Understanding Observed Global Climate Change

Global-scale records of surface temperatures, based on thermometer observations of surface air temperatures over land and measurements of sea surface temperatures, are available from the late 19th century onwards. From these observations, various research groups have developed global temperature datasets (see Figure 2.1) using different procedures for processing the available raw data, such as the treatment of gaps in observations (see Section 2.3.3). Based on these independently produced global temperature datasets, a best estimate of GMST has been calculated, which represents changes over both land and ocean. This estimate shows that GMST rose 0.85°C over the period 1880–2012 (based on a linear trend, with a 90% uncertainty range between 0.65°C and 1.06°C) (Hartmann et al., 2013²²). The last three full decades (1980–2010) have been the warmest on record, with the longest dataset extending back to 1850 (see Figure 2.1) (Hartmann et al., 2013²²). Global temperatures in the last three years with complete records (2015, 2016, and 2017) are the three warmest years on record for the globe on average (WMO, 2018), at more than 1°C above pre-industrial average levels (Blunden and Arndt, 2016³, 2017⁴; WMO, 2017⁸⁹, 2018⁹⁰; Hawkins et al., 2017²³).

Annual GMST has not increased in a steady linear progression since the late 19th century (see Figure 2.1). During several periods, warming was more pronounced (e.g., 1900−1940 and 1970 onwards) or less pronounced (e.g., 1940−1970). These fluctuations arise from natural variations within the climate system (internal climate variability) and outside (external) forces, including human factors (see Section 2.3.3).

Almost the entire globe experienced warming on a century scale (1901–2012). This warming was not uniform from one region of the Earth to another, owing to a range of factors, including internal climate variability, and regional variations in climate feedbacks and heat uptake (Hartman et al., 2013²²). In general, warming has been strongest at high northern latitudes and stronger over land than oceans. Since Canada has a large land mass, much of which is located at high northern latitudes, warming across Canada has been about twice the global average (see Chapter 4, Section 4).

Cold and warm extremes of temperature can have large impacts on human and natural systems. Based on multi-decadal observational datasets and rigorous statistical analysis, the IPCC AR5 reports that, for global land area as a whole, the number of warm days and nights⁴ very likely increased and the number of cold days and nights very likely decreased over the period 1951–2010. Robust statistical assessment of heat waves and warm spells is more challenging. The IPCC AR5 assesses with medium confidence that, since the mid-20th century, the length and frequency of warm spells, including heatwaves,⁵ has increased for global land areas as a whole (Hartmann et al., 2013²²). At the continental scale, it is likely that heatwave frequency has increased in some regions of Europe, Asia, and Australia over this period. For North America and Central America, there is medium confidence that more regions have experienced increases in heatwaves and warm spells than have experienced decreases (Hartmann et al., 2013²²).

Increasing global temperatures have impacts on the hydrological (water) cycle. The amount of moisture the atmosphere can hold increases with rising temperatures (about 7% per degree Celsius of warming). It is very likely that global specific humidity — a measure of the amount of water vapour in the air — near the surface and in the troposphere⁶ (see Figure 2.2) has increased since the 1970s, consistent with the observed temperature increase over this period (Hartmann et al., 2013²²).

The effects of increasing atmospheric concentrations of greenhouse gases (GHGs) on the hydrological cycle and precipitation are more complex than for temperature. Precipitation varies substantially over time and space, to a greater extent than does temperature. Long-term precipitation trends are smaller, compared to the range of precipitation variability, than are temperature trends relative to the range of temperature variability. Therefore, a greater density of monitoring stations with long records of precipitation is required for robust assessment of precipitation trends than is the case for temperature. Owing to a lack of data, there is low confidence in estimates of precipitation changes over land at the global scale before 1951 and medium confidence thereafter. Annual average precipitation for global land areas increased slightly over the period 1901–2008, and different datasets vary in the magnitude of observed changes (Hartmann et al., 2013²²). It remains a challenge to determine long-term trends in precipitation for the global oceans. At the regional scale, average annual precipitation for the mid-latitude land area in the Northern Hemisphere shows a likely overall increase since 1901, with medium confidence before 1951 and high confidence after that date (Hartmann et al., 2013²²). The changes in precipitation across Canada are discussed in Chapter 4.

