Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, Climate Science. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 26 June 2022

Sun-Climate Connectionsfree

Sun-Climate Connectionsfree

  • Judith L. LeanJudith L. LeanUnited States Naval Research Laboratory


Emergent in recent decades are robust specifications and understanding of connections between the Sun’s changing radiative energy and Earth’s changing climate and atmosphere. This follows more than a century of contentious debate about the reality of such connections, fueled by ambiguous observations, dubious correlations, and lack of plausible mechanisms. It derives from a new generation of observations of the Sun and the Earth made from space, and a new generation of physical climate models that integrate the Earth’s surface and ocean with the extended overlying atmosphere. Space-based observations now cover more than three decades and enable statistical attribution of climate change related to the Sun’s 11-year activity cycle on global scales, simultaneously with other natural and anthropogenic influences. Physical models that fully resolve the stratosphere and its embedded ozone layer better replicate the complex and subtle processes that couple the Sun and Earth.

An increase of ~0.1% in the Sun’s total irradiance, as observed near peak activity during recent 11-year solar cycles, is associated with an increase of ~0.1oC in Earth’s global surface temperature, with additional complex, time-dependent regional responses. The overlying atmosphere warms more, by 0.3oC near 20 km. Because solar radiation impinges primarily at low latitudes, the increased radiant energy alters equator-to-pole thermal gradients, initiating dynamical responses that produce regions of both warming and cooling at mid to high latitudes. Because solar energy deposition depends on altitude as a result of height-dependent atmospheric absorption, changing solar radiation establishes vertical thermal gradients that further alter dynamical motions within the Earth system.

It remains uncertain whether there are long-term changes in solar irradiance on multidecadal time scales other than due to the varying amplitude of the 11-year cycle. If so the magnitude of the additional change is expected to be comparable to that observed during the solar activity cycle. Were the Sun’s activity to become anomalously low, declining during the next century to levels of the Maunder Minimum (from 1645 to 1715), the expected global surface temperature cooling is less than a few tenths oC. In contrast, a scenario of moderate greenhouse gas increase with climate forcing of 2.6 W m−2 over the next century is expected to warm the globe 1.5 to 1.9oC, an order of magnitude more than the hypothesized solar-induced cooling over the same period.

Future challenges include the following: securing sufficiently robust observations of the Sun and Earth to elucidate changes on climatological time scales; advancing physical climate models to simulate realistic responses to changing solar radiation on decadal time scales, synergistically at the Earth’s surface and in the ocean and atmosphere; disentangling the Sun’s influence from that of other natural and anthropogenic influences as the climate and atmosphere evolve; projecting past and future changes in the Sun and Earth’s climate and atmosphere; and communicating new understanding across scientific disciplines, and to political and societal stakeholders.


  • Climate Systems and Climate Dynamics
  • Future Climate Change Scenarios

Earth’s Energy

The Sun alone, among all astronomical objects, defines the Earth as we know it. Solar radiant energy establishes the basic thermal structure of the surface, ocean, and atmosphere that makes Earth habitable. “The Sun’s rays are the ultimate source of every motion which takes place on the surface of the Earth” (Bahcall, 2000; Herschel, 1883).

The Sun, a middle-aged star approximately 4.5 billion years old, radiates electromagnetic energy from X-ray to radio wavelengths and beyond with a spectrum similar to that of a black body at 5770 K. This is the approximate temperature of the Sun’s visible surface, its photosphere. At an average distance of 150,000,000 km (1 Astronomical Unit, AU) from the Sun, the Earth receives (at the top of its atmosphere) a total solar radiative energy flux of 1361 W m−2, termed the total solar irradiance. Figure 1 shows the wavelength dependence of this energy flux, which peaks at visible wavelengths; approximately 99% of solar radiant energy is at wavelengths longer than 295 nm, which as Figure 1 also shows, penetrates the atmosphere to heat the troposphere and surface.

Figure 1. Shown in (a) is the solar irradiance spectrum compared with the spectrum of black body radiation at 5770 K. Also shown is the spectrum of solar radiation that reaches the Earth’s surface. The wavelength ranges of the primary atmospheric gases that absorb solar radiation are indicated. The change in the spectrum from high solar activity (February 2000) to low solar activity (July 2008) is shown in (b) in energy units and in (c) as percentages. The dashed line in (c) indicates the corresponding change (1 W m−2) in the total (spectrally integrated) irradiance.

Earth’s surface and atmosphere reflect, absorb, and scatter solar photons by different amounts at different altitudes; these processes, like solar irradiance itself, depend strongly on wavelength. More solar energy is deposited at lower latitudes because of the zenith angle dependence of the incident radiation. This establishes a global scale equator-to-pole thermal gradient that regional inhomogeneites such as land-ocean heating differences alter zonally. Heated globally to 255 K by primarily visible and near infrared solar energy, Earth itself radiates longer-wavelength infrared energy (near 10 micron), which “greenhouse” gases in the atmosphere absorb and reradiate, warming the surface a further 33 K. The principal greenhouse gases include water vapor, carbon dioxide, methane, nitrous oxide, and ozone. In equilibrium, Earth’s surface temperature is the consequence of the balance between incoming shorter-wavelength solar energy and outgoing longer-wavelength thermal energy (Hartmann, 1994). At 288 K (15oC) Earth’s average global surface temperature is in the “Goldilocks” zone—“just right” for biosphere habitability (Nisbet & Fowler, 2003).

Fortunately for the biosphere, Earth’s atmosphere completely absorbs biologically harmful solar radiation at ultraviolet wavelengths shorter than 295 nm (Bais et al., 2015). This absorption controls the vertical deposition of the Sun’s energy in the atmosphere, indicated in Figure 2 by the optical depth, and establishes the average vertical thermal structure also shown in this figure. When molecular oxygen absorbs solar ultraviolet radiation at wavelengths less than 242 nm it dissociates to oxygen atoms and heats the atmosphere above 85 km. Ozone, formed when molecular and atomic oxygen combine, absorbs solar ultraviolet radiation at longer wavelengths. This energy heats the atmosphere between 15 and 50 km (the stratosphere) and produces the peak in the atmospheric vertical temperature profile near 50 km evident in Figure 2. The tropopause occurs where stratospheric heating reverses tropospheric cooling with altitude.

Figure 2. Shown on the left are average height profiles of atmospheric temperature and ozone concentration. The shaded region indicates the altitude of unit optical depth for solar radiation incident overhead. On the right are time series of global measures of recent surface and atmospheric climatology. Shown are a) total ozone, (b) atmospheric temperature near 20 km (lower stratosphere), (c) atmospheric temperature near 5 km (middle troposphere), and (d) surface temperature. Colored lines indicate the observations. The black lines are representations of the climatology according to statistical models of the observations.

A perturbation in the flow of energy through the Earth system forces a new equilibrium temperature as the system adjusts to balance incoming and outgoing radiation (Stevens & Schwartz, 2012). Changes in solar radiation produce such a perturbation, as do changes in the concentrations of atmospheric greenhouse gases and volcanic and industrial aerosols. Climate system adjustments—“feedbacks”—to these “forcings” collectively define climate’s sensitivity, κ‎, which is likely between 0.5oC and 1oC per W m−2 forcing (IPCC, 2007). A radiative forcing, Δ‎F, produces a corresponding equilibrium surface temperature change, Δ‎T= κΔ‎F, when all feedbacks reach their full expression. Feedbacks involving water vapor and clouds rapidly alter convective flows and transfer heat on time scales of days to weeks. Large-scale flows that couple the atmosphere and ocean, such as the Hadley and Walker circulations, adjust to altered thermal gradients on times scales of months to years (Lucarini & Ragone, 2011). It may take decades to centuries for feedbacks such as sea ice, snow cover, and the deep ocean to reach their equilibrium responses (Hoffert, Callegari, & Hsieh, 1980).

Earth is changing on global scales. The time series in Figure 2 chronicle Earth’s climatology in the past three decades. Globally, the average temperature at the surface is warming, the troposphere less so, and the lower stratosphere cooling. Imposed upon these multidecadal trends are variations on time scales of months to decades, with distinctly different temporal signatures at the surface and in the lower stratosphere. Why is Earth’s climate and atmosphere evolving in this way, and what role does the Sun’s changing radiation play?

Unruly Sun

That the Sun is not a static astronomical object has long been known. In the early 17th century, Galileo detected dark sunspots moving across its disk when observing the Sun with the newly invented telescope. In the mid-19th century, Schwabe announced that the number of these sunspots waxed and waned with a cycle of 11 years. Hufbauer (1993) chronicles the history of the Sun since Galileo. As the sunspot number record in Figure 3 shows, the amplitude of the 11-year sunspot cycle varies considerably, with anomalously low epochs in the 17th and 18th centuries, called the Maunder and Dalton minima, respectively, and overall higher amplitude in the 20th-century Modern Maximum. In 1874 the Royal Greenwich Observatory (RGO) began monitoring the Sun’s visible disk using photographic plates, recording changes in the areas and locations of sunspots, as well as in bright regions—called faculae—until 1976 (Willis et al., 2013). The driver of the sunspot cycle and solar activity is a subsurface dynamo that causes sunspots to emerge at high heliographic latitudes at the beginning of a new cycle, and move steadily to lower latitudes.

Figure 3. The sunspot number during the past four centuries is shown in (a), determined for two different realizations of historical observations. The Maunder and Dalton minima, identified, are two epochs of lower-than-average solar activity, as indicated by low sunspot numbers for a few decades. The image on the left is a drawing by Galileo of sunspots on the Sun’s surface; on the right is a modern solar image showing dark sunspots and bright faculae. Shown in (b) is the total solar irradiance reconstructed from indices of solar activity, including the two sunspot number records in (a), calibrated against contemporary space-based observations. Also shown in (c), for comparison, is the concentration of carbon dioxide in the atmosphere during the past four decades.

