Solar Variations, Ozone, and the Middle
Weather and climate are experienced in the troposphere, which extends upward to the tropopause at about 15 km. The Earth's middle atmosphere (the stratosphere and the mesosphere) extends from the tropopause to approximately 90 km (Figure 1.2). Understanding the middle atmosphere is crucial for global change studies for two primary reasons. Firstly, this is where approximately 90 percent of the Earth's ozone shield resides and, secondly, the region constitutes the upper boundary of the troposphere. Changes in the ozone layer modulate the amount of ultraviolet (UV) radiation reaching the biosphere and are thus of direct concern for life on Earth. A decrease in column ozone of 1 percent causes the UV dose in the spectral region damaging to deoxyribonucleic acid (DNA) to increase by about 2 percent. Changes in ozone, temperature, and other trace gases have been widely implicated in a variety of the mechanisms by which changes in the middle atmosphere might influence the biosphere by physically coupling to the troposphere and modifying the climate (NAS, 1982; Lacis et al., 1990; Schwarzkopf and Ramaswamy, 1993; Hauglustaine et al., 1994; Rind and Balachandran, 1994). Crucial to these mechanisms is ozone, which is highly responsive to, and also has a controlling influence on, the state of the middle atmosphere. Ozone is
directly influenced by changes in solar radiative and energetic particle inputs to this region.
All of the Sun's ultraviolet energy input at wavelengths between about 150 and 300 nm is deposited in the Earth's middle atmosphere. This energy plays an essential role in the chemistry, radiation, and dynamics of the region. Figure 1.2 shows the altitude at which solar radiation (from an overhead Sun) reaches 1/e of its original intensity and gives an approximate measure of the penetration depth for different wavelengths. The inversion in the atmospheric temperature profile at about 15 km (Figure 1.2) that defines the tropopause is a direct consequence of heating by solar UV energy abosorbed by ozone in the middle atmosphere. The only significant solar radiation of shorter wavelength reaching the middle atmosphere is a small and highly sporadic contribution from X-rays, and the strong H I Lyman line at 121.6 nm.
The ozone layer exists because of the interaction of solar UV radiation with the constituents of the middle atmosphere. Photodissociation of molecular oxygen by solar UV radiation at wavelengths from 170 to 242 nm (in the O2 Schumann-Runge bands and Herzberg continuum) is the chief source of atomic oxygen and hence ozone production. Formed by combination of atomic and molecular oxygen, ozone is in turn photodissociated, mainly by solar UV radiation at wavelengths between 240 and 300 nm (in the strong O3 Hartley bands and continuum), but also by longer wavelength visible solar radiation. Ozone's strong absorption in the UV region of the spectrum serves the dual role of heating the middle atmosphere and protecting the surface of the Earth from damaging doses of ultraviolet radiation. Solar UV radiation also creates other important trace constituents, such as chlorine (Cl) and hydroxyl (OH), that participate in catalytic reactions that destroy ozone. Current understanding of the processes affecting ozone and other trace constituents of the middle atmosphere has been reviewed in a number of reports (e.g., WMO, 1988), and is not repeated here.
Stratospheric ozone has received intensive study in recent years, and is the subject of research under the Biogeochemical Dynamics science element of the USGCRP. Ozone is known to be influenced by human related sources such as chlorofluorocarbons (CFCs), carbon dioxide (CO2), and methane (CH4) and by natural occurrences such as volcanoes and solar variability. As summarized in Figure 3.1, definite changes in the total
amount of stratospheric ozone are thought to have occurred throughout the past few decades. Long term changes in stratospheric temperatures have also been found (WMO, 1988; Randel and Cobb, 1994). A significant fraction of the long term changes in ozone and temperature is thought to be caused by human related emissions (of CFCs and other gases). Recent analysis of data from the Total Ozone Mapping Spectrometer (TOMS) on board the Nimbus 7 satellite indicates a global (65 N to 65 S l latitude) ozone decrease of 0.27 percent ± 0.14 percent per year since 1978 (Stolarski et al., 1991; Hood and McCormack, 1992). This is thought to arise from anthropogenic effects.
A significant fraction of the ozone variance during the past decade is also due to solar forcing (Figure 3.1). The ozone content of the middle atmosphere varies with the 11-year solar cycle because the solar radiation and particle environment responsible for creating and destroying ozone varies. Superimposed on the long term downward ozone trend deduced from TOMS data during the past 11 years is a solar cycle variation whose amplitude is estimated to be 1.8 percent ± 0.3 percent (Hood and McCormack, 1992). These solar-related changes in ozone exacerbated the downward anthropogenic trend from 1982 to 1986 (the descending phase of the solar activity cycle) and masked it almost completely during 1988–1991 (the ascending phase of the cycle).