As climate warming has made more moisture available in the atmosphere, this additional atmospheric moisture can lead to increased intensity of extreme precipitation events that varies by location. Observed changes in extreme precipitation are generally larger than those in total annual precipitation. At the global scale, extreme rainfall over land, measured as the number of heavy precipitation events, has likely increased in more regions than it has decreased since the 1950s. There is large variability among regions and between seasons, but the highest confidence in observed results is for central North America, where there was very likely a trend toward heavier precipitation events since the 1950s (Harman et al., 2013²²).

While changes in precipitation patterns may be expected to contribute to changes in drought and floods, there is low confidence in global-scale trends for both of these hazards (Hartmann et al., 2013²²). However, regional-scale trends are evident in some areas, with a likely increase in frequency and intensity of drought in the Mediterranean and West Africa and a likely decrease in central North America (mainly central United States but including parts of southern Canada) since 1950. Perspectives on changes in the frequency and magnitude of droughts and floods in a Canadian context are provided in Chapter 6 (see also Chapter 4, Section 4.4 for a discussion of the 2013 Alberta flood).

A number of changes observed over the past century provide evidence of a warming global ocean (Rhein et al., 2013⁶²) (see Figure 2.2). Comprehensive estimates of global mean temperatures in the upper ocean (to a depth of 700 m) reveal that warming since the early 1970s is virtually certain. The global average warming for the upper 75 m of the ocean over the period 1971–2010 was an estimated 0.11°C (90% uncertainty range from 0.09°C to 0.13°C) per decade. There is greater uncertainty in measurements of ocean temperatures before 1971 due, in part, to the scarcity of observations, but the IPCC AR5 reports that global average ocean warming (0−700 m) from the 1870s to 1971 was likely. Warming has also been observed deeper in the oceans, although the trends are not as strong. The IPCC AR5 reports that the increasing ocean heat content (absorbed heat that has been stored in the ocean; see Figure 2.2) accounts for roughly 90% of the energy accumulated globally over the period 1971–2010 (high confidence). This accumulation of energy in the ocean is strong evidence of excess energy in the Earth system, with less energy leaving the Earth system than entering (see Section 2.3.1; Rhein et al., 2013⁶²). In addition to absorbing excess heat, the Earth’s oceans have also been absorbing excess carbon dioxide (CO₂) from the atmosphere, increasing their acidity (see Chapter 7, Section 7.6.1).

Global sea level rises primarily as a result of the expansion of ocean waters due to warming (thermal expansion) and the addition of water to the ocean from land ice (glacier and ice sheets) that is delivered to the oceans by melting and increased ice discharge. Tide-gauge records from around the world and, more recently, satellite altimeter data, indicate that the global mean sea level has been rising since the late 19th century (see Figure 2.2). The level has risen by an estimated 0.19 m (90% uncertainty range from 0.17 m to 0.21 m), based on a linear trend over the period 1901–2010, and the rate of this sea level rise has likely increased since the early 20th century (Rhein et al., 2013⁶²).

Both rising global sea level and increasing ocean heat content are strong evidence of a warming world. Influences of these global changes on the oceans surrounding Canada are detailed in Chapter 7.

The cryosphere refers to portions of the Earth with sufficiently cold temperatures for water to freeze, and includes snow, sea ice, land ice (glaciers and ice caps), freshwater ice (lake and river ice), permafrost, and seasonally frozen ground. The IPCC AR5 assessed changes in the cryosphere around the globe and found, with very high confidence, that almost all glaciers worldwide have continued to shrink and that the Greenland (very high confidence) and Antarctic (high confidence) ice sheets have lost mass (based on two decades of data) (Vaughan et al., 2013⁸⁵). The IPCC AR5 reported that, over the period 2003−2009, the greatest losses of glacier ice were from glaciers in Alaska, the Southern Andes, Asian mountains, the periphery of the Greenland ice sheet, and the Canadian Arctic (Vaughan et al., 2013⁸⁵).

There is very high confidence that the extent of Arctic sea ice (both newly formed annual ice and multi-year ice) declined over the period 1979–2012, and that declines occurred in all seasons but were most pronounced in summer and autumn (high confidence). Annual mean sea ice extent in the Arctic very likely declined at a rate of 3.5%–4.1% per decade over this period. Antarctic sea ice extent very likely increased over the same period at a rate of 1.2%–1.8% per decade. The causes of variations in Antarctic sea ice properties and trends remain less well known than those for the Arctic, and the World Meteorological Association (2018) reports that, since the increase was reported in 2013, Antarctic sea ice extent was at or near record low levels throughout 2017. There is also very high confidence that snow cover extent has declined in the Northern Hemisphere (especially in spring) and high confidence that permafrost temperatures have increased in most regions since the 1980s, which is related to regional warming. Overall, the net loss in mass of ice from the global cryosphere (due to changes in glaciers, ice sheets, snow cover, sea ice extent, melt period, and ice thickness) is evidence of strong warming at high latitudes (see Figure 2.2) (Vaughan et al., 2013⁸⁵). Further details on these changes and implications from a Canadian perspective are found in Chapter 5.