Observations of certain emission and absorption features in the solar spectrum began in the 20th century, leading to new understanding of physical aspects of the Sun’s surface and atmosphere. Observations of emission in the near-ultraviolet Ca K Fraunhofer line capture the distribution of bright plage that overlie their photospheric counterparts, the visible faculae. Observations of the Zeeman splitting of solar spectral features, such as visible-wavelength Fe lines, identify the magnetic polarity of sunspots. These emblematic ground-based observations of locally enhanced emission and the Sun’s magnetic field characterize the inhomogeneous distribution of the Sun’s brightness in dark sunspots and bright faculae, and their relationships to similarly inhomogeneous magnetic fields. Like sunspot numbers, all exhibit pronounced 11-year cycles in their occurrence and strength.

Multiple ground-based time series of spatially and spectrally resolved observations of the Sun now exist. They diagnose wide-ranging aspects of the changing Sun, and contribute to understanding the causes of the Sun’s changing radiative output. Many of the initial observations extend in one form or another to the present. There are also reliable ground-based measurements of irradiance indices (i.e., averaged over the solar disk and projected to Earth at 1 AU). One of the longest such records is the radio flux at 10.7 cm, monitored since 1946 (Tapping, 2013), which changes by a factor of two or more during the 11-year solar activity cycle. Its variation is indicative of changes in the Sun’s outer atmosphere (chromosphere and corona), rather than its lower atmosphere. Additional ground-based programs began in the late 1970s to specifically measure the variability of “Sun-as-a-star” (i.e., its global output, aka irradiance), in key, targeted wavelengths such as the Ca K emission feature (Livingston, Wallace, White, & Giampapa, 2007). As well, ground-based observations of Sun-like stars were undertaken when it became evident that stars of similar mass and age exhibited activity cycles similar to the Sun’s (Hall, Lockwood, & Skiff, 2007).

But Earth’s atmosphere renders ground-based observations incapable of elucidating intrinsic changes in the Sun’s total radiative output, which is the physically relevant quantity for interpreting changes in Earth’s climate and atmosphere. As recently as 1976 the Sun’s total irradiance was effectively considered to be constant (the Solar Constant), for lack of reliable information to the contrary. Even four decades of mountain-top observations in the first half of the 20th century were inconclusive because of the interference of the Earth’s atmosphere; subsequent statistical analysis of ground-based solar irradiance measurements failed to detect variability at levels of 0.3% (Foukal et al., 1997).

Instruments launched on balloons, aircraft, and rockets avoid atmospheric extinction and extend spectral irradiance measurements to shorter ultraviolet wavelengths. These observations were able to detect, in the 1950s, that the Sun’s ultraviolet emission spectrum was sufficient to explain the formation of the ozone layer. Balloon, aircraft- and rocket-based observations continued in the 1960s and 1970s and beyond (Friedman, 1986) but the measurements were not able to measure the solar “constant” or the solar spectrum with the needed accuracy to determine its true absolute value or to detect real changes.

In summarizing a comprehensive community assessment, White (1977) reported the most probable magnitude of the total solar irradiance as 1373±20 W m−2 and that “it is sadly true that at present the limits on (its) temporal change . . . are poorly known . . . when we look at the solar spectral irradiance measurements with the goal of stating the absolute fluxes and their degree of variability, we find the time record to be seriously incomplete, even for one solar cycle.” The total solar irradiance is now known to be 1361±0.5 W m−2 (Kopp & Lean, 2011) and the tens of percent variability in the ultraviolet spectrum estimated from rocket observation in the 1970s is now known to be a factor to 5 to 10 larger than the true UV spectral irradiance changes during the solar cycle. The detection of real changes in total and spectral irradiance awaited regular observations by high precision solar radiometers on space-based platforms.

Searching for Sun-Climate Signals

Growing evidence for an inconstant Sun motivated enthusiastic pursuit of solar signals in terrestrial variables in the 19th century. Fundamental recognition of the Sun as the source of Earth’s energy encouraged belief that such signals should—indeed must—be present in observations of weather, climate, and the Earth’s atmosphere. “Since we are the children of the Sun, and our bodies a product of its rays . . . it is a worthy problem to learn how things earthly depend on this material ruler of our days. But . . . We know so little of the way it does many things here that we are still often only able to connect the terrestrial effects with the solar cause by noting what events happen together” (Langley, 1888).

Initially, the search for Sun-climate signals focused on time scales of the 11-year cycle, which the sunspot record demonstrated to be a primary mode of solar variability. Sunspot numbers were correlated with assorted terrestrial phenomena. Nevertheless, collective evidence remained ambiguous about the reality of connections between the Sun and climate. Correlations found in terrestrial records often disappeared or changed sign in different epochs. Exacerbating the confusion was the lack of knowledge of how solar irradiance—both total and in the UV spectrum—actually changed since this could not be conclusively detected in ground-based measurements.

Despite sustained effort, reliable linkages of Earth’s changing climate to the changing Sun able to survive peer scrutiny failed to materialize from early investigations of ground-based datasets. By the mid-20th century the idea of the Sun influencing the weather or climate, or even the ozone layer, was in disrepute due to inadequate justification of the relationships. Research in subsequent decades would reveal that even a few decades of ground-based, site-specific observations such as the level of Lake Victoria or Central England temperature lacked the duration and stability to establish reliable Sun-climate connections. Ironically, modern statistical analyses of longer records have subsequently reaffirmed the Sun’s solar cycle influence on lake levels in Lake Victoria (Stager, Ruzmaikin, Conway, Verburg, & Mason, 2007) and possibly also on Central England temperature in winter (Lockwood, Harrison, Woollings, & Solanki, 2010; but see also Lilensten, Dudok de Wit, & Matthes, 2015).

Toward the end of the 20th century the pursuit of connections between the Sun and climate shifted from the solar cycle to longer time scales, motivated by new circumstantial evidence that the Sun’s changes were likely larger on time scales much longer than the 11-year activity cycle. Contrary to the prevailing belief by solar physicists at that time, that the solar cycle was the primary manifestation of solar variability, examination of historical sunspot archives established that the Maunder Minimum was an epoch of real solar quiescence (Eddy, 1976). Furthermore, cosmogenic isotope indices of solar activity archived in tree-rings and ice-cores tracked the long-term sunspot record, capturing the Spörer (1460–1550), Maunder (1645–1710) and Dalton (1790–1830) minima of anomalously low sunspot activity. And these episodes coincided with colder than average temperatures in Europe during the Little Ice Age from 1450 to 1850 (Eddy, 1976).

Responding to renewed impetus to understand how changes in the Sun’s output may cause changes in the Earth’s climate and atmosphere, the USA National Research Council undertook a comprehensive exploration of possible physical mechanisms. While recognizing the lack of concrete evidence, the resultant publication, Solar Variability, Weather and Climate (NRC, 1982), recommended that “The question of possible solar influence on weather and climate should be treated as part of the more general problem of the Sun’s effects on the atmosphere as a whole . . .,” including the development and testing of physical models and expanded and strengthened databases.

As debate over the causes of 20th-century climate change intensified in the 1980s and 1990s, with recognition of a probable influence of increasing concentrations of anthropogenic gases, the search for Sun-climate signals reemerged with vigor. Longer-term solar changes, represented by smoothed sunspot numbers or length of sunspot cycles, were shown to correlate with Earth’s temperature in the past 150 years (Friis-Christensen & Lassen, 1991; Reid, 1991). On longer time scales, climate was reported to track changes in the Sun inferred from cosmogenic isotope records (Jirikowic & Damon, 1994; Stuiver & Braziunas, 1993).

The suggestion that changing solar output might account for much of Earth’s temperature increase in the 20th century, and global warming a recovery from the Little Ice Age rather than a response to increasing greenhouse gas concentrations, enjoyed brief, widespread notoriety. A number of studies rapidly challenged this hypothesis (Kelly & Wigley, 1992; Schlesinger & Ramankutty, 1992). Solar changes, although poorly known, were likely modest compared with well-measured increases in greenhouse gases. A consequence of attributing essentially all of Northern Hemisphere surface temperature change in the past century to the Sun was to negate any contribution from increasing greenhouse gas concentrations. This was inconsistent with physical climate models that readily simulated significant warming to this anthropogenic influence. Lack of knowledge of true long-term changes in solar radiative output and the inability of climate models to replicate aspects of the proposed Sun-climate connections confounded resolution of the debate (Physics Today, 2008).

The inability of 20th-century research to secure unambiguous evidence for an influence of solar variability on Earth’s climate led many to believe that there was, in fact, no influence, at least on time scales of the 11-year solar cycle. Hoyt and Schatten (1997) assessed the main arguments against the Sun’s influence on climate and concluded that climate models suggested solar cycle effects to be so small as to be undetectable, whereas on longer time scales a lack of observations and theory precluded knowing how the Sun changes and climate responds, since terrestrial Sun-climate correlations were indeterminate.

Contemporary Sun-Climate Connections

Modern satellite observations revolutionized understanding of how the Sun and Earth are changing, and the linkages between them. When launched on space-based platforms, solar- and earth-viewing radiometers and spectrometers circumvent Earth’s atmosphere entirely, thereby achieving unprecedented accuracy and repeatability. Earth-observing instruments additionally achieve global coverage, unattainable from ground-based observations. Ever-lengthening datasets of the radiative output of the Sun “as a star” and global-scale changes in Earth “as a planet” facilitate investigation of the Sun’s impact on the Earth’s climate and atmosphere with a fidelity not possible using ad hoc site-specific correlations and sunspot numbers (Benestad, 2006; Lilensten et al., 2015).

Recognition that Earth might be changing because of human activities galvanized the acquisition of sustained, coordinated, validated space-based observations of the Sun and Earth that continues today. Attributing warming of the globe to increasing concentrations of greenhouse gases necessitates knowledge of how the Sun’s total radiative output, which heats the Earth, also changes. Detecting and understanding how chlorofluorocarbons released into the atmosphere deplete Earth’s protective ozone layer necessitates knowledge of how solar ultraviolet irradiance, which both produces and destroys ozone, also changes.