Longer-term variations related to solar variability are clearly possible. To develop a reliable understanding of true anthropogenic effects on ozone therefore requires that its natural variability in response to solar influences be fully defined; knowledge of solar-induced changes will be particularly important in the future in verifying whether an observed slowing of the ozone depletion rate is the result of limits on anthropogenic sources or a response to changing solar energy inputs.
As mentioned in Chapter 2, solar-induced variations in stratospheric temperature structure may be affecting the troposphere through modulation of the Hadley circulation in the tropics or planetary wave genration in the extratropics. Change in the latitudinal temperature gradient in the stratosphere and planetary wave generation in the extratropics affect stratospheric wind speeds and thus the ability of long-wave energy to propagate out of the troposphere, further altering tropospheric dynamical patterns and subsequently radiative parameters (Rind and Balachandran, 1994).
Solar Ultraviolet Radiation
Whereas total solar irradiance (discussed in Chapter 2) was, until recently, assumed to be invariant, solar emissions at shorter UV wavelengths are known to vary in phase with the solar activity cycle, with the shortest wavelengths varying the most (see Figure 1.1). The reason is that, compared with the visible radiation that contributes most to the total solar irradiance, the ultraviolet radiation is formed higher in the Sun's atmosphere and is more susceptible to the impact of magnetic fields that erupt into the solar atmosphere during times of enhanced solar activity. Accurate knowledge of the amplitude and temporal structure of solar UV spectral irradiance variations is clearly critical for studies of solar forcing of the middle atmosphere. The true extent of the UV irradiance changes over the 11-year solar activity cycle has long been debated and remains somewhat uncertain.
Estimates of the amplitude of the solar spectral irradiance variations shown in Figure 1.1 were derived from a combination of direct measurements and inferences from empirical variability models. Ultraviolet radiation is currently thought to vary over the 11-year solar cycle by 5 percent to 10 percent at 200 nm, 50 to 100 percent at H I Lyman , and more than a factor of two at the shortest extreme ultraviolet (EUV) wavelengths, reaching maximum values during times of activity maximum, as indicated by the sunspot number. The solar cycle variations at wavelengths from 125 to 300 nm shown in Figure 1.1 are significantly less at all wavelengths than inferred directly from the rocket and satellite data base (Lean, 1987; Rottman, 1988), but still cannot be considered entirely reliable. Uncertainties remain even for the H I Lyman line, whose variability is considered to be better known than for any other portion of the solar spectrum. Superimposed on the 11-year UV irradiance cycle is a 27-day rotational modulation whose amplitude near high activity levels may as much as half the 11-year cycle amplitude.
Measurements of Solar UV Spectral Irradiance
Various instruments have been launched on rockets, satellites, and the Space Shuttle during the past 20 years (about two solar activity cyles) with thegoal of acquiring knowledge of the Sun's ultraviolet spectral
irradiance at wavelengths less than 400 nm. Solar Backscatter Ultraviolet (SBUV)-type instruments have been flown since November 1978, initially on the Nimbus 7 satellite (Health and Schlesinger, 1986) and subsequently on a series of National Oceanic and Atmospheric Administration (NOAA) satellites (Donnelly, 1988; Cebula et al., 1992). Coincident with these observations from 1982 to 1989 are measurements by the solar spectrometer on board the Solar Mesosphere Explorer (SME) (R (Rottman, 1988; London and Rottman, 1990). Two solar spectral irradiance monitors for the region from 115 to 415 nm are currently operating on board the Upper Atmosphere Research Sateellite (UARS), launched in September 1991 near the maximum of activity in solar cycle 22 (Rottman et al., 1993; Brueckner et al., 1993). Other solar UV spectroradiometers have been flown intermittently on rockets and balloons, with half a dozen recent measurements from the Space Shuttle.
Measurements from rockets (some of the more than 50) and spacecraft of the solar Lyman irradiance are shown in Figure 3.2. The primary satellite data bases are the Orbiting Solar O Observatory-5 (OSO-5), from 1970 to 1974 (Vidal-Madjar 1975; Vidal-Madjar and Phissamay, 1980), Atmospheric Explorer-E (AE-E) from 1977 to 1980 (Hinteregger et al., 1981), SME from 1982 to 1989 (Rottman, 1988; Barth et al, 1990; White et al., 1990) and UARS, from late 1991 to the present (London et al., 1993); none overlap.