In summary, these changes documented in the atmosphere, oceans, and cryosphere since the late 19th century (see Figure 2.2), as well as additional changes documented in the IPCC AR5, provide a strong, coherent picture of a warming planet based on multiple, independent lines of evidence. For this reason, warming of the climate system is robustly demonstrated; that is, it is found to be unequivocal.

Scientists have understood the basic workings of Earth’s climate for almost 200 years. Studies in the 19th century had already identified the key role of Earth’s atmosphere and of CO₂ in raising the temperature of the planet (Fourier, 1827¹⁷; Tyndall, 1859⁸³; Arrhenius, 1896¹). The fundamentals of the climate system, including factors that determine climate and that can drive climate change, have been included in every major IPCC assessment as foundational background information (IPCC, 1990²⁶, 1996²⁷, 2001²⁸, 2007²⁹, 2013a³¹).

Earth’s long-term climate and average temperature are regulated by a balance between energy arriving from the sun (in the form of shortwave radiation) and energy leaving the Earth (in the form of longwave radiation) (see Box 2.1). When this balance is disrupted in a persistent way, global temperature rises or falls. Factors that disrupt this balance are called “climate drivers” or “climate forcing agents,” evoking their influence in forcing climate toward warmer or cooler conditions. Their effect on Earth’s energy balance is called “radiative forcing,” which is defined as the net change in the energy balance of the Earth system due to an external perturbation. The strength of radiative forcing is measured in units of watts per square metre (W/m²). Positive radiative forcing indicates excess energy is being retained in the climate system — less energy is leaving than is entering the system — leading to a warmer climate, whereas negative radiative forcing indicates more energy is leaving the climate system than entering it, leading to a cooler climate (Le Treut et al., 2007⁴⁵; Cubasch et al., 2013¹⁰). Radiative forcing provides a useful means of comparing and/or ranking the influence of different climate drivers.

Climate drivers can be either natural or anthropogenic — resulting from human activities. The fact that Earth’s average temperature and climate have varied significantly over geologic time indicates that natural factors have varied in the past. On shorter timescales of decades to centuries, the main climate drivers are changes in solar irradiance, volcanic eruptions, changes in atmospheric composition, and changes to the land surface. The latter two are influenced by human activities. How changes in these climate drivers influence incoming or outgoing radiation is described below.

Solar irradiance, the strength of solar radiation received at the Earth’s surface, fluctuates by a small amount over a solar cycle of approximately 11 years, and these fluctuations can explain global temperature variations of up to approximately 0.1°C between the strongest and weakest parts of the cycle. Small, multi-decadal trends (increasing and decreasing) in solar irradiance can also occur, with similarly small effects on global climate (Masson-Delmotte et al., 2013⁴⁷).

Volcanic eruptions periodically eject large volumes of gases and dust into the stratosphere (upper atmosphere). Sulphate aerosols (tiny airborne particles) that form from these gases reflect solar radiation and thereby induce a cooling effect.⁷ Since volcanic eruptions are episodic, and sulphate aerosols remain in the stratosphere for only a few years, the cooling effects are short-lived. The global cooling effect of large volcanic eruptions, such as the eruption of Mount Pinatubo in the Philippines in 1991, is clearly evident in the global temperature record (see Section 2.3.3 and Figure 2.9).

Human activities affect Earth’s reflectivity (albedo) by changing the atmospheric composition and the land surface. For example, the combustion of fossil fuels emits a variety of pollutants, in addition to GHGs, into the lower atmosphere, where they form aerosols of various chemical compositions. These aerosols may either reflect or absorb solar radiation and are important drivers of climate change. Aerosols in the lower atmosphere also serve as particles on which water vapour can condense to form clouds (cloud condensation nuclei). Changes in aerosol concentrations can therefore induce changes in cloud properties which, in turn, can affect Earth’s albedo. While the interactions between aerosols and clouds are complex and involve a number of different processes, an increase in aerosol concentration is known to produce brighter clouds, which reflect more solar radiation, inducing a cooling effect. Human alterations of the land surface also tend to increase albedo. When forested lands are cleared for cultivation this tends to produce more reflective land surfaces (Le Treut, H. et al., 2007⁴⁵; Cubasch et al., 2013¹⁰).