The imperative to address such fundamental societally relevant challenges as global warming and ozone depletion forged a new discipline of Earth System Science in the 1980s. Characterized by the so-called Bretherton diagram, depicted in Figure 4, the Earth System is recognized to be a myriad of interactions among the land, ocean, atmosphere, and biosphere, with the Sun a driving external influence (Mooney, Duraiappahb, & Larigauderiec, 2013). In the United States, Congress mandated regular reports on the state of the ozone layer and established the U.S. Global Change Research Program (USGCRP), which the U.S. National Research Council proceeded to review (NRC, 1990). Solar Influences on Global Change (NRC, 1994) was the seventh of seven interdisciplinary science elements of the USGCRP. NASA designated solar irradiance variability 1 of 22 key variables whose changes are crucial for Earth Science. The Intergovernmental Panel on Climate Change, which included activities of all nations, published its first report (Houghton et al., 1990).

Figure 4. The “Bretherton” diagram is a schematic that identifies the multiple connections among Earth’s atmosphere, surface, ocean and biosphere. The Sun, indicated on the left-hand side, is an “external forcing” of change in the coupled system.

The space era of high fidelity global-scale observations of the Sun-Earth system began in earnest in late 1978 with the launch of the Nimbus 7 spacecraft carrying a complement of instruments that measured total solar irradiance and Earth’s radiation budget, atmospheric ozone and temperature, aerosols and winds, until 2000. Launched soon after, in 1980, the Solar Mesosphere Explorer (SME) initiated continuous space-based observations of solar UV irradiance, ozone, and other atmospheric parameters in the stratosphere and mesosphere (Barth et al., 1983). Building on the experience of these early missions, the Upper Atmosphere Research Satellite, launched in 1992, observed both the Sun and the Earth with state-of-the-art instrumentation (Dessler et al., 1998), seeking to properly establish how and why the stratosphere changes, including how it responds to changing solar UV radiation (Woods et al., 1996).

Many instruments on numerous space platforms launched by a number of countries now provide information about the Sun’s changing radiative energy and Earth’s changing environment (King, Parkinson, Partington, & Williams, 2007), characterizing and investigating the interconnections of the Bretherton diagram. Multiple observations of solar irradiance, in addition to those from Nimbus 7, SME, and UARS, have been made from space-based platforms during the 37 years since 1978 (Rottman, 2006). These include solar radiometers on the Solar Maximum Mission (SMM), ACRIMSAT, Solar Heliospheric Observatory (SOHO), and the Solar Radiation and Climate Experiment (SORCE, Rottman, 2005; Lean, Rottman, Harder, & Kopp, 2005). Figure 5 shows the solar irradiance datasets. Space-based observations of the changing Earth are also lengthening and expanding. The time series of Earth’s global temperature in Figure 2 is the product of a sequence of Microwave Sounding Units (MSU) flown on multiple satellites (Christy et al., 2007). An alternative processing of the MSU observations (Mears, Wentz, & Thorne, 2012; Mears, Wentz, Thorne, & Bernie, 2011) produces an overall similar time series that has a slightly more positive long-term trend. The time series of global total ozone in Figure 2 is a composite of observations made by Solar Backscatter Ultraviolet (SBUV) instruments, including and subsequent to that on Nimbus 7 (McPeters, Bhartia, Haffner, Labow, & Flynn, 2013). An alternative total ozone record compiled from observations by Total Ozone Mapping Spectrometers (TOMS, McPeters, Labow, & Logan, 2007) shows similar variability but with a more positive long-term trend (Lean, 2014).

Figure 5. Contemporary observations of solar irradiance are shown since 1978, compared with the sunspot number, shown in (a). Shown in (b) are daily values of total solar irradiance made by multiple independent instruments on various space-based platforms, in (c) are concurrent observations of solar Lyman α‎ irradiance at 121.5 nm (the strongest emission line in the Sun’s ultraviolet spectrum, Figure 1), in (d) solar ultraviolet radiation in a band from 200 to 250 nm, which Earth’s atmosphere absorbs, and in (e) solar visible-wavelength radiation in a band from 500 to 600 nm that is incident at Earth’s surface.

Unequivocal Solar Irradiance Variability

Evidence of the Sun’s changing radiative output came rapidly and conclusively following the launch of sophisticated radiometers on earth-orbiting satellites. Such radiometers measure the Sun’s total radiative energy incident on a black “cavity” with an aperture of known area in terms of equivalent electrical power. They detected significant decreases in total irradiance on time scales of days to weeks that were soon traced to the transit of dark sunspots across the Sun’s disk (Hudson, Silva, Woodard, & Willson, 1982), the same motions that Galileo observed almost 400 years previously. Space-based spectroradiometers that disperse incident solar radiation in wavelength also detected solar UV irradiance changes during the Sun’s 27-day rotation, but with opposite phase to total irradiance. These changes approximately tracked the ground-based 10.7 cm flux (Donnelly, Heath, Lean, & Rottman, 1983) and were traced to the passage of bright regions on the Sun’s disk, namely the chromospheric plage and photospheric faculae evident in ground-based solar images of Ca K emission. Figure 6 illustrates typical solar irradiance changes on time scales of days to months. The variations occur primarily because the Sun’s 27-day rotation alters the distribution of sunspots and faculae on the hemisphere of the Sun projected to Earth. Sunspot numbers are shown for comparison.

Figure 6. The Sun’s 27-day rotation modulates solar radiative energy output at Earth and indices such as the sunspot number. Because these changes occur on time scales somewhat shorter than that of sensitivity drifts in space-based instruments, they are relatively reliable determinations (compared with solar cycle time scales) of true solar irradiance changes. Daily sunspot numbers are shown in a) for two epochs of solar activity, in 1993 and 2003. Shown in (b) are simultaneous changes in total solar irradiance, demonstrating the out-of-phase relationship with sunspot numbers on solar rotation time scales. In (c) are observations of solar Lyman α‎ irradiance at 121.5 nm and in d) solar ultraviolet radiation in a band from 200 to 250 nm, showing that ultraviolet irradiance varies in phase with sunspot numbers. The solar rotation modulation of solar visible-wavelength radiation in a band from 500 to 600 nm, shown in (e), like total solar irradiance, varies out-of-phase with sunspot numbers during solar rotation.

The overall decline in total solar irradiance from 1980 to 1986, when initially observed, proved more elusive to confirm and understand than had the shorter-term changes related to the Sun’s rotation. On these longer time scales, difficulties in tracking in-flight instrument calibration changes can lead to spurious trends in observations. It was speculated, briefly, that the Sun’s energy output might be systematically declining but the decrease in total solar irradiance was soon associated with an overall decrease in solar activity; as solar activity increased from 1986 to 1990 so too did total solar irradiance (Foukal & Lean, 1988). Whereas dark sunspots are the dominant cause of short-term total solar irradiance decreases, over the longer duration of the solar cycle an overall increase in facular emission more than compensates sunspot-produced decreases, producing a net increase in total solar irradiance at cycle maximum. Multiple irradiance observations confirm the reality of solar cycle increases in total and ultraviolet solar irradiance, concurrent with increasing sunspot number, 10.7 cm flux, and various plage and facular indices (Lean, 1991).

An 11-year cycle is apparent in all the individual total irradiance datasets in Figure 5b with amplitude about 0.1%, higher when sunspot number is higher. Even though uncertainties in the absolute values of total irradiance reported by individual radiometers produce a spread in the datasets in Figure 5b in excess of 0.7%, the radiometers’ calibrations are nevertheless sufficiently stable to detect irradiance changes during the solar cycle. The total irradiance cycle manifests from corresponding cycles in solar spectral irradiance, with pronounced wavelength dependence, as shown in Figure 1.

Except at the shortest ultraviolet wavelengths (e.g., the Lyman α‎ emission at 121 nm, Figure 5c), definitive characterizations of solar spectral irradiance variability from direct observations is not yet possible. This is partly because the independent measurements lack overlap, precluding detection of the sought-for variability due to insufficient calibration accuracies. Uncertainties in absolute values produce the offsets among independent measurements of solar spectral irradiance evident in Figures 5c, d, and e. It is also partly because the spectroradiometric calibrations are not sufficiently stable, and this lack of repeatability produces spurious drifts in the time series (Lean & DeLand, 2012). Solar cycle changes in spectral irradiance are therefore inferred from coordinated analysis of observations and models (Lean et al., 1997).

Most of the solar spectrum, including the visible radiation that is the dominant source of Earth’s energy, varies over the solar cycle in phase with sunspot numbers. Solar irradiance at 600 nm increases about 0.06% during the solar cycle, at 200 nm by 6%, and in Lyman alpha emission at 121.5 nm by of order 50%. However, solar radiation at near-IR wavelengths, formed in the deepest layers the solar atmosphere, may vary out of phase with sunspot numbers. This occurs because the increase in facular emission near solar maximum is insufficient to counter the concurrent emission deficit in sunspots. It is apparent in Figure 1c, for example, at wavelengths between 1600 and 2200 nm (Coddington, Lean, Pilewskie, Snow, & Lindholm, 2016) and may extend over a larger wavelength range, from 1300 to 2400 nm (Krivova, Balmaceda, & Solanki, 2007).

As solar irradiance observations lengthened, composite records were constructed by cross-calibrating and combining assorted individual observations (Fröhlich & Lean, 2004); the search for longer-tem solar irradiance changes began. As Figure 7 shows, different composite records of total solar irradiance have somewhat different long-term trends for the duration of observations. They disagree, as well, about the level of total solar irradiance during solar cycle minima. Since sunspots and faculae are small or absent at times during minima of the Sun’s activity cycle, might inter minima irradiance changes signify solar irradiance changes on time scales longer than the 11-year cycle? Discrepancies among total irradiance composite records thus far preclude knowing the answer. That neither of the long, continuous records of the 10.7 cm radio flux or the Ca K index measured from ground-based observatories indicate large inter-minima trends further argues against the reality of such trends in solar irradiance in the past three solar cycles.

Figure 7. Shown in (a), (b), and (c) are three long-term observational records of total solar irradiance during the last three solar cycles, constructed by cross-calibrating selected independent observations from among those shown in Figure 5b. For comparison, two models of the variations in total solar irradiance arising from sunspot and facular features are shown in (d) and (e). The lines correspond to long-term trends in each of the four time series, determined for common daily values during the period 1984 to 2014. A primary cause of differences in the slopes is that the values of irradiance during solar cycle minima differ among the time series. Since sunspot and facular influences are minimal during solar cycle minima, these differences during cycle minima are the consequence of different assumptions about the observations’ cross-calibrations and in-fight sensitivity drifts, in the case of the observational composite records in (a), (b), and (c), and of different relationships of solar indices to irradiance in the case of the models in (d) and (e).