Figure 3.2 typifies, in many respects, the current state of the entire data base of solar irradiance at wavelengths shorter than of 400 nm. No portion of the spectrum has been monitored continuously for even one 11-year solar activity cycle. Instrumental effects contribute significantly to the scatter of the data. Absolute accuracies are typically quoted to be 10 to 50 percent, but inconsistencies between measurements made by different experimental groups are clearly apparent and signify larger systematic uncertainties. Information about instrument responsivity drifts during spacecraft missions is absent (in the case of AE-E), limited (in the case of SME), or speculative (in the case of SBUV). The SBUV diffuser reflectivity is thought to have degraded over seven years by about 30 percent at 250 nm (Herman et al., 1990). Although the SME instrument included a stored diffuser for periodic reference, this instrument suffered from problems associated with t temperature-dependent wavelength calibration drifts. Pronounced long term drifts are also suspected in the AE-E data.
Only the two UARS solar UV radiometers have the capability for end-to-end, in-flight sensitivity monitoring using, in the case of the Solar Ultraviolet Spectral Irradiance Monitor (SUSIM), a bank of four deuterium lamps and redundant optics, and for the Solar Stellar Irradiance Comparison
Experiment (SOLSTICE), the UV fluxes from a collection of bright stars (Woods et al., 1993).
Despite the inadequacies in the experimental data base, there is no doubt that the Sun's full disk UV spectral irradiance varies. Changes of a few percent or more occur in response to the modulation of solar activity by the 27-day solar rotation and have been reliably measured at wavelengths as long as 250 nm. The amplitude of the rotational modulation is largest at the shortest wavelengths and can be as much as 30 to 40 percent at H I Lyman (e.g., Lean, 1991). It has proven more difficult to extract from the data base reliable information about the magnitude of solar irradiance variations over the 11-year activity cycle. These variation are expected to be somewhat larger than those associated with solar rotation (Lean et al., 1992b), but are uncertain because of the degradation of instrument sensitivies during long-duration space flight, and because of the difficulty in comparing measurements by different instruments whose inacurracies typically exceed the expected amplitude of the spectral irradiance variability.
Irradiance Variability Parameterizations
In lieu of an adequate experimental data base, information about the Sun's UV spectral irradiance variations has been acquired by critically analyzing the available data in conjunction with knowledge of solar activity derived independently, often from ground based observations (Lean et al., 1992b; DeLand and Cebula, 1993). In concert with rising solar activity during the 11-year cycle is an overall increase in the solar ultraviolet spectral irradiance because of increased emission from magnetic active regions whose radiation is enhanced at these wavelengths. Superimposed on this 11-year cycle are intermediate term variations over time scales of three to nine months that reflect the birth, growth, and decay of individual active regions associated with surges of activity. In addition, the UV irradiance is modulated by the Sun's 27-day rotation, which brings active regions onto the face of the Sun viewed from the Earth (e.g., Lean, 1987).
Irradiance variability parameterizations range from linear transformations of solar activity surrogates to detailed representations of active region source of enhanced UV emission. Certain solar emissions that can be measured from the ground, such as the emission core of the Ca II K
Fraunhofer line, behave similarly to the UV irradiances in their responses to solar magnetic activity. These ground based data reflect the UV enhancement in active regions, exhibiting variations throughout the 11-year solar cycle in response to active region evolution and during the 27-day solar cycle. Estimates of solar cycle UV irradiance variability have been obtained either by extrapolating known 27-day changes, assuming a linear relationship with a suitable solar activity indicator (Health and Schlesinger, 1986; Cebula et al., 1992; Lean et al., 1992b) or, in the case of Lyman , by linearly regressing extant solar UV irradiance data against full-disk emission surrogates (e. g., White et al., 1990). Calculations of the UV spectral irradiance directly from ground based, spatially resolved Ca II K observations of the areas, locations, and spectral brightness of active regions have also been attempted with the purpose of elucidating the origins of the UV irradiance variations in terms of solar magnetic activity (Cook et al., 1980; Lean et al., 1982).
Inferences about the amplitude of the 11-year solar cycle variation in the UV spectral irradiance from 120 to 300 nm derived from measurements of the rotational modulation (Cebula et al., 1992; Lean et al., 1992b) are in general agreement with the variations reported from the SME observations (Rottman, 1988; London and Rottman, 1990). From 210 to 250 nm the solar cycle variation appears to be of the order of 3 to 5 percent, increasing to 7 to 9 percent at 200 nm. In the region of 120 to 200 nm the UV spectral emission lines vary significantly more than does the underlying continuum. For example, 1 nm spectral regions dominated by the H I Lyman and O I 133.5 nm lines are estimated to vary by about 50 percent, with the underlying continuum varying by about half this amount.