Changes in solar irradiance, volcanic eruptions, and changes in albedo affect Earth’s energy balance by altering the amount of incoming energy available to heat the Earth, but the primary driver of the amount of heat leaving the Earth is changes to the chemical composition of the atmosphere. While the two most abundant gases in Earth’s atmosphere — nitrogen (78%) and oxygen (21%) — are transparent to outgoing longwave radiation, allowing this heat to escape to space, some trace gases absorb longwave radiation, creating the greenhouse effect, and are referred to as GHGs (see Box 2.1). GHGs have both natural and human sources. The main GHGs are water vapour, CO₂, methane (CH₄), ozone (O₃), nitrous oxide (N₂O), and groups of synthetic chemicals referred to as halocarbons (see Box 2.2). Changes to the atmospheric concentrations of GHGs affect the transparency of the atmosphere to outgoing heat. Individual GHGs differ in their capacity to trap heat, and most are more powerful GHGs than CO₂. However, CO₂ is by far the most abundant GHG (Myhre et al., 2013⁵²) aside from water vapour. The build-up of atmospheric GHGs has reduced heat loss to space and is therefore a positive radiative forcing, with a warming effect on the climate system (Le Treut et al., 2007⁴⁵; Cubasch et al., 2013¹⁰).

Determining the relative contribution of different forcing agents perturbing the Earth’s energy balance provides a useful first-order assessment of the causes of observed climate change (see Section 2.2). However, the climate system does not respond in a straightforward way to changes in radiative forcing. An initial perturbation can trigger feedbacks in the climate system that alter the response. These climate feedbacks either amplify the effect of the initial forcing (positive feedback) or dampen it (negative feedback). Therefore, positive feedbacks in the climate system are cause for concern because they amplify the warming from an initial positive forcing, such as increases in atmospheric concentrations of GHGs.

There are a number of feedbacks in the climate system, operating on a wide range of timescales, from hours to centuries (Cubasch et al., 2013¹⁰; see, in particular, Fig 1.2 and associated text in this reference). Important positive feedbacks that have contributed to warming over the Industrial Era include the water vapour feedback (water vapour, a strong GHG, increases with climate warming) and the snow/ice albedo feedback (snow and ice diminish with climate warming, decreasing surface albedo) (see Box 2.3). There is very high confidence that the net feedback — that is, the sum of the important feedbacks operating on century timescales — is positive, amplifying global warming (Flato et al., 2013¹⁶; Fahey et al., 2017¹⁴). Some feedbacks are expected to become increasingly important as climate warming continues this century and beyond. These include feedbacks that change how rapidly the land and ocean can remove CO₂ from the atmosphere and those that may lead to additional emissions of CO₂ and other GHGs, such as from thawing permafrost (Ciais et al., 2013⁹; Fahey et al., 2017¹⁴) (see Chapter 5, Section 5.6).

The Industrial Era refers to the period in history, beginning around the mid-18th century and continuing today, marked by a rapid increase in industrial activity powered by the combustion of fossil fuels. Burning these carbon-based fuels releases CO₂, as well as other gases and pollutants, to the atmosphere. The Industrial Era is recognized as the period when human activity has substantially affected the chemical composition of the atmosphere by increasing the concentration of trace gases, including GHGs (Steffen et al., 2007⁸⁰).

In summary, as concluded by the IPCC (IPCC, 2013b³²), total radiative forcing is positive and has led to an uptake of energy by the climate system. The largest contribution to total radiative forcing is caused by the increase in the atmospheric concentration of CO₂ since the beginning of the Industrial Era. This factor is the dominant driver of global warming and climate change over this period. Natural forcing from solar irradiance changes and volcanic aerosols made only a small contribution throughout the last century, except for brief periods following volcanic eruptions.

