2.2 Solar Radiative Forcing

CHAPTER 2.2

SOLAR RADIATIVE FORCING

Natalie A. Krivova1 and Ilaria Ermolli2

1 Radiative forcing

Radiative forcing (RF) quantifies the impact of various factors on Earth’s surface temperature and climate. In equilibrium, the global average temperature of the Earth is determined by a balance between the incoming energy absorbed in the atmosphere and by the surface and the energy of the thermal infrared radiation emitted back into space, so that the net radiative flux (globally and annually averaged) at the top of the atmosphere3 is zero. When the amount of the incoming or outcoming radiation changes, then before a new equilibrium is established the net flux at the top of the atmosphere is not zero. Thus, radiative forcing is the instantaneous change in the value of the net downward radiative flux. Radiative forcing is defined positive when the amount of energy retained by the Earth- atmosphere system increases. This could be due to an increase in the incoming or a decrease in the outcoming energy and leads to an increase in the Earth’s temperature.

2 Solar irradiance

The dominant energy source to Earth’s climate system is the Sun. Solar energy is produced in the Sun’s core, where hydrogen atoms fuse to form, via a chain reaction, helium. Photons released in this process travel inside the Sun until, tens or even hundreds of thousands years later, they reach up to the so-called solar visible surface (or the photosphere), above which the solar atmosphere is transparent for the photons. They escape as visible light and part of them reaches the Earth just minutes later.

1 Max-Planck-Institut fürSonnensystemforschung, 37077 Göttingen,Germany 2 INAF, Osservatorio Astronomico di Roma, Monte Porzio Catone, Italy 3 Radiative forcing can also be defined at the tropopause. c EDP Sciences 2015 DOI: 10.1051/978-2-7598-1733-7.c108 68 Earth’s climate response to a changing Sun

The total solar electromagnetic energy flux received at the top of the Earth’s atmosphere and reduced to the mean distance from the Sun to the Earth (one astronomical unit, AU) is called the total solar irradiance (TSI). This definition takes care to differentiate between changes in the solar radiative output itself and changes in the Earth-Sun distance (see Infobox 2.1) or the interaction between the solar radiation and the Earth’s atmosphere. Solar electromagnetic energy is emitted over essentially entire spectrum. Al- most half of this energy originates in the range visible to the human eye (roughly 400 to 800 nm), more than 40% come from the infrared, IR, range (wavelengths above 800 nm), and only less than 10% are contributed by the ultraviolet, UV, radiation (wavelengths below 400 nm). Solar irradiance measured at a certain wavelength or within a wavelength interval (in units of W m−2 per unit wavelength interval, e.g. W m−2nm−1) is called spectral solar irradiance (SSI). Total solar irradiance is one of the more recently discovered variables of the Sun, generally being referred to as the solar constant in the older literature.

3 Measurements of solar irradiance

3.1 TSI Measurements of TSI are done with absolute radiometers. They include a cavity with a blackened inner side, operating as a black body to efficiently absorb the incident radiation, which heats the cavity. A built-in electrical calibration heater maintains the heat flux constant by adjusting the supplied heating power. The electrical power required to maintain the equilibrium is proportional to the incoming radiative power absorbed by the cavity. Due to atmospheric extinction, measurements of total solar irradiance sufficiently precise to reveal its variability are only possible from space and have been regularly carried out since 1978 by over a dozen of different radiometers (Figure1). On an absolute scale, results from individual experiments differ significantly. Recent tests and experiments have helped to identify main sources of the offsets. The TSI value of 1360.8 0.5 Wm−2 recorded by SORCE/TIM in 2008 is currently believed to best represent± the solar minimum conditions. The relative TSI changes measured by space-borne radiometers agree quite good with each other on time scales of days to years. Thus, TSI varies on all observable time scales, from minutes to decades. Most prominent are the short- term fluctuations on time scales of days to roughly a week and the modulation in phase with the 11-year solar activity cycle (see Figure1). But there are also differences between the data coming from individual experiments, most importantly in the longer-term trends. This is because changes in the sensitivity (i.e., the calibration) of the instruments due to their aging and degradation once they are launched and exposed to the solar light, are quite individual and often tricky to assess. Owing to the offsets in the absolute values, combining the data from individual instruments into a single composite record is non-trivial. Natalie A. Krivova and Ilaria Ermolli: Solar radiative forcing 69