However, parameterizations of solar UV irradiance variability derived from auxiliary solar data remain to be verified by more reliable and extended long term observations than SME, made by instruments such as those on UARS having complete in-flight sensitivity monitoring. This is true even for the Sun's strongest emission feature, the H I Lyman line. Although its irradiance has been measured over more than three solar cycles, and far more frequently than observations of any other portion of the solar UV spectrum, unequivocal confirmation of Lyman 's true 11-year cycle variability has still proven difficult, confounded by apparent inconsistencies between different data sets and the suspicion of instrument
artifacts in the data base. As demonstrated in Figure 3.2, none of the four Lyman data bases overlap, which severely impedes the separation of the data into solar and instrumental effects. Data from the solar spectrometer on the SME satellite have been widely used for aeronomic applications, either directly for the period from 1982 to 1989, or over longer periods via parameterizations with the Ca II K (White et al., 1990) and He I 1083 nm EW (Lean, 1990) ground based indicators of chromospheric variability. However, measurements made by the AE-E (Fukui, 1990) and, more recently, the SOLSTICE on UARS imply significantly higher Lyman irradiances (Figure 3.2) and greater variability than is estimated from parameterizations based on rotational modulation or SME observations. Pioneer Venus Orbiter Langmuir probe observations of the integrated solar EUV and Lyman fluxes also provide indirect evidence for a Lyman 11-year irradiance cycle in excess of 50 percent and poorly correlated with the Ca II and He I indices near the maximum of solar cycle 22 (Hoegy et al., 1993). Persistent inconsistencies among the measurements themselves and with parameterizations derived from solar activity surrogates caution against supposing that all is known even about the Sun's photon output variation at H I Lyman or at any other UV wavelength.
Solar Proton Events
Eruptions of activity on the Sun, such as solar flares and coronal mass ejections, frequently give rise to fluxes of energetic particles. Interaction of these particles with the Earth's environment depends on the site of the eruption on the solar disk and on the particles' transport to the Earth via the solar wind and the Earth's near-space environment (discussed in Chapter 5). Typically, at 10 million electron volts (MeV), a large solar energetic-particle event reaches its peak at the Earth a few hours to a day after the initiating eruption and may persist for one to a few days. The particles are mainly protons and alpha particles with a small component of medium-mass and heavy ions. Fluxes at energies > 10 MeV can reach 100,000 cm-2sec -1sr-1.
Solar energetic particles are excluded from low magnetic latitudes by the Earth's magnetic field, but the polar regions are exposed to the full interplanetary flux. Figure 3.3 shows the regions in the two hemispheres (the polar caps) within which the effects are particularly felt, although during intense geomagnetic storms these boundaries can expand to considerably lower latitudes for brief periods. The depth in the atmosphere to which the particles penetrate depends on their initial energy; the energy spectrum of the particles is highly variable from event to event.
At times, solar protons can bombard the middle atmosphere at high latitudes for periods ranging from hours to days at a time. Penetration into the stratosphere (i.e., altitudes below about 50 km) requires energies in excess of 30 MeV for a proton, while direct penetration into the high-latitude troposphere requires energies greater than about 1 billion electron volts (GeV). Events with significant fluxes above this latter threshold are rare, but ground level neutron monitors, which detect thermal neutrons generated as secondary particles by the interaction of very energetic solar protons (several hundred MeV) with atmospheric constituents, have recorded many events.
The energy loss mechanism for solar protons in the Earth's atmosphere is almost exclusively through ionization. The impact between a solar particle and an atmospheric molecule releases a secondary electron with energy in the range of a few hundred electron volts that goes on to produce more ionization. Primary and secondary electrons both dissociate atmospheric nitrogen efficiently, while subsequent positive-ion reactions dissociate water vapor. The net result is the production of a wide range of odd-nitrogen (NOx) and odd-hydrogen (HOx) compounds, primarily in the middle atmosphere between heights of roughly 20 to 100 km. At heights of 60 to 70 km, where the solar energetic particle ionization reaches its maximum, it can exceed the normal background ionization rate by a factor of a million. These relatively large ionization rates give rise to significant increases in the concentration of free electrons and consequent severe disturbances in radio propagation at polar latitudes. In fact, the radio propagation effects provided much of the early information on solar energetic particle events before direct spacecraft measurements became available.