Even when a strong anthropogenic forcing drives climate change (see Section 2.3.2), signals of climate change may be difficult to detect against a backdrop of a climate system that is naturally chaotic – “noisy”. This chaotic behaviour is due to internal climate variability and natural external forcings, which can be large over short periods (for example, forcing by volcanic eruptions). Internal climate variability, such as El Niño–Southern Oscillation (ENSO) (see Box 2.5), is variability that arises within the climate system, independent of variations in external forcing.

Global mean surface temperature (GMST), as calculated by a linear trend, has increased significantly since the 1880s, especially since the 1950s (see Section 2.2.1). However, the changes in GMST have been far from uniform, with substantial variability among years, decades, and periods spanning several decades. These short-term fluctuations are superimposed on an underlying externally forced trend (see Figure 2.9) (Morice et al., 2012⁵¹; Karl et al., 2015³⁷; Hansen et al., 2010²¹).

To analyze the causes of the short-term fluctuations in GMST, we first need to be confident that the observed variability is real and not an artifact, an error introduced by the way the data were collected or analyzed. Long-term GMST time series have been produced by a small number of scientific teams using data collected from around the world. Values are reported as an anomaly: a departure from the average over a reference period (1961–1990 for Figure 2.9). Differences among the estimates are due mainly to differing choices made in processing the underlying raw observations. For example, one estimate (HadCRUT4.4) is an average for only those grid cells where observations exist, whereas the other estimates (NOAA-Karl and GISTEMP) use infilling; if observations are missing for certain locations, they are estimated based on values for neighbouring locations. These estimates of GMST, and others, are routinely updated as errors are identified and adjusted (see Box 4.1). Correcting and updating long-term datasets with new observations as these become available is imperative for tracking global change from year-to-year, decade-to-decade, and century-to-century.

Some of the ups and downs over time shown in Figure 2.9 are associated with the ENSO, the fairly periodic internal variation in sea surface temperatures over the tropical eastern Pacific Ocean, affecting much of the tropics, subtropics, and some areas outside the tropics, including Canada (Box 2.6). The warming phase is known as El Niño and the cooling phase as La Niña. ENSO events can be powerful enough to be recorded as significant signals in GMST. The 1997/1998 El Niño was regarded as one of the most powerful El Niño events in recorded history, resulting in widespread droughts, flooding, and other natural disasters across the globe (Trenberth, 2002⁸²). It terminated abruptly in mid-1998 and was followed by a moderate-to-strong La Niña, which lasted until the end of 2000 (Shabbar and Yu, 2009⁷⁴).

Some other ups and downs shown in Figure 2.9 are associated with natural external forcing agents, such as large volcanic eruptions. The 1991 eruption of Mount Pinatubo, in the Philippines, was the second largest terrestrial eruption of the 20th century. It ejected a massive amount of particulate matter into the stratosphere and produced a global layer of sulphuric acid haze. GMST dropped significantly in 1991–1993 (McCormick et al., 1995⁴⁸). Similarly, the 1982 eruption of El Chichón, the largest volcanic eruption in modern Mexican history, ejected a large amount of sulphate aerosols into the stratosphere (Robock and Matson, 1983⁶⁶). The cooling impact of the of El Chichón eruption on GMST from 1982 to 1984 was partly offset by global warming associated with a very strong El Niño event during this time (Robock, 2013⁶⁵).

Naturally occurring variations in GMST, whether internally generated or externally forced, should be viewed in the context of global mean radiative forcing caused by human activities (Fyfe et al., 2016¹⁸). The combined radiative forcing from human activities has increased over time (see Figure 2.9) (Meinshausen et al., 2011⁵⁰). The periods in Figure 2.9 labelled “big hiatus” and “warming slowdown” correspond to times when the dominant mode of internal decadal variability in the Pacific — the Interdecadal Pacific Oscillation (IPO) — was in its negative (cold) phase. In addition, during the “big hiatus” period, radiative forcing increased relatively slowly, owing to cooling contributions from increasing tropospheric aerosols, as well as stratospheric aerosols from the Mount Agung eruption in 1963 (e.g., Fyfe et al. 2016¹⁸). In the intervening period, the IPO was in its positive (warm) phase. A given phase, warm or cold, of the IPO typically lasts from 20 to 30 years, much longer than the timescale associated with ENSO. Recent computer models (Meehl et al., 2013⁴⁹; Kosaka and Xie, 2013⁴¹; England et al., 2014¹³) and studies based on observations (Steinman et al., 2015⁸¹; Dai et al., 2015¹¹) indicate that the IPO plays an important role in changes in GMST over time.

