarily because of the sparseness of data. Improved analysis of tree ring records also suggest that current average temperatures over the land areas of the extratropical Northern Hemisphere are about 1.3°C warmer than that of the Little Ice Age (the coldest of the past 1300 years) and 0.7°C warmer than that of the Medieval Warm Period (the previous warmest), both of which appear to be linked to natural solar forcing. Borehole studies in the Canadian high Arctic also show evidence of a pronounced warming since the mid 19th century (although the high degree of spatially variability in the data suggests that local factors must be carefully considered when analyzing the data). In general, these proxy indicators, together with related model studies confirm that the warm climates of the past few decades cannot be solely explained by long term natural variability, but are consistent with anthropogenic forcing related to rising greenhouse gas and aerosol concentrations in the atmosphere410-421.
Figure 8. Reconstructed trends in Northern Hemisphere temperatures, using multiple proxies. While details vary from study to study, all indicate that temperature anomalies during the past half century are without precedence within at least the past millennium. (Mann et al., 2008, ref. #415) These proxy-based investigations also indicate that other aspects of climate, particularly water resources, have changed dramatically in the past, often with major consequences for ecosystems and human societies. For example, in Pakistan, where summer monsoonal intensity is closely linked to regional temperatures and solar radiation, the past century was the wettest of the past millennium. Since wet periods are also periods of food abundance, this has greatly benefited the population there. In contrast, in the atmospheric subsidence regions of central America, large scale droughts linked to changes in atmospheric circulation and periods of high tropical temperatures during the last two to three millennia caused the collapse of ancient civilizations, including those of the Olmecs and the Mayans. Diatom data from lake sediments in the southern Canadian Prairies indicate three major periods of long-term shifts in precipitation regimes in that region during the past 6000 years, the most recent occurring almost two thousand years ago. Within North America, multi-decadal megadroughts linked to intense La Nina events, occurred regularly throughout the past millennium, but more frequently during the first half than the second. The 20th century was likely the wettest of the past 700 years. In the north-east Canadian Arctic, many northern lakes that have been present for millennia have begun to dry up in recent decades, with catastrophic effects on ecosystems within them422-432.
5.2 Trends of the past century 5.2.1 Temperature Global Trends. Analyses of global surface temperature data sets that have been corrected for various biases related to factors such as network densities, systematic observational changes and urban heat island effects indicate that mean global temperatures are now about 1°C higher than in the late 19th century. Correction of a cold bias introduced into sea surface temperatures in the mid-1940s as a result of poor representation of ship data in the final years of the Second World War has removed much of the mid-century cold anomaly in the Northern Hemisphere data reported in previous analyses. These and other improvements in the ocean part of the data base result in a new estimate for global sea surface temperature rise since 1850 of about 0.71°C and 0.64°C in the Northern and Southern Hemispheres, respectively. Most of this rise occurred in recent decades. However, despite these improvements, some critics still indicate that there are unresolved concerns about the quality of land data433-438.
Although there is abundant new evidence that the troposphere has warmed significantly in recent decades, there remains controversy about the magnitude of this warming and its regional distribution, particularly in tropical regions. Key factors in this controversy are the different methods used to select and analysis the related satellite and radiosonde data. However, most studies, using improved methodologies and data from radiosondes and satellite-based MSU sensors corrected for known biases, now suggest average tropospheric warming of 0.2°C/decade since 1979, very similar to that at the surface. The upper troposphere may be warming more rapidly, with one study using calculations based on thermal wind measurements suggesting a warming rate at the top of the troposphere during the past few decades of as much as 0.65°C/decade. Corrected radiosonde data over Antarctica also suggests a mid-tropospheric warming in that region between 1971 and 2003 of 0.7°C/decade. These results are, in general, also in good agreement with reanalysis outputs and model simulations439-448.
Satellite instruments show strong cooling trends in the mid-stratosphere and at higher levels that are consistent with model projections of the impacts of enhanced greenhouse gas concentrations and depleted ozone levels. In the thermosphere, some 400 km above the Earth’s surface, this cooling has caused atmospheric density to increase by 1.7% per decade over the past three decades. This could increase to 2.7%/decade during the next solar cycle, impacting satellite orbits due to the added drag449-451.
The heat energy content of oceans also continues to rise. Various analyses of trends in the upper 700 or so meters converge on an average increase since 1960 of about 0.24 W/m2, with considerable variability over time and space. Some regions, such as the North and Baltic Seas, have warmed much more than this, particularly in summer, causing noticeable ecological changes in these waters. In general, these increases in the upper ocean are larger than that projected by climate model simulations. On the other hand, recent corrections for biases in bathythermographic data have resulted in revised estimates for total heat gains in the deeper ocean layer between the surface and three km depth between mid 1950s and early 1990s that are almost 40% less than that previously estimated. Despite these corrections, experts caution that there are still significant concerns about the quality of the ocean data used in these analyses. One example that underscores these concerns is a recent report that approximately 20% of the upper ocean heat gains since the mid 1950s had been abruptly lost between 2003 and 2005. A subsequent reanalysis found errors in both the data used and the calculations involved, and that the report was erroneous288,452-461.