Perhaps one of the most surprising aspects of solar terrestrial physics to emerge in the past decade has been the number and intensity of major
solar particle events that can occur. Whereas the peak of solar cycle 20(1969) showed substantial flaring and many moderate events, only a few ofthe particle disturbances (e.g., during August 1972) were considered reallyimportant from the point of view of solar-atmospheric coupling. In solarcycle 21, which peaked in 1979–1980, there were no flares as intense or asspectrally hard as the August 1972 event.
In solar cycle 22, however, the Sun provided many surprises. A number of exceptionally large eruptive events, comparable to the August 1972 event, occurred on the Sun during 1989. Very large events were seen in March and August, while in late September and early October an unusually intense and spectrally hard event occurred for which the flux of protons with energies > 10 MeV matched the highest values ever recorded. As well as wreaking havoc on sophisticated electronic systems
(Allen et al., 1989; Chapter 5), the solar protons associated with these major storms are thought to have led to signficant ozone depletions (Stephenson and Scourfield, 1991; Reid et al., 1991). Since September 1991, the Particle Environment Monitor (PEM) on board the UARS has been monitoring charged particle energy inputs to the middle and upper atmosphere (Sharber et al., 1993), including perhaps the most intense storm on record, thereof November 1991.
Figure 3.4 provides a perspective for this recent activity by comparing the flux versus energy for the September 1989 event (cumulative proton dose at geostationary orbit for 29 September to 1 October above several energy thresholds) with the August 1972 total flux above 30 MeV. This 1972 point was essentially the entire significant flux for all of solar cycle 20 and was matched by the September 1989 point alone.
When all events are considered, solar cycle 22 emerged as one of the most active and most intensely disturbed periods known, with concomitant impact on the middle atmosphere.
Energetic electrons, like energetic protons, are capable of penetrating to the middle atmosphere. Enhanced relativistic electron fluxes (2 to 15 MeV) in the outer trapping zone of the magnetosphere have been observed at geostationary orbit since 1979. The presence or absence of multi-MeV electrons in the Earth's outer magnetosphere is controlled largely by high speed solar wind streams, which in turn show a strong solar cycle dependence related to the buildup of solar coronal hole structures in the declining phase of the 11-year sunspot cycle. The population dwindles as solar minimum is reached and is largely absent near the solar maximum. A 27-day periodic enhancement of the relativistic electrons is observed in association with concurrently measured solar wind streams in excess of 600 km/s.
The depth of penetration into the middle atmosphere of energetic electrons precipitated from the magnetosphere is a strong function of the electron energy (e.g., Baker, et al., 1987). However, the precipitating electron energy spectrum and its flux at the top of the atmosphere for relativistic electron precipitation (REP) events are still uncertain. A 1 MeV electron can penetrate to about 55 km, while a 10 KeV electron only
penetrates to about 100 km. Thus, if there is a significant population andflux of trapped high-energy relativistic electrons (energy > 700 KeV) in themagnetosphere, precipitation in auroral and subauroral latitudes may occur, withenergy deposition and resulting production of hydrogen oxides
and nitrogen oxides in the lower mesosphere and upper stratosphere. If the energetic events are sufficiently frequent, the downward transport of nitrogen oxide molecules during the winter and spring could result in a significant effect on the budget of odd-nitrogen in the lower stratosphere (e.g., Dahe et al., 1991).
Measurements from the Spectrometer for Energetic Electrons (SEE) on spacecraft 1979-053 and 1982-019 do provide evidence for large fluxes of relativistic electrons at 6.6 Earth radii (RE) that undergo variations on 27-day and 11-year time scales (Baker, et al., 1987). Precipitation of 30 to 40 percent of these electrons from the largely trapped trajectories seen at synchronous orbit into the atmosphere could have significant effects on middle atmosphere chemistry (Baker et al., 1993). However, measurements of the energy deposited into the middle atmosphere made by sounding rockets (e.g., Goldberg et al., 1984) have typically indicated much smaller fluxes of relativistic electrons than measured by the SEE instrument. Figure 3.5 shows a comparison of a typical energy deposition rate from an REP event measured by Goldberg et al. (1984) with energy deposition rates of EUV and galactic cosmic rays. In the lower mesosphere and stratosphere the energy deposition rates are almost two orders of magnitude larger for the Baker et al. REP event than for the Goldberg et al. event. On the other hand, recent rocket observations (Herrero et al., 1991) indicate that the fraction of the high altitude flux of relativistic electrons that reaches the middle atmosphere near noon (when daytime REP fluxes peak) may actually be significantly larger, and penetrate deeper, than the nighttime REP events measured by earlier rockets (Goldberg et al., 1984). At about 100 km, electron spectra were estimated to be 5 to 25 percent of the geostationary orbit fluxes at the same local time, and Baker et al. (1993) argue that precipitated fluxes of electrons with energies > 1 MeV could be as much as 30 to 50 percent of the trapped flux levels. New measurements being made by the Solar, Anomalous, and Magnetospheric Particles Explorer (SAMPEX) may help to resolve these differences and determine whether REPs are as important for long term ozone variability as has been inferred from some studies (Callis et al., 1991).