Finally, the “warming slowdown” — a slowdown in the rate of increase of GMST observed over the early 2000s — has been much debated (Karl et al., 2015³⁷; Lewandowsky et al., 2015⁴⁶; Rajaratnam et al., 2015⁶¹). Observations indicate that the rate of global mean surface warming from 2001 to 2015 was significantly less than the rate over the previous 30 years (Fyfe et al., 2016¹⁸). It is now understood that both internal variability and external forcing contributed to the warming slowdown (Flato et al., 2013¹⁶; Fyfe et al., 2016¹⁸; Santer et al., 2017⁶⁸). The contribution from external forcing has been ascribed to: 1) a succession of moderate volcanic eruptions in the early 21st-century (Solomon et al., 2011⁷⁸; Vernier, 2011⁸⁶; Fyfe et al., 2013¹⁹; Santer et al., 2014⁶⁹; Ridley et al., 2014⁶⁴; Santer et al., 2015⁷⁰); 2) a long and anomalously low solar minimum during the last solar cycle (Kopp and Lean, 2011⁴⁰; Schmidt et al., 2014⁷²); 3) increased atmospheric burdens of sulphate aerosols from human activity (Smith et al., 2016⁷⁷); and 4) a decrease in stratospheric water vapour (Solomon et al., 2010⁷⁹). In the last several years, GMST has warmed substantially (e.g., Hu and Fedorov, 2017²⁵), with an exceptionally strong ENSO in 2015/2016, suggesting that the warming slowdown is now over.

Establishing the causes of observed changes in climate involves both “detection” and “attribution.” Specifically, “detection” means demonstrating that an observed change is inconsistent with internal climate variability; in effect, this is a task of detecting a signal from the “noise” of background climate variability. “Attribution” means identifying the causes of an observed change in terms of different forcings (Bindoff et al., 2013²). The IPCC AR5 included a chapter (Bindoff et al., 2013²) assessing the evidence for attributing global and regional changes in a range of variables to GHG increases and other forcings. Understanding the causes of climate change on the global scale is important for understanding the causes of regional climate change discussed in Chapters 4 to 7 of this report. In this subsection, we summarize relevant findings from the IPCC AR5 assessment and more recent findings on global-scale attribution. The relatively new science of attribution of individual events, as opposed to longer-term changes, is discussed in Chapter 4, Section 4.4.

Detection and attribution studies compare observed climate changes with simulations from different types of climate-model experiments: 1) simulations of the response to external forcings of interest; and 2) simulations with no variations in external forcing that show the effect of internal climate variability. Confidence in such analyses is increased by using simulations from multiple climate models developed in centres around the world, and by validating simulated internal variability by comparison with observations. If an observed change is inconsistent with simulated internal variability alone, then a response to external forcing is detected. If the observed change is consistent with model simulations including a particular forcing, such as GHGs, and inconsistent with simulations omitting it, then the observed change is attributed, in part, to that forcing. Since more than one forcing drives trends in climate, an observed change is generally not wholly attributable to variations in one forcing. The sections below summarize attribution of observed changes in each component of the climate system.

Atmosphere and surface

The IPCC AR5 assessed contributions of greenhouse gases, other anthropogenic forcings (mainly aerosols), and natural forcings to the observed trend in GMST that increased approximately 0.6°C from 1951 to 2010, based on several studies that had assessed these trends quantitatively using detection and attribution methods. The trend attributable to combined forcings from human activities (mainly changes in GHGs and aerosols) is likely between 0.6°C and 0.8°C (see Figure 2.10) and extremely likely more than half of the observed increase (Bindoff et al., 2013²). Note that, as expected, IPCC AR5 assigned a lower likelihood level to the narrower confidence interval (0.6°C to 0.8°C) and a higher likelihood level to the broader one (greater than half the observed warming). However, when the GMST response to forcings is separated into contributions from GHG forcing and aerosol forcing, uncertainties are larger due to several factors: large uncertainties in aerosol forcing, differences in the simulated responses to these forcings among models, and difficulties in separating the response to GHG increases from the response to aerosol changes. Nonetheless, more than half of the observed increase in GMST was very likely due to the observed human-caused increase in GHG concentrations. The combined effect of aerosols from volcanic eruptions and variations in solar irradiance made only a small contribution to observed trends over this period (statistically, the contribution was not significantly different from zero). Similarly, internal variability made only a small contribution to trends over this period. Warming was also observed over the first half of the 20th century, and this warming was very unlikely to have been due to internal variability alone, but it remains difficult to quantify the contributions of internal variability, anthropogenic forcing, and natural forcing to this warming (Bindoff et al., 2013²).