Analyses of temperature-depth profiles across Northern Hemisphere land areas also show a net terrestrial heat gain over the last 50 years of 4.8 x 1021 J 462.
Regional Trends.The pattern of recent changes in temperature at the continental and sub-continental scale is much more complex, largely due to the added impacts of regional scale climate variability and landscape influences.
Within Canada, average surface temperatures have increased by 1.2°C since 1955, with the greatest warming in winter and spring seasons, and in western regions. There is little evidence of a change in the diurnal temperature range for the country as a whole. Local effects and other factors have modified changes at the regional scale. Reduced winter ice cover on Lake Superior, for example, has led to earlier stratification of lake waters in the spring and contributed to an increase in summer surface water temperatures of 2.6°C since 1976. Over the Prairies, local albedo and hydrological changes associated with reduced farm fallowing may have contributed to a regional decrease in daily maximum temperatures, diurnal range and solar radiation in early summer, accompanied by an increase in rainfall. Changes in the North Pacific ocean circulation since the late 1970s also appear to have contributed to altered atmospheric circulation that enhanced winter warming in the MacKenzie River Basin by as much as 4.7°C over the past 60 years. Meanwhile, in the Arctic, both instrumental data and vegetation proxies indicate that the eastern high latitude region of North America, which had been cooling during the 1970s, is once again in a warming trend. There are indications that the dramatic changes observed there during the past decade have occurred at least partly due to an abrupt northeastward shift in the centers of the AO/NAO circulation patterns, which in turn appears to have triggered a major change in Arctic atmospheric circulation patterns. However, there is as yet no clear evidence of the amplified warming in this region as projected by model studies, perhaps because of the critical non-linear role of sea ice in such amplification. Recent dramatic changes in sea ice may now change that. One study, based on analysis of ERA-40 reanalysis data, has also suggested that enhanced latent heat transport into the region has caused the Arctic troposphere to warm much more rapidly than surface temperatures. Others, however, have challenged both the evidence for amplification and the suggested causes, arguing that the ER-40 is poorly suited for trend analysis and that snow and ice feedbacks dominate over latent heat transport processes184,463-473.
The entire high-latitude Southern Hemispheric region between 50 and 90°S has warmed by 0.37°C since 1958. Temperature changes were largest during winter seasons, rising by 0.17°C/decade. Recent studies also confirm that there has been significant warming over the Antarctic Peninsula during this period, but that changes in the Antarctic interior have been relatively minor. Composite model simulations replicate this behavior. They further project that the warming will also become most significant in the interior over the next century474-475.
Greenland has also been warming rapidly during the past decades, with average temperatures rising by 1.7C between 1991 and 2006, with the four warmest summers all occurring since 2001. Some parts have experienced increases of up to 3°C since the early 1980s. However, some researchers argue that stronger rates may have occurred during the first half of the century1920s476-479.
5.2.2 Hydrological Precipitation and soil moisture. Recent trends in both global surface specific humidity and in total atmospheric moisture content over oceans appear to have been increasing significantly in a manner consistent with a constant relative humidity response to enhanced radiative forcing. While past reports have also suggested that global cloud cover has been increasing, those trends now appear to be artifacts of incorrect interpretation of satellite viewing geometry, and thus erroneous 221,480-481.
There has been an overall global increase in both average precipitation and in soil moisture during the past half century. The largest increases in soil moisture are found in North America. There are multiple factors that can affect the regional and temporal characteristics of these changes. For example, in North America, long term variations are significantly influenced by changes in ENSO behavior and possibly by the Atlantic Multi-decadal Oscillation. In the tropics, non-uniform sea surface temperature response to an enhanced greenhouse effect can affect regional atmospheric circulation and hence precipitation patterns. Some argue that, in heavily polluted regions such as north-eastern Europe, a primary influence is the effects of aerosols on solar irradiance, and thereby indirectly affecting the regional hydrological cycle. Direct effects of rising CO2 concentrations on plant respiration can also affect regional hydrological cycles, particularly in the tropics. However, while increased precipitation may tend to increase soil moisture, there is now evidence that, since 1970, increased evaporation due to warmer temperatures have begun to offset the precipitation effect, especially in high northern latitudes. Furthermore, despite the rise in soil moisture, major droughts still occur with regularity – and increasingly so in some regions. During the past century, some 30 regions of the world have experienced periods of abrupt net decreases in precipitation of 10% or more, and serious droughts conditions. In recent decades, regions of west Africa and south Asia have been among the most significantly affected by such droughts. Meanwhile, there has also been a significant increase in winter snow over much of Asia that appears to be unprecedented in at least the past millennium431,482-487.