Galactic Cosmic Rays
The Sun is not the only source of energetic particles that penetrate the Earth's atmosphere. Galactic cosmic rays (GCR), which are mostly protons and alpha particles, have typically higher characteristic energies (peaking in the range 0.1– 1 GeV per nucleon near Earth orbit) but lower fluxes than the solar (cosmic ray) protons. Carbon, nitrogen, oxygen, and other medium-mass nuclei are also present, together with a significant heavy ion population. Like solar energetic protons and relativistic electrons, galactic cosmic rays are strongly shielded from reaching the equatorial regions of the Earth's surface by the terrestrial magnetic field and by the thick neutral atmosphere. This shielding is energy dependent, deflecting incident particles with energies as high as 15 GeV at the magnetic equator. However, in the polar regions of the Earth and at high altitudes (> 5 Earth radii) the GCR component has direct access to the middle atmosphere.
At energies less than about 1– 2 Gev, the main mechanism for energy loss in the atmosphere is through ionization, and the particles stop well above the surface. At higher energies, nuclear interactions with atmospheric constituents become increasingly more important, giving rise to energetic secondary particles that can penetrate to the surface of the Earth. Among the products of these interactions are neutrons, which are created with relatively high energies and can subsequently be thermalized as they travel down through the atmosphere. The fact that these neutrons can be readily measured at the Earth's surface provides a convenient means of monitoring cosmic ray fluxes on a routine basis.
There is a relatively strong (5 to 15 percent) solar cycle modulation near Earth's orbit for galactic cosmic rays. This modulation, which is also energy dependent and increases at lower energies (below 1 GeV) to well over a factor of two, is in antiphase with the 11-year solar activity (sunspot) cycle and reflects the effects of changes in magnetic structures in the solar wind on the flow of cosmic rays into the solar system. Significant cosmic ray (Forbush) decreases also occur during large magnetic storms associated with coronal mass ejections and usually with solar flares (see Herman and Goldberg, 1978; Tinsley and Deen, 1991).
Cosmic ray secondary neutrons have an important by-product in the form of 14 C, or radiocarbon, produced by the reaction of a neutron with
14N, the most abundant atmospheric isotope. The 14 C becomes incorporated into the terrestrial carbon cycle and eventually into the carbon of living organisms. Once created, the 14C decays with a half-life of 5730 years, thereby providing the basis for the radiocarbon dating technique that has revealed much about biosphere evolution on millennial time scales. Laboratory measurements of the 14 C:12 C ratio in tree-ring dated wood have provided a chronicle of 14 C input variations and hence of variations in solar activity that now extend almost 10,000 years into the past (e.g., Stuiver and Braziunas, 1993).
Solar Forcing of the Middle Atmosphere
Changes in both solar UV radiation and in the flux of energetic particles influence the middle atmosphere directly. Determining the impact of human activities on upper stratospheric ozone and temperature requires a knowledge of the impact of this solar forcing background.
Effects from Variations in UV Irradiance
As discussed previously, variations in the solar flux during the 11-year activity cycle are still not well understood, particularly at the ultraviolet wavelengths that are responsible for the primary production and destruction mechanisms of middle atmosphere ozone. Changes in ozone and temperature resulting from the 27-day solar irradiance rotation cycle are better understood (Hood, 1987; Hood and Douglass, 1988; Chandra, 1991; Brasseur, 1993) and have confirmed the response of the middle atmosphere to changes in solar UV energy inputs. Ozone changes related to the 11-year irradiance cycle are expected to be larger than those detected over solar rotation time scales (Chandra, 1991) because the UV irradiance variation is larger (Lean, 1991) and the sensitivity of ozone to longer period forcing is greater (Brasseur, 1993).