Sun, Climate, and Troposphere

A solar cycle signal was not expected to be found in observations of Earth’s global surface temperature. Energy balance model simulations indicated that surface warming from an increase of 1.1 W m−2 in total solar irradiance during the 11-year cycle would be 0.025oC globally, and therefore beyond detection (Wigley & Raper, 1990). But observational evidence abounds for solar cycle-related increases in the temperature of the ocean, surface, and lower atmosphere, of order 0.1oC globally. The signals are detected in a variety of observations, using assorted statistical techniques. Initial detections were of solar cycle signals in basin-wide ocean temperatures at the surface and in the ocean mixed layer from analysis of bathythermographs (White, Lean, Cayan, & Dettinger, 1997), then in the Microwave Sounding Unit global tropospheric temperature record (Douglass & Clader, 2002; Michaels & Knappenberger, 2000). Subsequent analyses of reanalyzed datasets characterized the solar cycle signal in regional as well as global surface temperatures (Camp & Tung, 2007; Gray et al., 2013; Lean & Rind, 2008), and vertically in the atmosphere as well as at the surface (Coughlin & Tung, 2004; Crooks & Gray, 2005; Gleisner & Thejll, 2003; Haigh, 2003).

Observations of climate record the net response to all forcings of Earth’s energy flow, including by the Sun, superimposed on internal variability. Quantifying the solar signal therefore requires isolating this signal in observational time series, for example, using multiple regression or other statistical techniques. Figure 8 shows the individual components of variability in the global temperature of Earth’s surface and in the atmosphere near 5 km (in the middle troposphere), extracted using multiple linear regression. Compared with the modest but persistent solar-induced 11-year surface warming cycle of 0.1oC (in monthly averages), large volcanic eruptions can cool the surface in excess of 0.3oC globally, but there were only two large events in the past 35 years. Fluctuations in the El Niño southern oscillation (ENSO) continuously warm and cool the globe, by as much as 0.2oC on time scales of a few years. For comparison with these natural components of climate variability, the magnitude of global surface warming attributable to anthropogenic forcing in Figure 8a is 0.17oC per decade. Relative to the surface responses shown in Figure 8a, the global middle tropospheric response to the solar cycle is diminished slightly, the volcanic and ENSO signals enhanced, and the anthropogenic trend notably smaller, 0.08oC per decade (Figure 8b).

Figure 8. Changes from 1979 to 2016 in monthly averaged global surface temperature are shown in (a) (on the left) and of monthly averaged global temperature in the atmosphere near 5 km in (b) (on the right), from Figure 2. Also shown are statistical models of the changes constructed by combining components whose individual contributions to the observed changes are identified in the lower two panels; these include the El Niño southern oscillation (ENSO), volcanic aerosols, the solar irradiance cycle, and changes in the concentrations of anthropogenic gases.

During the solar cycle, some regions of Earth warm more and some less than the global average, and some regions cool. Figure 9 shows the regional patterns of the surface and middle troposphere temperature responses to solar cycle irradiance changes, extracted from spatial grids of the observations using multiple regression. Mid latitudes warm more, when zonally averaged, than do low latitudes with regions of cooling interspersed with warming throughout both the Northern and Southern hemispheres. A pattern of alternating mid-latitude warming and minimal change or cooling with a spatial scale of about 1000 km in longitude is particularly evident at 5 km, especially in the Southern Hemisphere, and is also detectable at the surface. At higher latitudes the scale of the warming and cooling increases to continental spatial scales.

Figure 9. Compared on the left are the regional changes from the minimum (between March and May 1996) to the maximum (between January and March 2002) of the 11-year cycle in (a) total ozone, (b) temperature in the atmosphere near 20 km, (c) temperature in the atmosphere near 5 km, and (d) surface temperature. On the right are zonally averaged changes in the regional maps (without cosine adjustment with latitude). Integrated over all regions, these changes reproduce the global solar cycle components shown in Figures 8 and 11.

Additional seasonal and time-lagged dependencies are also present in the Earth’s temperature response to solar cycle irradiance changes. The response patterns in Figure 9 are determined for all months at a lag of 1 month at the surface and 2 months in the middle troposphere. The seasons modulate these patterns, with distinct regional responses especially in the Northern Hemisphere winter (Ineson et al., 2011; van Loon, Meehl, & Shea, 2007). The response pattern also evolves with time. After 2 to 3 years the primary surface temperature response manifests mostly in the North Atlantic (Gray et al., 2013).

Surface temperature records, such as that shown globally in Figure 8a, have reasonable global coverage since 1850 (albeit with sparse observations in polar regions) but the lack of reliable solar irradiance observations before 1978 makes it more difficult to determine the Sun’s influence on climate change prior to the space era. One approach is to alleviate the lack of space-based solar irradiance observations prior to 1978 by reconstructing past solar irradiance using proxy models developed from contemporary observations. An historical reconstruction, such as shown in Figure 3b, is then combined with historical estimates of other natural and anthropogenic influences to statistically decompose a long-term climate record into its multiple variability components (Lean, 2010; Lean & Rind, 2008). Figure 10a shows the components of surface climate change extracted from linear regression of historical global surface temperature data. The approach suggests a long-term solar-induced warming of 0.05oC in the 100 years from 1900 to 2000 compared with 0.6oC warming due to increasing greenhouse gases; the solar contribution to 20th-century global warming is 8% of the anthropogenic contribution.

Figure 10. Changes from 1885 to 2005 in monthly averaged global surface temperature are shown in (a) (on the left) from observations and (b) (on the right) according to simulations made by a physical climate model. Also shown are statistical models of the changes in both the observations and physical model simulations constructed by combining components whose individual contributions to the observed changes are identified in the lower four panels: the El Niño southern oscillation (ENSO), volcanic aerosols, the solar irradiance cycle, and changes in the concentrations of anthropogenic gases.

Another approach for isolating solar (and other) signals in historical surface temperature observations is the use of statistical pattern recognition based on the regional responses suggested by climate models (Stevens & North, 1996). Input to the climate models are past solar cycle irradiance changes inferred from contemporary observations using sunspot numbers. The solar cycle global warming thus determined is 0.06oC (in annual averages), which is larger than expected from model simulations but somewhat smaller than determined using multiple linear regression; the corresponding long-term solar-induced warming in the past 100 years is 0.05oC.

Sun, Ozone, and Stratosphere

Unlike decadal solar cycles in surface temperatures, a dependence of ozone concentration on solar activity was anticipated, and solar signals were readily found in space-based observations of stratospheric temperature and ozone concentrations. Changing solar ultraviolet radiation alters stratospheric ozone and temperature, and these relationships depend on geographical location, season, and altitude. Initial model simulations suggested that a 30% increase in solar ultraviolet radiation at wavelengths 180 to 340 nm would increase ozone at 35 km by 10%, and total ozone by 5% (Penner & Chang, 1978). Although the solar cycle change in ultraviolet radiation is of order a few percent (not 30%), responses of stratospheric ozone and temperature to irradiance changes during the Sun’s 27-day rotation and the 11-year activity cycle were nevertheless detected (Chandra & McPeters, 1994; Hood, 1986; McCormack & Hood, 1996) and are now well established (Soukharev & Hood, 2006; WMO, 2011). Globally averaged total ozone increases a few percent, and globally averaged lower stratospheric temperature a few tenths oC from the minimum to the maximum of the solar cycle (Hood, 1997).

Figure 11 shows the climatology of global temperature in the lower stratosphere (near 20 km) and global total ozone since 1979. The temperature record is a composite of Microwave Sounding Unit observations from multiple spacecraft (Christy, Norris, Spencer, & Hnilo, 2007), and the ozone record is a composite of multiple Solar Backscatter Ultraviolet (SBUV) spectrometer measurements (MOD V8.6, McPeters et al., 2013). Like Earth’s surface temperature, atmospheric temperature and ozone change in response to multiple factors. In addition to solar activity and anthropogenic ozone-depleting substances (such as chloroflurocarbons), volcanic aerosols, the quasi-biennial oscillation (QBO, in the tropical lower stratosphere), the El Niño southern oscillation (ENSO, in the tropical Pacific) and greenhouse gases also influence the stratosphere. All influences must be considered when determining the dependence of temperature and ozone on solar activity from observations. Figure 11 compares the relative strengths of individual influences in the stratospheric temperature and ozone records, extracted using multiple regression.

Figure 11. Changes from 1979 to 2016 in monthly averaged global temperature in the atmosphere near 20 km are shown in (a) (on the left) and of monthly averaged global total ozone in (b) (on the right), from Figure 2. Also shown are statistical models of the changes constructed by combining components whose individual contributions to the observed changes are identified in the lower two panels; these include the quasi-biennial oscillations (QBO), the El Niño southern oscillation (ENSO), volcanic aerosols, the solar irradiance cycle, changes in the concentrations of anthropogenic greenhouse gases (GHG), and the effective equivalent stratospheric chlorine (EESC) of ozone-depleting substances.

Despite their strong interconnections, temperature in the lower stratosphere and total ozone have distinct temporal differences, as a result of their different dependencies on individual natural and anthropogenic influences, evident in Figure 11. Both, however, vary in phase with the Sun’s 11-year irradiance cycle with magnitudes comparable to the net anthropogenic (greenhouse gas plus ozone-depleting substances) change in the 25 years from 1980 to 2015. Compared with the solar cycle change of 0.3oC, the lower stratosphere cooled 0.7oC in the past 35 years from the combined influences of increasing greenhouse gases and increasing, followed by decreasing, ozone-depleting substances (Figure 11a, lower panels). Compared with the solar cycle amplitude of 4 DU (where 1 Dobson Unit, DU, is the amount of ozone in a layer 0.01 mm thick at standard temperature and pressure), total ozone decreased 6 DU from anthropogenic influences over the same time period (Figure 11b, lower panels).