Atmospheric moisture content over Canada has increased since 1955, particularly over the west and the Great Lakes region. Humidity increases have been slightly greater at night than during the daytime. Analysis of data from climate stations in southern Quebec and New Brunswick also indicate a significant increase in winter precipitation for much of those regions. However, recent trends towards reduced moisture transport into the Arctic appear to have reduced winter cloud cover across much of the region. Changes in western Arctic precipitation have also contributed to significant changes in lake area of the Tuktayuktuk Peninsula adjacent to the Beaufort Sea (increasing between 1978 and 1992, and decreasing since). Meanwhile, many lakes and ponds in the high Arctic that have been present for millennia are now drying up during summers, causing collapse of aquatic ecosystems within them429,472,489-490.
Across the US, there has been a significant increase in the presence of warm moist air masses in recent decades, and a decrease in dry ones. Although some of this relates to natural variability in atmospheric and ocean circulation patterns, these trends are also consistent with model projections of anthropogenic forcing. Despite this increase in average atmospheric moisture, mean duration and frequency of prolonged droughts have increased significantly in much of continental United States during the past 40 years491-492.
While global net water discharge from rivers into the oceans appears to have increased by almost 8% over the past century, the changes during the past half century appear to be small. That is because large increases in some regions have been offset by decreases elsewhere. Rivers in northern Europe, for example, show large increases in discharge, while some mid-latitude rivers show as much as a 60% decrease. While some of these changes can also be linked to human management of water drainage, most appear to be caused by changes in precipitation patterns. Changes in evapotranspiration due to the direct effect of rising CO2 concentration on vegetation respiration may also be a factor. For the Arctic Ocean basin, the large increase in European river discharge has only been partially offset by decreases from North American rivers, with a net increase of 5.6 km3 per year between 1964 and 2000. This rate of net discharge may also be accelerating. Such estimates are somewhat larger that projected by models responding to enhanced anthropogenic forcings, and imply that models may have underestimated related changes in Arctic Ocean salinity and freshwater flux into the North Atlantic Ocean493-498.
Figure 9. Changes in discharge of global rivers between 1951 and 2000. Shaded areas indicate water scarce regions with less than 100 mm/year in annual runoff. (Milliman et al. 2008, ref. #496). In north-western Canada, melting permafrost appears to be contributing to an increase in groundwater discharge into the Yukon River over past few decades. These ground waters also add more inorganic carbon to the river water, and decrease its average organic carbon and nitrogen content499.
5.2.3 Sea Ice Arctic. Since 1979, Arctic Ocean ice cover has been declining by between 3.4 to 4%/decade. This rate of decline accelerated to about 10%/decade after 1996, resulting in record low conditions by 2007 that were 37-38% below the climatological average. Less than 5% of the Arctic ice is now 7 years old or older. The large reduction in multi-year and second year ice has resulted in an estimated 50% decrease in the average late summer ice thickness in the transpolar drift region of the Arctic Ocean since 2001. Submarine data indicate that ice thickness had already been reduced by about 1.25 m during the preceding two decades. Consequently, while total winter ice extent in the Arctic has not changed significantly, it is on average much younger and more likely to melt in the following summer. The zone of oceans that fluctuates seasonally between ice cover and ice free waters has also expanded at an accelerating rate. Winds have pushed much of the remaining old Arctic ice against the Canadian high Arctic islands and into the eastern Beaufort Sea, where mean ice thickness has increased slightly as a result. Much of the old Arctic ice grown in situ amongst the Canadian high latitude islands has also melted or broken up and moved out. However, more of the old ice from the Arctic Ocean ice pack is now entering the Canadian Arctic straits. As a result, despite the increased melt, there has been little net change in old Arctic ice within these straits over the past 40 years. Hence, old ice will likely continue to be a problem in the region until all such ice disappears from the Arctic Ocean. Experts suggest that, if current Arctic Ocean ice melting trends continue, that could occur as early as 2040, much sooner than previously estimated. It would take several years of well below normal cold temperatures or a dramatic reduction in old ice export from the Arctic Ocean to restore ice cover to former levels500-514.