Trend studies of middle atmosphere ozone and temperature observations provide evidence for long term changes associated with solar cycle variations in UV radiation, albeit with large uncertainties in the magnitude of the effect, which has thus far been measured for only one 11-year solar cycle. The results indicate that over decadal time scales solar cycle forcing
is a sizable contributor to trends in global ozone and upper stratospheric temperatures. This is illustrated in Figure 3.1, where the solar cycle in global total ozone is compared with the long term trend deduced from the TOMS observational record. Time series statistical analyses of the TOMS data (corrected for instrument degradation) indicate solar cycle changes in globally averaged total ozone of the order of 1.8 ± 0.3 percent, in phase with the activity cycle (Stolarski et al., 1991, Chandra, 1991, Hood and McCormack, 1992). Ground based data signify a slightly smaller effect, in the range 1.2 ± 0.4 percent to 1.5 ± 0.9 percent (Reinsel et al., 1988; Bojkov et al., 1990). The trends are larger at higher latitudes than in the tropics (Randel and Cobb, 1994; Reinsel et al., 1994a).
Both spacecraft and ground based observations indicate that significant solar cycle changes are also occurring in the vertical ozone profile. Global upper stratosphere (about 45 km) ozone is estimated from SBUV data to increase by 3 to 4 percent from the minimum to the maximum of the 11-year solar cycle (Hood et al., 1993). Similar results were found by Reinsel et al. (1994b) using northern hemisphere mid-latitude Umkehr data (revised for aerosol effects). Changes of this magnitude may also be occurring in the lower stratosphere (about 20 km), with a reduced solar cycle effect of about 1 percent near 30 km. For comparison, the data indicate upper stratosphere depletion at northern mid-latitudes of the order of 6 percent per decade and about 3 percent per decade in the lower stratosphere (Reinsel et al., 1994b), which thus may be masked by solar-cycle effects for a number of future solar cycles (Figure 3.1). In addition, solar cycle variations have been reported in both stratospheric temperatures and winds (Mohanakumar, 1989; Kodera and Yamazaki, 1990; Chanin and Keckhut, 1991; Hood et al., 1993).
That changing UV radiation throughout the 11-year solar activity cycle should evince a measurable effect on upper stratospheric ozone and temperature has been predicted by atmospheric modeling studies over the past decade. Assuming an irradiance increase at 205 nm from solar minimum to solar maximum ranging from 6 to 9 percent based on available spacecraft data, and using H I Lyman and He I EW surrogates for the variations in UV flux with time, Wuebbles et al. (1991) determined from a two-dimensional chemical-radiative-transport model of the global middle atmosphere a corresponding increase in total ozone of 1.2 to 1.7 percent. The largest local changes, about 2 to 3 percent, occurred at about
40 km altitude, while maximum temperature changes, 0.75 to 1 K, were determined near the stratopause. Contrary to the observations, significant solar-induced ozone changes in the lower stratosphere are not simulated by two-dimensional middle atmosphere models; equally small changes were found in a 2D model study by Huang and Brasseur (1993). Furthermore, when downward transport of the upper atmosphere nitric oxide (NO) enhancement at solar cycle maximum is included in 2D model simulations, there is significant ozone depletion, resulting in an antiphase global total ozone change, also contrary to the observations.
Clearly, substantial uncertainties exist in determining the response of middle atmosphere ozone and temperature to solar radiative forcing. In part, this is because the magnitude of the solar flux variations at the important wavelengths (200–300 nm) are still uncertain. Perhaps more importantly, a primary limitation of the 2D models is that dynamical feedbacks are not available to amplify initial perturbations of the middle atmosphere by changing UV energy inputs (Balachandran and Rind, 1994). To clarify the discrepancies between observed and modeled ozone response to solar forcing it is essential to continue to monitor the relationships between variations in the middle atmosphere parameters and observed solar irradiance variations, both for the 27-day rotation and the 11-year solar cycle, as well as longer term trends. Because the prime space based global data sets extend over only one 11-year solar activity cycle, more definite studies of, for example, the role that might be played by dynamics and by the QBO in the observed solar cycle variation of ozone, await longer observational records.
Effects from Solar Proton Events
During a solar proton event, the primary energetic protons produce secondary electrons by impact on atmospheric molecules, and these electrons can efficiently dissociate molecular nitrogen, producing a range of odd-nitrogen species by subsequent chemical reactions. The odd-nitrogen, in turn, catalytically destroys ozone in the stratosphere. This affects the Earth's surface environment directly through an increase in the solar ultraviolet flux and may have indirect effects via coupling of the middle atmosphere to the troposphere. Ozone is also depleted at higher altitudes in the mesosphere as a consequence of the dissociation of water
vapor, producing odd-hydrogen that destroys ozone at these altitudesthrough a catalytic reaction sequence.