Since the publication of the IPCC AR5, studies have shed further light on aspects of detection and attribution. For example, the influence of observational uncertainty on estimates of the GMST trend attributable to GHGs was found to be small relative to other sources of uncertainty (Jones and Kennedy, 2017³⁵). Another study found that considerable differences remain among models in the simulated response to forcings from human activity, particularly to non-GHG forcing (Jones et al., 2016³⁶). However, the conclusions of these studies remain consistent with the IPCC AR5 (Bindoff et al., 2013²). Even when using a novel approach to detection and attribution (Ribes et al., 2017⁶³), the assessed range for the contribution to observed warming trends from human activities remains consistent with the range in the IPCC AR5 (Bindoff et al., 2013²).

The IPCC AR5 also assessed that it was likely that forcings from human activity have contributed to warming of the lower atmosphere (troposphere) since 1961 (Bindoff et al., 2013²). Recent research continues to support this assessment. A new study found that apparent differences between the rate of warming of the lower atmosphere in climate models and in satellite observations since 1979 are smaller than previously reported (Santer et al., 2017⁶⁸).

There is medium confidence that human activities have contributed to observed increases in atmospheric specific humidity and to global-scale changes in precipitation over land since 1950, including increases in the Northern Hemisphere at mid- to high latitudes (Bindoff et al., 2013²). Large uncertainties in observations and models, and large internal variability in precipitation, precluded greater confidence. Research since the IPCC AR5 (e.g., Hegerl et al., 2015²⁴; Polson et al., 2016⁶⁰) has examined sources of uncertainty in more detail, but the overall conclusions remain consistent with those of IPCC AR5 (Bindoff et al., 2013²).

Ocean

Several aspects of observed global-scale change in the oceans have been attributed to human activity. In particular, it is very likely that human-caused forcing made a substantial contribution to upper ocean warming since 1970 and to a rise in global mean sea level since the 1970s (Bindoff et al., 2013²). It is very likely that human-caused increases in CO₂ have driven acidification of ocean surface waters, through uptake of CO₂ from the atmosphere, decreasing pH by 0.0014 to 0.0024 per year (see Chapter 7, Section 7.6.1). Recent research continues to support the attribution of ocean warming and sea level rise to human influence (e.g., Slangen et al., 2014⁷⁶; Weller et al., 2016⁸⁷), with new estimates of the heat content of the upper ocean showing a larger warming trend than that assessed in the IPCC AR5 (Durack et al., 2014¹²).

Cryosphere

It is very likely that human-caused forcings have contributed to Arctic sea ice loss since 1979 (Bindoff et al., 2013²). This conclusion was based on model simulations, which were able to reproduce the observed decline only when including human-caused forcings. There is low confidence in the understanding of an observed increase in the extent of Antarctic sea ice. However, since that assessment was made in 2013, Antarctic sea ice extent has decreased, with September 2017 sea ice extent being the second lowest on record (NOAA, 2017⁵⁵). It is likely that human-caused forcing contributed to the observed surface melting of the Greenland Ice Sheet since 1993 and to the observed retreat of glaciers since the 1960s, but there is low confidence in attribution of the causes of mass loss from the Antarctic Ice Sheet. There was likely a contribution of human activity to observed reductions in Northern Hemisphere snow cover since 1970 (Bindoff et al., 2013²). New research strengthens the evidence for attribution of the decrease in Arctic sea ice extent (e.g., Kirchmeier-Young et al., 2017³⁹), and Northern Hemisphere snow cover (e.g., Najafi et al., 2016⁵³) to human influence.

Extremes

On the global scale, it is very likely that human-caused forcing has contributed to observed changes in the frequency of daily temperature extremes since 1950, including increases in hot extremes and decreases in cold extremes (Bindoff et al., 2013²). For regions with sufficient observations, there is medium confidence that human-caused forcing has contributed to increased intensity of heavy precipitation events since 1950. New research further strengthens the evidence for attribution of changes in temperature and precipitation extremes to human influence (Zhang et al., 2013⁹¹; Kim et al., 2016³⁸; Fischer and Knutti, 2015¹⁵; Christidis and Stott, 2016⁸).