[Figure 11: Ozone and lower stratospheric temperature observations and models]

Total ozone and lower stratospheric temperature change regionally in different ways as solar irradiance changes. Solar-induced warming occurs primarily at low latitudes but the solar cycle signature in total ozone is larger at higher latitudes, especially in the Southern Hemisphere. This is evident in Figure 9, which compares the simultaneous lower stratospheric temperature and total ozone regional responses to the solar cycle irradiance increase that accompany the global solar cycle changes in Figure 11. The regional patterns in Figures 9a and 9b are average responses determined for all months (i.e., averaged over all seasons); the solar signals are generally larger in the winter hemisphere (Weber et al., 2003).

Ozone’s response to changing solar irradiance depends strongly on altitude. The largest relative solar cycle-produced change, an increase of order 4%, occurs in the upper stratosphere, where ozone is photochemically produced, but the primary contributions to the total ozone solar cycle changes (shown globally in Figure 11 and regionally in Figure 9) are dynamical changes in the lower stratosphere, where the solar cycle increase is 1 to 2% (Hood, 1997).

Paleo Evidence

Space-based observations elucidate changes in the Sun and climate during the past four or so decades, coincident with the 20th-century Modern Maximum of relatively strong solar activity cycles (Figure 3). Ground-based observations provide information during recent centuries; sunspot numbers and a few surface temperature observations extend back nearly to the 17th-century Maunder Minimum. On longer climatological time scales, such as during the last 10,000 years of the Holocene (the current interglacial epoch, Ruddiman, 2001), changes in the Sun and climate are inferred from terrestrial archives that record these changes indirectly (Alverson, Bradley, & Pedersen, 2003; Wanner et al., 2008).

Past Changes in the Sun

The primary sources of information about solar variability prior to the sunspot record are the 14C and 10Be cosmogenic isotopes archived in tree-rings and ice-cores, respectively (McCracken & Beer, 2007). Although the atmospheric processes responsible for deposition in their respective terrestrial repositories are quite different, the two cosmogenic isotopes have in common their production by galactic cosmic rays whose transit though the heliosphere to the Earth varies with solar activity (McCracken, Beer, Steinhilber, & Abreu, 2013). Validation of these indicators of solar activity is well established from comparisons with sunspot numbers and geomagnetic activity in the recent period (Beer, 2000). There are uncertainties in the magnitude and timing of the fluctuations attributable to solar activity because climate and atmospheric variability also modulate the deposition rates. Nevertheless, the isolation of common variance in the two independent cosmogenic isotope records provides strong evidence for their ability to mimic past changes in solar activity.

Episodes of enhanced solar activity comparable to the Modern Maximum are evident throughout the past 10,000 years, notably in the eighth and tenth centuries and in the Medieval Period. There are also episodes of anomalously low solar activity, of which the Maunder, Dalton, and Spörer minima are the most recent. Primary periods of solar variability present in the isotope and historical records are at 11 years, 80–90 years (Gleissberg cycle), 200–210 years (De Vries/Suess cycle), and 2,500 years (Halstead cycle).

Historical Sun-Climate Connections

There are many associations of paleoclimate records with 14C and 10Be cosmogenic isotopes during the past few thousand years. The associations manifest as coincident peaks and troughs and common cycles (especially the De Vries cycle) in the solar cosmogenic indices and in indices of climate change inferred from lake levels, sediments, ice-cores, ice-rafted debris, stalagmites, and historical records. Paleo Sun-climate connections are particularly prevalent in tropical Asian and Atlantic rainfall and in Northern Hemisphere mid-to-high latitude temperature.

Paleo evidence suggests that solar activity modulates rainfall in tropical regions by altering the position and strength of the Intertropical Convergence Zone (ITCZ, Kerr, 2005); particularly susceptible are those regions near the latitudinal extremes of its seasonal migration (such as Cariaco and Oman in the Northern Hemisphere and Venezuela in the Southern Hemisphere). In the Americas, river flow subsides in the Amazon (Antico & Torres, 2015) and increases further south (Mauas, Buccino, & Flamenco, 2011) during times of high solar activity. In the Cariaco Bay in the Southern Caribbean, drought associated with low solar activity may have hastened the collapse of the Mayan civilization (Haug et al., 2003; Hodell, Brenner, Curtis, & Guilderson, 2001). Low solar activity corresponds to glacial advances and cool temperatures in the Venezuelan Andes (Polissar et al., 2006). The monsoon in Oman and drought and rainfall in Kenya vary with solar activity (Neff et al., 2001; Verschuren, Laird, & Cumming, 2000).

Paleo evidence also suggests that solar activity modulates temperature and pressure at Northern Hemisphere mid-to-high latitudes, especially in continental regions, by altering the North Atlantic Oscillation (NAO) and its more zonal expression, the Northern Annular Mode (NAM). High solar activity corresponds to warming in the North Atlantic (as inferred from a decrease in drift ice, Bond et al., 2001), subarctic Alaska (Hu et al., 2003) and Beijing (Tan, Hou, & Liu, 2004), and drought in the western United States (Cook, Meko, & Stockton, 1997). Greenland, however, is cooler when solar activity is higher (Kobashi et al., 2015). Northern Hemisphere mid-to-high latitude temperatures were cooler in the Little Ice Age and warmer in the Medieval Warm Period, which coincide respectively with the Maunder Minimum and Medieval Maximum in solar activity (Mann et al., 2009).

Considered collectively, paleo Sun-climate connections show overall consistency with the inhomogeneous regional pattern of surface temperature response to the solar cycle inferred from contemporary observations. Figure 12 superimposes some paleo Sun-climate relationships on the response pattern of the surface temperature to the 11-year solar cycle (from Figure 9).

Even though paleo Sun-climate connections at individual sites may suggest a significant influence of the Sun, Earth’s global climate response to the changing Sun during the past few thousand years, at least, appears to be less than 0.2oC. Solar-induced climate change of this magnitude, coupled with volcanic activity, best reproduces preindustrial surface temperature reconstructions, but neither solar nor volcanic influences can account for climate change in the industrial era (Crowley, 2000).

Figure 12. A variety of Sun-climate connections obtained from paleo evidence over thousands of years at individual sites are superimposed on the regional response of contemporary surface temperature to increasing solar irradiance during the 11-year solar activity cycle (Figure 9d). Identified are examples of the climate changes reported to occur when solar activity increases.

Understanding Sun-Climate Connection Mechanisms

Earth receives more incoming energy when solar irradiance increases. Of an increase of 1 W m−2 from the minimum to the maximum of an 11-year solar cycle, 30% is reflected back into space and the remainder, when averaged over the globe (converting from Earth’s cross-sectional to spherical area), provides climate forcing of 0.18 W m−2. This solar radiative forcing over five years (the time of the solar cycle increase) approximately equals radiative forcing by greenhouse gas increases over the same time period (Hansen & Lacis, 1990). How Earth responds, and temperature changes, depends on feedbacks to these forcings.

As recently as a few decades ago, it was assumed (at least in physical model simulations) that the climate response to radiative forcing, including to cycling solar irradiance, involved deep-seated thermal storage in the oceans. This was thought to delay atmospheric warming from secular climate forcing by a few decades, and attenuate substantially and delay the response to a cyclic forcing (Hoffert et al., 1980). The observational indication of a decadal solar cycle signal of order 0.1oC in global surface temperature, lagging the solar activity cycle peak by only a month or two, alternatively suggests that the actual mechanisms involve (much) faster feedbacks, and engage the atmosphere rather than—or in addition to—the ocean.

Interpretation of observations and targeted model simulations (Benestad, 2006; Gray et al., 2010) show that the mechanisms connecting the changing Sun and climate involve similar convective and dynamical processes that distribute the Sun’s energy from its low latitude input to higher latitudes. Water and winds move incident solar energy meridionally from the equator to the poles, zonally between inhomogeneously heated land and ocean surfaces, and vertically between the surface and atmosphere. These energy flows, which Figure 13 shows schematically, establish the average large-scale structure of the atmosphere. Changes in solar energy modulate this structure, producing what Labitzke and van Loon (1995) term a “ten-to-twelve year oscillation” (also called a “quasi-decadal” oscillation) throughout the troposphere and stratosphere.

Figure 13. The mechanisms thought to be responsible for the response of Earth’s climate and atmosphere to increases in solar radiative energy involve the transport of the incident energy at low latitudes to higher latitudes by meridional flows in tropospheric circulation cells, and from the summer to winter hemisphere in the stratosphere. The upper schematic depicts these meridional flows. The bottom panel indicates zonal inhomogeneities in the meridional flows, which manifest primarily at the interfaces of the tropospheric circulation cells and alter the longitudinal character of, for example, the circumpolar vortex (Polar Jet) and Intertropical Convergence Zone (ITCZ). The red arrows in both the upper and lower schematics indicate the changes that occur when solar activity increases.

Solar-induced changes in stratospheric temperature are integral to the modulation of tropospheric meridional and zonal flows. Because solar ultraviolet radiation varies by an order of magnitude more than does visible and infrared radiation, the stratosphere warms, relative to the surface, when solar irradiance increases. This lessening of the thermal difference between the stratosphere and surface decreases the “static stability” of the atmosphere and perturbs the stratosphere’s large-scale Brewer-Dobson circulation. Interaction of the altered stratospheric circulation with the troposphere below (Salby & Callaghan, 2005) modulates the upward propagation of energy in planetary and gravity waves. These processes depend on season and on the phase of the equatorial stratospheric quasi-biennial osculation (Kodera & Kuroda, 2002; Rind & Balachandran, 1995).


The greater input of solar radiant energy at low latitudes imposes a meridional thermal gradient on Earth’s surface temperature. This thermal gradient drives heat from the equator to the poles via rising and falling motions in the Hadley, Ferrell, and Polar cells. Figure 13 depicts these cells, which extend vertically from the surface to the (local) tropopause. The tropopause defines the altitude where solar heating of the stratosphere reverses the cooling with altitude in the troposphere (Figure 2). It is higher over the equator (~17 km) than at the poles (7 to 10 km, depending on season), and depends, itself, on the incident solar energy, varying annually, seasonally and also with the solar cycle (Gage & Reid, 1981).