Figure 10. Comparison of observed seasonal sea ice extent in the Northern Hemisphere in 2005, 2007 and 2008 with mean values for five year intervals between 1980 and 2004. (Adapted from Comiso et al., 2008, Ref #501. The rapid decline in Arctic ice cover in the past decade has significantly exceeded projections by models. This has raised many questions about causal mechanisms, and the ability of models to replicate them. The mechanisms involved both melting from the bottoms and tops of ice surfaces and changes in ice dynamics associated with altered atmospheric and ocean circulation patterns. Oscillations in Arctic atmospheric circulation patterns, such as the Arctic Oscillation (AO) and the Northern Annual Mode (NAM), can explain as much as 60% of the variability of late summer ice concentrations in the Arctic Ocean ice pack. However, their role in long term trends in ice decline over the past three to four decades is more controversial. Experts do agree that increased anticyclonic circulation has been exporting more of the high Arctic multi-year sea ice into the marginal seas, where it melts more rapidly, and through Fram Strait into the North Atlantic Ocean. Furthermore, excessive summer melt of first year ice means that old ice is no longer being replaced. Circulation changes and ice melt processes may also be interlinked through complex feedbacks. For example, in the summer of 2007, the development of anomalous anticyclonic circulation over the Arctic, in addition to causing wind induced ice movement, brought in air masses with warmer temperatures, reduced humidity and lower cloud cover. While not necessarily the dominant factors in ice loss that year, both the higher air temperatures and increased solar insolation were likely contributors. A wind induced increase in influx of warm Pacific Ocean waters into the Arctic may also have been an important factor in enhanced bottom melt and hence ice loss in the late 1990s. The albedo feedback is also important. Reduced ice cover decreases local albedo, enhancing the summer warming of air and water, and thereby contributing to ice melt. During the summer of 2007, for example, the large open water presence is estimated to have caused a 500% increase in solar absorption in the near surface waters of the Beaufort Sea that accelerated bottom melt and ice retreat. Finally, some suggest that increased storm activity, also indirectly linked to changed circulation patterns, may also be implicated in some of the regional changes249,514,515-527.
Changes in Arctic ice conditions are now beginning to impact land and marine regional ecosystems and the global climate system. For example, the expanded seasonal ice zone (that occupied by first year ice) has resulted in increased salinity in surface waters below areas of new seasonal ice formation. This in turn affects ocean circulation. Modelling results also suggest that rapid retreat of sea ice will impact the state of permafrost inland. Such impacts are discussed more extensively in section 6 of this report508,517,528.
Antarctic. Southern Ocean sea ice cover change has been much less dramatic than that for the Arctic. While ice concentrations are decreasing in some regions, increases elsewhere exceed these reductions. As a result, despite the rise in regional air and ocean temperatures, total ice extent has still been growing at a long term average of about 1%/decade. Related modeling studies suggest a negative regional feedback may be an important factor. These studies indicate that, when warmer climates reduce regional sea ice growth, reduced salt rejection decreases local surface ocean salinity and hence convective overturning. This reduces the upwelling of warmer deep waters, reducing bottom sea ice melt more than the original reduction in ice growth501,529-530.
5.2.4 Land ice and Sea Level Rise Greenland Ice Sheet. A variety of studies using surface measurements, satellite gravity and altimeter data and models confirm that the Greenland ice sheet has been melting at an accelerating pace, and will likely continue to decline in volume. While all studies agree that net ice volume loss has been large in recent years, the range of estimates for this loss is also large. Some studies imply loss rates as low as about 100 km3/year, others as high as 249 km3/year. The latter is equivalent to a sea level rise of 5 mm/decade, and is significantly higher than that simulated by ice sheet models. This implies that, if rates are indeed at the high end of the above range, the models are missing or inadequately capturing some key processes involved. Much of the Greenland ice loss was caused by a large and accelerating increase in ice surface melt at lower elevations that is highly correlated with increases in air temperature. The area of surface melt increased since 1988 at an average rate of about 1.3%/year, although total melt at lower elevations was partially offset by increased snow accumulation at high elevations. However, during the anomalously warm summer of 2007, melt area was about 60% greater that the second highest melt year (1998), and expanded for the first time to elevations above 2000m. The rate of volume loss implies that more than just warmer air temperatures may be involved. Some experts suggest that melt water, as it penetrates through the ice, also lubricates and enhances flow of outlet glaciers, increasing ice loss through iceberg calving. Others suggest variability and trends in ocean temperatures near the outlet glaciers may also affect calving rates through bottom melt of the glacier face. Finally, glacier flow also appears to respond to non-linear interior processes that cause natural fluctuations in flow rates531-551.
Figure 11. Annual estimates for total Greenland ice sheet mass balance (TMB), as well as surface mass balance (SMB) and ice discharge (D) components of the total ice mass budget. TMB is the different between SMB and D. Dark blue circles indicate TMB values estimated from observations, while light blue diamonds indicate linear regression model estimates for TMB for years with missing discharge data. Green circles represent annual anomalies in SMB for 1958–2007, while red squares and pink triangles represent interpolated and observed anomalies, respectively, in D. (Rignot et al, 2008, ref. #548).