The odd-nitrogen and odd-hydrogen effects on ozone have been observed directly by satellites and rockets during solar proton events (e.g., Weeks et al., 1972; Crutzen et al., 1975; Heath et al., 1977; Thomas et al., 1983) and have been studied theoretically by several groups (e.g., Jackman et al., 1990; Reid et al., 1991). As demonstrated in Figure 3.4, the August 1972 solar proton event was one of the largest events in the past 30 years. Substantial increases in nitrogen oxides were expected to be associated with this event. Available satellite measurements and modeling studies indicate ozone depletions >20 percent in the upper stratosphere, with depletions >15 percent persisting for about two months after the event, indicating a slower recovery time than is estimated by two-dimensional modeling studies (Jackman et al., 1990).
In a sense, each major solar proton event can be looked on as a controlled experiment, in which a calculable amount of NOx is introduced into the atmosphere, and the subsequent effects can be monitored. These events are thus a potentially valuable source of information on the behavior of the atmosphere, providing a unique means of testing the validity of model predictions (Jackman and McPeters, 1987; Jackman, 1991). Also, over the longer time scales of the 11-year solar cycle, variations in the downward transport of nitrogen oxides produced by auroral particle precipitation at altitudes in the thermosphere (Garcia et al., 1984) and other solar particle effects could further affect stratospheric temperature and ozone. Understanding these processes, however, requires accurate monitoring of the particle influx at the top of the atmosphere, since this is the basic information needed to calculate the NOx production. Since the most relevant effects are those occurring at the lowest levels in the atmosphere, the high-energy spectrum must be reliably known beyond energies of 100 MeV. Routine spacecraft monitoring of energetic-particle fluxes has generally been carried out only for energies much less than this.
Effects from Relativistic Electron Precipitation
Various studies have suggested that relativistic electron precipitation events may be important contributors to the odd-nitrogen budget of the middle atmosphere (Thorne, 1980; Baker et al., 1987; Callis et al., 1991).
Thorne (1980) indicated that the odd-nitrogen and odd-hydrogen formed in the upper stratosphere and lower mesosphere by the ion chemistry associated with these precipitation events can lead to destruction of ozone. Although these studies suggest that most of the production of the odd-nitrogen from such events occurs at subauroral latitudes in the lower mesosphere, the downward polar transport during the winter and spring may provide a source of stratospheric nitrogen oxides. Callis et al. (1991) have suggested that this phenomenon may explain a significant fraction of the measured lower stratospheric ozone decrease in the 1978-1986 period. However, as discussed earlier, there remain many uncertainties associated with the amount of nitrogen oxides produced and the resulting effects on lower stratospheric ozone.
Ultraviolet Radiation Reaching the Biosphere
The amount of solar ultraviolet radiation reaching the biosphere is extremely sensitive to the amount of overhead ozone, especially at the UV-B wavelengths that coincide with a sharp decrease in the ozone (O3 ) absorption at about 310 nm (Figure 1.2), the edge of the O3 Hartley band. Therefore, the amount of ozone in the middle atmosphere, and its response to natural and anthropogenic influences, is extremely important for determining potential impacts on the UV-sensitive biosphere (Madronich, 1992). Small reductions in ozone column density are expected to cause an increase in the incidence of skin cancer (Madronich and de Gruijl, 1993) and in blindness due to UV-B induced cataracts, as well as decreased agricultural and fisheries production. Biological immune systems, as well as the oxidizing capacity of the troposphere, are also expected to be affected.
Although measurements indicate that total ozone has been decreasing over the past two decades, and theoretical calculations predict a corresponding increase in the UV radiation reaching the Earth's surface (Madronich, 1992), experimental evidence for this increased UV exposure is at present inconclusive. Factors such as tropospheric ozone, sulfate particles in the atmosphere, and cloudiness can produce opposite effects (Crutzen, 1992). Substantial downward, upward, and null trends have all been reported (Justus and Murphey, 1994). At the same time, there are
questions about the adequacy of the primary device used in these measurements, the Robertson-Berger meter. The measurements require not only accurate absolute radiometry, but also high wavelength resolution and accuracy, because the ozone absorption spectrum at the edge of the Hartley band is so steep. The spectral intensity of solar radiation received at Earth drops by eight orders of magnitude in the spectral band from 300 to 280 nm. The data from the existing ground based network probably lack adequate calibration stability, limiting their suitability for trend analyses.