An increase in solar radiation alters tropospheric circulation (Gray et al., 2010; Haigh, 1996; Rind & Balachandran, 1995). The Hadley cell, which transports heat convectively from the Earth’s surface to the tropopause, expands. The subtropical jets at the interfaces of the Hadley and Ferrell cells move poleward (Figure 13). It is unclear whether the Hadley cell actually strengthens (Labitzke & van Loon, 1995) or weakens (Haigh, 2003) with increasing solar radiation. The increase in zonally averaged temperatures with solar activity evident in Figure 9 at tropical latitudes (15oS and 15oN) is of order 0.1oC at the surface and less than 0.1oC in the middle troposphere. This indicates that increased solar irradiance warms the surface and troposphere slightly at low latitudes. Since tropospheric air warms as it descends, this suggests that increased solar irradiance produces a net downward tropospheric flow. The net flow possibly reflects the competing effects of surface warming, which accelerates the rising branch of the Hadley cell, and stratospheric warming relative to the surface, which suppresses tropopause height. Clearly evident in the zonally averaged temperatures in Figures 9c and 9d is an increase of 0.2 to 0.3oC in the vicinity of 30o to 40o latitude, north and south, during high solar activity, attributed to increased downward motion in the Ferrell cells (Gleisner & Thejll, 2003; Haigh, 2003).

By increasing the equator-to-pole thermal contrast (a consequence of the greater solar energy deposition at the equator than at the poles), higher solar irradiance stabilizes the circumpolar vortex, reducing the latitudinal range (“waviness”) of its embedded jet stream. In contrast, the reduced thermal gradient during solar minima epochs weakens the polar vortex, the jet stream meanders over a wider range of latitudes, and cold polar air can spill into mid latitudes. So-called blocking patterns, where areas of high or low pressure defined by the jet stream reside for many days over a large-scale geographical region, lengthen and shift (Barriopedro, Garcıa-Herrera, & Huth, 2008). Winter blocking events are more frequent during solar minimum conditions and Eurasian winters are likely colder (Sirocko, Brunck, & Pfahl, 2012; Woollings, Charlton-Perez, Ineson, Marshall, & Masato, 2010;). Figure 9 shows the corollary of this, enhanced surface warming in Eurasia during high solar activity. Circumpolar flow that is strong and zonally confined enhances the pressure gradient between the Icelandic low and Azores high (Figure 13, bottom), which defines the positive phase of the North Atlantic Oscillation. There are many associations of high/low solar activity with positive/negative phases of the North Atlantic Oscillation (Crooks & Gray, 2005; Kodera, 2003) and alteration of the Arctic Oscillation (Shindell, Schmidt, Miller, & Rind, 2001). The Medieval Warm Period (high solar activity) and the Little Ice Age (low solar activity) were characterized by positive and negative phases of the North Atlantic Oscillation, respectively (Trouet et al., 2009).

Changing solar radiation may also alter the location and breadth of the Intertropical Convergence Zone (ITZC, Yan et al., 2015), driving changes in rainfall throughout South America (Antico & Torres, 2015; Mauas et al., 2011). A positive cross equatorial temperature difference (which shifts the Intertropical Convergence Zone northward) is evident in the surface temperature response to increased solar irradiance in Figure 9, where tropical Atlantic Ocean warming in the Northern Hemisphere exceeds that in the Southern Hemisphere. Since the meridional circulation of the Atlantic Ocean carries warm water from the equator to the North Atlantic, changing solar activity may thus modify this large-scale ocean circulation. Simultaneous wind anomalies in the North Atlantic driven by changes in the upper troposphere and lower stratosphere possibly augment these solar-induced tropical changes in modulating the Atlantic meridional circulation (Menary & Scaife, 2014).


Flows in the Hadley, Ferrell, and Polar cells are not uniform in longitude, especially in the vicinity of their interfaces, which are located near ±30o and ±60o latitude (Figure 13, in the summer hemisphere). Mid-latitude synoptic scale (1000 km) thermal gradients disrupt the regular cellular flow and produce semipermanent pressure systems that are local regions of enhanced variability in synoptic weather. Such “centers of action” in the Northern Hemisphere are the Aleutian Low, Hawaiian High, Icelandic Low, and Azores High, which Figure 13 identifies.

Solar-induced bands of regional temperature inhomogeneities are prominent in Figure 9 at mid latitudes, especially in the middle troposphere. These regional patterns are consistent with changing solar radiation affecting the interfaces of the Ferrell and Polar cells, and centers of activity. During high solar activity the Aleutian Low moves westward and the Hawaiian High migrates northward (Christoforou & Hameed, 1997); mid-latitude storm tracks shift poleward, in part because of changing overhead (stratospheric) winds (Haigh, 1996).

Significant zonal inhomogeneities manifest also in the tropical Pacific. Increased solar energy preferentially warms the western tropical Pacific relative to the east because there the thermocline (which separates the upper mixed layer from the deep ocean) is deeper. The resultant thermal east-to-west gradient accelerates the Trade winds and cools the Eastern tropical Pacific, producing a La Niña-like response pattern (Bal, Schimanke, Spangehl, & Cubasch, 2011; Mann, Cane, Zebiak, & Clement, 2005). Consistent with this hypothesis, the tropical Pacific exhibited a La Niña-like state in the Medieval Warm Period, when solar activity was high, and an El Niño-like state during the Little Ice Age, when solar activity was low (Mann et al., 2005). The Walker circulation cell, which flows from west to east in the tropical Pacific troposphere, couples the ocean surface changes with simultaneous solar-driven changes in the overlying stratosphere. Furthermore, solar-induced zonal responses in the tropical Pacific may affect the Hadley cell, coupling to the North Pacific and the Aleutian Low (van Loon et al., 2007).

Modeling the Changes

Models simulate changes in the Sun and Earth by synthesizing and parameterizing the physical mechanisms understood to produce the changes. Comparisons of model simulations with observations validate the purported processes and their representations (Lucarini & Ragone, 2011; Stevens & Schwartz, 2012). Modeling the Sun’s changing radiative output and Earth’s climate response has twofold utility. Firstly, reliable understanding of the causes of global change require it and, secondly, “A knowledge of the terrestrial climate response to a known solar variation could provide a gauge against which climate models could be calibrated. These models, in turn, are the only tools available for society to estimate the seriousness of the growing influences human activity may have on climate” (Schneider, 1977).

Solar Irradiance

Changing solar irradiance is the outcome of variability in the Sun’s magnetic field, of which dark sunspots and bright faculae are primary manifestations. These particular magnetic features correspond to regions of “closed” magnetic flux because they are composed of loops of magnetic field anchored beneath the Sun’s visible surface. Models of solar irradiance variability combine indicators of sunspots and faculae to reproduce observations. Figure 7 compares with observations the total solar irradiance in the past three decades that two such models reproduce (Figures 7d and 7e).

One approach to modeling solar irradiance variations is to statistically regress sunspot and facular indices against observations to quantify their relative contributions (e.g., the Naval Research Laboratory Total Solar Irradiance model, NRLTSI2, Coddington et al., 2016). The sunspot index is constructed from information about the locations and areas of sunspots in spatially resolved observations of the visible disk (such as made by the Greenwich Solar Observatory, Willis et al., 2013). The facular index is the Mg core-to-wing ratio (Snow, Weber, Machol, Viereck, & Richard, 2014). This space-based index, analogous to the ground-based Ca K index, measures the emission in the core of ultraviolet Fraunhofer lines relative to that in the wings.

Another approach models solar irradiance variability by deriving the relative sunspot and facular contributions from solar magnetograms (e.g., the Spectral and Total Irradiance Reconstruction, SATIRE, Krivova et al., 2007). A threshold absolute value is assigned above which magnetic flux is attributed to sunspots and below which to faculae. Physical models of the solar atmosphere estimate the wavelength-dependent contrasts of the magnetic features in the (SATIRE) model.

Figure 14 compares the NRLTSI2 model determination of total solar irradiance variations with observations made by the Total Irradiance Monitor (TIM) on SORCE, and shows the model’s sunspot and facular components. This model accounts for 92% of the observed variance, and the long-term trend in the residuals is within the ±10 ppm per year repeatability of the TIM observations. Relative to NRLTSI2, the alternative SATIRE model has almost a factor of four larger downward trends in the three decades from 1984 to 2013 (Figure 7). Figure 14 also illustrates that the NRLSSI2 model reproduces much of the observed changes in solar spectral irradiance by accounting for the wavelength dependence of the sunspot and facular influences, self-consistently with the total solar irradiance changes. The exception is in the visible spectrum (Figure 14e) where the observations are sparse and lack stability. The SATIRE model estimates somewhat different spectral irradiance changes during the solar cycle than those shown in Figure 14, with larger positive (in-phase) changes in the wavelength region 300 to 400 nm and larger negative (out-of-phase) variations at near infrared wavelengths. Both models utilize a variety of cross-calibrated direct observations such as sunspot numbers to reconstruct irradiance changes since 1610 (Kopp, Krivova, Wu, & Lean, 2016).

Figure 14. Models of contemporary solar irradiance variations since 1996 are compared with observations during solar cycles 23 and 24. In (a) total solar irradiance modeled by NRLTSI2 is compared with the SORCE/TIM observations; in (b) are the bolometric sunspot and facular components that combine to produce the model total solar irradiance changes in (a). Spectral irradiance changes modeled by NRLSSI2 are compared with observations (adjusted to a common scale) of (c) solar Lyman α‎ irradiance at 121.5 nm, (d) solar ultraviolet radiation in a band from 200 to 250 nm, and (e) solar visible-wavelength radiation in a band from 500 to 600 nm. Spectral irradiance changes are modeled with wavelength-dependent combinations of the sunspot and facular time series in (b). When integrated over all wavelengths the NRLSSI2 model reproduces the variations in the NRLTSI2 model in (a).

Figure 3 shows two different estimates of total solar irradiance changes since 1610, estimated using the NRLTSI2 model and two different realizations of historical sunspot numbers. Deducing the long-term evolution of the Sun’s magnetic activity is key to estimating plausible magnitudes of long-term changes in solar irradiance such as shown in Figure 3. One approach is to first quantify the contributions of magnetic features in images of the present-day Sun, then numerically remove selected features to simulate solar quiescence. Such an approach using Ca K images estimates that removal of all magnetic features reduces total solar irradiance 1.5 W m−2 below its contemporary minimum value (Lean, Skumanich, & White, 1992). Another approach is to model the long-term evolution of magnetic flux on the Sun’s surface following its emergence from the convection zone below. Differential rotation transports flux nearer the Sun’s equator more rapidly than at the poles, meridional flow transports flux from the equator to the poles, and diffusion spreads the flux outward (Sheeley, 2005). A simulation of the accumulation of magnetic flux accompanying the increase in flux emergence with increasing solar cycle amplitude since 1710 estimates a total solar irradiance increase of 0.5 W m−2 from the Maunder Minimum to contemporary minima (Lean et al., 2005; Wang, Lean, & Sheeley, 2005). Other approaches suggest a wider range of solar irradiance values during the Maunder Minimum, from comparable to contemporary minima to as much as 6 W m−2 less.

Estimation of solar irradiance changes prior to 1610 relies on the 14C and 10Be cosmogenic isotopes. These indices relate indirectly to solar irradiance in that variations in both reflect changes in the amount of magnetic flux on the Sun’s surface. But whereas the sunspots and faculae that modulate solar irradiance are regions of closed flux, it is the Sun’s “open” flux that modulates the flow of galactic cosmic rays that produce the cosmogenic isotopes (e.g., Lean, Wang, & Sheeley, 2002). Unlike closed flux, where both ends of a magnetic field loop are anchored below the surface, open flux extends outward from the Sun into the heliosphere. Using the observational irradiance record to calibrate cosmogenic isotope fluctuations, Steinhilber, Beer, and Fröhlich (2009) report that the increase in total solar irradiance from the Maunder Minimum to the present is 0.9 W m−2.

Sun-Climate Connections

Models of Earth’s climatology range from very simple to extremely complex. Simple energy balance models provided some of the first estimates of Earth’s response to observed changes in solar irradiance. These models estimate surface temperature by rebalancing the altered incoming solar (short wave) energy and outgoing thermal (long wave) energy; they lack the components that transfer the energy dynamically, underestimate climate response to the solar irradiance cycle by a factor of 2 to 3 and, by directing incoming energy to the deep ocean, project maximum response lagged by a number of years. For example, Hoffert and colleagues (1980) estimate that it may take 50 to 100 years for the climate system to reach 90% of its equilibrium response to an imposed step function forcing.

Responding to the availability in the mid-1990s of historical irradiance reconstructions (Hoyt & Schatten, 1993; Lean, Beer, & Bradley, 1995), general circulation models began to simulate time-dependent climate responses to changing total solar irradiance since the Maunder Minimum. Although the models lacked fully coupled oceans and did not properly resolve the stratosphere, these simulations provided some of the first plausible estimates of the Sun’s role in climate change. For an adopted solar irradiance increase of 3 to 4 W m−2 from the Maunder Minimum to present day (a climate forcing of 0.5 to 0.7 W m−2), the models estimated global temperature increases of 0.4oC to 0.5oC. Maximum response lagged 5 to 10 years, but this response time could be shortened by omitting heat flow to the deep ocean. Distinct regional inhomogeneities were evident as a result of land-ocean contrasts and enhanced warming in subtropical latitudes where cloudiness is low (Cubasch et al., 1997; Rind, Lean, & Healy, 1999).

Over the following decade, the atmospheric component of some general circulation climate models expanded to 80 km to encompass the stratosphere, incorporated ozone parameterizations, and coupled optionally to ocean models. In addition to the direct surface impacts of changing solar radiation at visible and near infrared wavelengths, these models were now able to simulate atmospheric temperature and ozone responses to changing solar ultraviolet radiation, and the consequent indirect impacts on climate of stratospheric-tropospheric coupling (Rind et al., 2004). Including the stratosphere allowed the models to involve the Artic/North Atlantic Oscillation in Earth’s response to changing solar radiative output, simulating a negative phase during the Maunder Minimum.

Simulations of global surface temperature changes (anomalies) in the past century made with the Community Climate System Model, Version 4 (Meehl et al., 2012), are directly compared in Figure 10b with the observed changes in Figure 10a. Statistical decomposition of the simulated changes to quantify individual components, also compared in Figure 10, suggests that this model may underestimate somewhat the global response (at lag one month) to solar cycle irradiance changes. Figure 15 shows regional patterns of response to the solar irradiance cycle determined statistically (using multiple regression at every grid point) in 50-year climate change simulations made with the GISS “high-top” Model 3 (Rind et al., 2008) whose top boundary extends to 80 km. Comparison of these simulated patterns of regional responses with those in Figure 9 derived from observations illustrates that present-day climate models capture the broad, qualitative features of the Earth’s response to the changing Sun but not yet the quantitative detail.

Figure 15. Shown on the left, for comparison with the observationally derived regional changes in Figure 9, are the regional changes from the minimum to maximum of the 11-year cycle in solar irradiance simulated by GISS Model 3 in (a) total ozone, (b) temperature in the atmosphere near 20 km, (c) temperature in the atmosphere near 5 km, and (d) surface temperature. On the right are zonally averaged changes in the regional maps.

State-of-the-art “high-top” general circulation models now couple the earth’s atmosphere from the surface to 80 km with the land surface and oceans at resolutions of a few degrees in latitude and longitude. Thirteen “high-top” general circulation climate models contributed to the Coupled Model Intercomparison Project, Phase 5, for IPCC (2007), for which the models simulated climate change in response to both natural and anthropogenic forcings since 1850. Extensive analysis of the models’ responses to changing solar irradiance provides the most comprehensive assessment yet of present-day model capabilities. Climate models generally simulate weaker and less rapid atmospheric responses to solar cycle irradiance changes (0.07oC in global mean surface temperature, lagged by a few years) than is observed (0.1oC, lagged by a few months). Not all models simulate solar-induced ozone changes, or polar vortex and surface changes. Model deficiencies are attributed to a possible underestimation of the solar ultraviolet irradiance changes adopted for the simulations and, in many of the models, the lack of spectral resolution, interactive ozone chemistry and a self-generating quasi-biennial oscillation (Hood et al., 2015; Misios et al., 2015; Mitchell et al., 2015).

Future Scenarios

How Earth’s climate changes in the future depends, among other things, on how its incoming and outgoing energy evolve with time. The Sun provides incoming energy, greenhouse gases in the atmosphere affect outgoing energy. Solar radiative forcing is about 0.18 W m−2 from minimum to maximum of recent 11-year cycles; longer-term changes are plausible, probably small, but have yet to be detected. Figure 16a shows plausible scenarios for solar and anthropogenic radiative forcings during future decades, based on past behavior. The Sun’s irradiance continues to cycle (but with amplitude 75% of Modern Maximum cycles), anthropogenic forcing continues to increase at a rate of 0.4 W m−2 per decade, and the ozone-depleting gases decline because of the Montreal Protocol (World Meteorological Association, 2011).

Figures 16b, c, d, and e show future changes in Earth’s temperature and ozone corresponding to the assumed forcing scenarios in Figure 16a, according to statistical associations established in the past three decades (shown in Figures 8 and 11). At the surface, solar-produced cycles in global temperature are superimposed on a much larger upward trend. In the middle troposphere and lower stratosphere the projected (respectively, upward and downward) temperature trends are smaller and the solar-produced cycles are relatively more prominent. Projection of future ozone levels in Figure 16b show that estimates of ozone’s recovery from depletion by ozone-depleting substances depends on the ozone dataset used to construct the statistical model. According to the statistical model of the ozone MODV8.6 observations, ozone concentrations do not recover to 1980 levels by 2100. In contrast, the statistical model of the MODV8 observations suggests that ozone concentrations may reach 1980 levels in the next few decades, sooner than chemistry climate model simulations indicate (Lean, 2014). Ozone changes during the solar cycle will complicate detection of this recovery (Reinsel et al., 2005).

Figure 16. How Earth’s surface and atmosphere evolve in future decades depends on future levels of anthropogenic gases and solar irradiance, and on other natural influences such as the El Niño southern oscillation and volcanic aerosols. Shown in (a) are projected scenarios for the solar irradiance cycle, anthropogenic forcing and ozone-depleting substances. Corresponding global changes shown are (b) total ozone, (c) temperature in the atmosphere near 20 km, (d) temperature in the atmosphere near 5 km and (e) surface temperature. The projections are forward extrapolations of the statistical models shown on Figures 8 and 11, for the scenarios in (a) and thus assume that the causes of climatological change in the past three decades manifest similarly in the near future. The solid and dashed projections in each panel correspond to statistical models developed from different datasets.

Were the Sun’s activity to decline to that of the 17th-century Maunder Minimum, how might decreased solar radiative output alter future projections of Earth’s future climate?.

Estimates of solar irradiance during the Maunder Minimum range from no change below contemporary solar minima to a (highly unlikely) reduction of as much as 6 W m−2 (equivalent to radiative climate forcing of 1 W m−2, accounting for 30% reflected radiation and converting from Earth’s cross-sectional to global area). The latter is considered implausible from solar and stellar perspectives (Judge et al., 2012) and incompatible with historical surface temperatures (Feulner, 2011). Global cooling in response to the more plausible total solar irradiance reduction of 0.6 W m−2 shown in Figure 3, with attendant climate forcing of 0.1 W m−2, is expected to be less than 0.05oC. For comparison, an increase in anthropogenic forcing of 2.6 W m−2 over the next century (a moderate scenario of anthropogenic forcing, Moss et al., 2010) is expected to warm the globe in the range 1.5 oC to 1.9oC. Changing solar irradiance will not counter future greenhouse warming (Feulner & Rahmstorf, 2010).

The projections in Figure 16 of Earth’s future temperature and ozone derive from current understanding and statistical modeling of changes in the Sun and Earth during past decades. But as Earth changes under the influence of increasing anthropogenic gases, the processes through which it responds to changing solar irradiance may differ in unexpected ways. Feedbacks prevalent in the contemporary epoch may disappear and new feedbacks may emerge. For example, as Arctic sea ice melts and the high latitude open ocean absorbs more solar energy, future solar forcing will impose upon an equator-to-pole thermal gradient reduced relative to that of the past three decades (Francis & Vavrus, 2012). As anthropogenic warming accelerates the Brewer-Dobson stratospheric circulation, future levels of ozone, which absorbs solar ultraviolet radiation, will also change and the QBO, which regulates the atmosphere’s response to solar activity, may disappear (Rind, Jonas, Balachandran, Schmidt, & Lean, 2014).


As the Sun’s radiative energy changes, Earth’s surface climate and atmosphere respond accordingly. This necessitates reliable knowledge of the magnitude of these changes for the foreseeable future; the consequence of not knowing is to render the Sun a global change “wild card.”

Making Measurements

Making measurements that are accurate, stable, and extend indefinitely is a pressing challenge for knowing how the Sun and Earth change climatologically. Spurious trends in observational databases confound detection and attribution of change. They foster speculation of spurious processes and preclude validation of models, compromising their utility as tools for projecting past and future changes.

Is the inter-minima decrease in total solar total irradiance from 1996 to 2008 real or the consequence of undetected instrumental instabilities and model imperfections? Which of the trends in two separately analyzed Microwave Sounding Unit global temperature datasets and ground-based radiosonde observations pertains to the real troposphere (Thorne et al., 2011)? Relative to the trends in middle troposphere and lower stratosphere global temperatures in the MSU times series of Christy et al. (2007) shown in Figure 2 (and Figures 8 and 11), the MSU dataset of Mears et al. (2011) indicates somewhat greater tropospheric warming and less stratospheric cooling in the past three decades; the projections in Figure 16 differ accordingly. Does the MOD V8.6 or MOD V8 total ozone database better replicate ozone climatology? Relative to the former, the latter implies a more rapid ozone recovery from ozone-depleting substances (Lean, 2014).

Advanced space metrology is essential to trace instrument calibrations to radiometric standards for in-flight calibration tracking and cross-calibration of overlapping independent observations. Because the uncertainties of space-based instrument accuracies exceed those sought for detecting climatological change, measurement overlap remains essential, at least in the near future, to achieve the cross-calibration needed to secure continuity. Perhaps the largest impediment to achieving high fidelity space- and ground-based datasets is programmatic; comprehensive, calibrated databases of the changing Sun and Earth require considerable resources (NASEM, 2015).

Advancing Models

Models of the Sun and the Earth are the only tools available for projecting past and future changes. One challenge is to reconcile differences among simulations made by different models (Rowlands et al., 2012); another challenge is to reconcile differences between models and observations (Hood et al., 2015; Mitchell et al., 2015).

A challenge for models of solar irradiance variability, such as the two shown in Figure 7, is to reconcile differences in their multidecadal representations of sunspot and facular components, including spectral dependencies. For reconstructing past irradiance changes on centennial time scales, the utility and relevance of cosmogenic isotopes that underpin paleo Sun-climate connections requires clarification; while there is little doubt that cosmogenic isotopes record changes in solar magnetic activity, the changes pertain to the Sun’s open flux (which extends into the heliosphere) whereas closed flux (which is anchored below the Sun’s visible surface) produces the sunspots and faculae that alter solar irradiance. Simulations with models that parameterize the transport of magnetic flux on the Sun’s surface (by differential rotation, meridional flow, and diffusion) indicate that significant temporal changes in open flux do not necessarily imply equivalent changes in closed flux (Lean et al., 2002). This physical understanding cautions against representing historical solar irradiance variations in terms of open flux indices alone.

An immediate challenge for climate models is better simulation of natural decadal change, globally, regionally, vertically, and seasonally. Arguably such limitations are the reason why climate models predicted more warming in the past 15 years than was observed, leading to the erroneous notion that global warming had “paused” (Lewandowsky et al., 2015). A combination of solar irradiance, El Niño southern oscillation fluctuations, and increasing anthropogenic gases readily reproduces observed surface temperature changes in the 15 years from 1995 to 2009 (Figure 8; Foster & Rahmstorf, 2011; Lean & Rind, 2009). Meridional and zonal thermal gradients are fundamental in organizing the dynamical flows in the troposphere that changing solar irradiance (and volcanic aerosols, the El Niño southern oscillation, and anthropogenic gas increases) alters. But physical climate models represent the equator-to-pole thermal gradient poorly (Rind, 2008) and do not produce realistic fluctuations in the El Niño southern oscillation. As well, only a subset of current climate models includes a fully resolved stratosphere, which is necessary to simulate mechanistic pathways through which Earth’s climate responds to stratospheric modulation by changing solar ultraviolet radiation.

Detecting Change

One approach for detecting how and why Earth evolves climatologically is the statistical analysis and interpretation of observational databases (such as shown in Figures 8 and 11). Another approach employed in climate change assessments (IPCC, 2007) is optimal signal detection in which observations are regressed against physical model simulations of the expected changes (globally and regionally) to known forcings in order to determine the most likely cause(s) (Stott et al., 2010). An overarching challenge is to achieve agreement between different approaches, cognizant of their limitations.

A challenge for purely statistical detection and attribution is the optimal selection of predictors against which to regress the observed changes, accommodating or reducing covariance among them, establishing the resilience of the statistically derived coefficients and estimation of realistic uncertainties (Roscoe & Haigh, 2007; von Storch & Zwiers, 1999). For example, depending on the particular epoch used in statistical regression models, the coincidence of the El Chichón and Pinatubo volcanoes with high solar activity can add degeneracy to these predictors. Similarly, long-term trends in solar irradiance and greenhouses gases project onto each other, more or less so depending on the epoch.

A challenge for statistical attribution and detection using patterns generated by physical climate models is to validate the realism of the adopted patterns. There are differences between observed and modeled regional responses of the climate and atmosphere to changing irradiance in the solar cycle (compare Figures 9 and 15). This suggests that current model-produced regional responses may not be optimal, which in turn may compromise the capabilities of localized detection.

Projecting the Past and Future

Climatological time scales are considerably longer than those of contemporary reliable observations of the Sun and Earth. This constrains the reliability of projections of past and future changes. In the USA, NOAA’s Space Weather Prediction Center (SWPC, operationally forecasts solar activity indices (sunspot numbers and 10.7 cm flux) days to weeks ahead for the purposes of forecasting the space environment, but there are no forecasts of solar irradiance, for which partitioning of emerging magnetic flux into sunspots and faculae must additionally be forecast. Despite many reports of an impending Maunder Minimum-type quiescence, the likelihood of this is simply not known.

Nor are climate forecasts yet operational, but a growing imperative motivates ad hoc efforts to quantify current prediction capability (Smith et al., 2013). Since changing solar irradiance alters the stratosphere, which in turn can influence large-scale surface weather, specification and forecasting of solar-induced changes is expected to improve medium- and long-range weather prediction, and is being implemented in the Navy Global Environmental Model (McCormack & Eckermann, 2015) and UK Met Office models (Ineson & Scaife, 2008). Crossing the “valley of death” from research to operations continues to be a looming challenge.

Communicating Understanding

It is a challenge to communicate the simultaneous limitations and potential of the changing Sun to change Earth’s climate and atmosphere, and the complexity of the connections. Some consider the Sun to be a major cause of climate change; others insist its influence is negligible. This bifurcation underscores public debate about the extent of natural versus anthropogenic influences on Earth’s climate. That the Sun’s radiative output changes is now indisputable; that these changes initiate myriad complex responses throughout the Earth system, currently and in preindustrial times, is also increasingly well established. But the magnitude of solar-produced climate change is significantly smaller than that imposed by increasing greenhouse gases in the past century, and expected during the next century.

The most optimistic avenue for communicating connections between the changing Sun and Earth’s climate and atmosphere is to observe, model, and project the ongoing changes with sufficient fidelity to alleviate uncertainty as to their extent and impact.


NASA and the Chief of Naval Research supported this work. Two decades of discussions with David Rind are gratefully acknowledged, as are ongoing collaborations with scientists at NRL and LASP. Very much appreciated are the sustained efforts of the many scientists who produce the numerous databases that enable the analyses of variability in the Sun and Earth. As Doug Hoyt and Ken Schatten acknowledged in 1997, that we understand so much more today than a century ago is a tribute to the vibrancy and persistence of a relatively small group of scientists for whom measuring, understanding and connecting changes in the Sun with those on Earth has motivated creative, persistent dedicated research.

Appendix 1: List of Acronyms


Active Cavity Radiometer Irradiance Monitor


Active Cavity Radiometer Irradiance Monitor Satellite


tlantic Meridional Oscillation


tlantic Oscillation


Coupled Model Intercomparison Project


Community Climate System Model


El Niño Southern Oscillation


Greenhouse Gases


Goddard Institute for Space Studies


Intergovernmental Panel on Climate Change




Intertropical Convergence Zone


Merged Ozone Dataset


Microwave Sounding Unit


North Atlantic Oscillation


Northern Annular Mode


National Aeronautical and Space Administration


National Academies of Science Engineering and Medicine


National Oceanic Atmospheric Administration


National Research Council


Naval Research Laboratory Total Solar Irradiance 2


Naval Research Laboratory Solar Spectral Irradiance 2


Physikalisch-Meteorologisches Observatorium Davos


Quasi-Biennial Oscillation


Royal Greenwich Observatory


Spectral and Total Irradiance Reconstruction


Solar Backscatter Ultraviolet


Solar Extreme Utraviolet Experiment


Sunspot Index and Long-Term Solar Observations


Solar Mesosphere Explorer


Solar Maximum Mission


Solar Heliospheric Observatory


Solar Stellar Intercomparison Experiment


Solar Radiation and Climate Experiment


Space Weather Prediction Center


Solar Ultraviolet Spectral Irradiance Monitor


Total Irradiance Monitor


Total Ozone Mapping Spectrometer


Total Spectral Irradiance


Upper Atmosphere Research Satellite


United States Global Change Research Program






Variability of Solar Irradiance and Gravity Oscillations


World Meteorological Organization