Fig. 1. Space-borne measurements of the total solar irradiance (TSI) covering the period from 1978 to 2014. Individual records are shown in diﬀerent colors, as labeled in the plot. The bottom part of the plot shows the monthly mean sunspot number. Courtesy of G. Kopp (http://spot.colorado.edu/koppg/TSI/). Abbreviations: Active Cavity Radiometer Irradiance Monitor (ACRIM) instruments on the Solar Maximum Mission (SMM), Upper Atmosphere Research Satellite (UARS), and ACRIMSat; Earth Radiation Budget (ERB) instrument on Nimbus-7; Earth Radiation Budget Satellite (ERBS), the Earth Radiation Budget Experiment (ERBE) on the Earth Radiation Budget Satellite (ERBS) National Oceanic and Atmospheric Administration (NOAA), Precision Monitor Sensor (PREMOS) on PICARD, Solar Variability Experiment (SOVA) on EUropean Re- trievable Carrier of ESA (EURECA), Solar Variability and Irradiance Monitor (SOVIM) on the International Space Station (ISS), Total Irradiance Monitor (TIM) on Solar Ra- diation and Climate Experiment (SORCE), Variability of Solar Irradiance and Gravity Oscillations (VIRGO) on SoHO.

Currently, three different composites exist (see Figure2). The most critical discrepancy concerns their conflicting secular trends, best seen as the difference in the TSI levels during activity minima in 1986, 1996 and 2008. Models of solar irradiance (see Section5) return an either downward trend (Figure2), as in the PMOD (Physikalisch-Meteorologisches Observatorium Davos) composite, or in case of most proxy models, no trend. None of the models returns an upward or alterating trend. 70 Earth’s climate response to a changing Sun

Fig. 2. The ACRIM (red), IRMB (green) and PMOD (blue) TSI composite records smoothed over 181 days. Also plotted is the SATIRE-S model reconstruction of TSI. The vertical dashed lines mark solar minima. From Yeo et al.(2014).

3.2 SSI

Measurement of the solar spectral irradiance is more sophisticated. Very low am- plitudes of the variability in the visible and IR ranges, a rapid increase in the variability in combination with significant decrease of the amount of the radiation with decreasing wavelength (Figure3) and the need for a meaningful spectral resolution are only some of the complicating factors. Thus in addition to the high radiometric accuracy and precision, a spectrometer has to maintain a sufficient wavelength accuracy and account for the wavelength-dependent long-term degradation effects. Regular monitoring of the solar spectrum in the UV (below 400 nm) also started in 1978. Most of our knowledge is based on the results of two experiments, SOLSTICE (Solar Stellar Irradiance Comparison Experiment) and SUSIM (Solar Ultraviolet Spectral Irradiance Monitor), onboard UARS (the Upper Atmosphere Research Satellite), in operation from 1991 to 2001 and 2005, respectively. SSI monitoring over a broad range from the UV to the IR (up to about 2 400 nm) with absolute radiometric calibration has been carried out by SORCE/SIM since 2003. Whereas most of the solar radiant energy is emitted at visible and infrared wavelengths, the variability is significantly stronger in the UV spectral range (Figure3). The amplitude of the variation grows from about 0.1% in the visible and even less in the IR to almost 100% around the Ly-α at 121.6 nm, which is the strongest line in the solar spectrum. The contribution of the UV (below 400 nm) part of the spectrum to the total irradiance variability most probably exceeds 50%, although the exact number is still a matter of debate (Figure4). Thus, the more recent observations by SORCE/SIM suggest stronger variability in the UV than had been measured before SORCE and predicted by the models. This exces- sive variability would need to be partly compensated by a presumably anti-phase (with the solar cycle) variation in the visible (400–700 nm). This pattern of SSI variability is, however, in contradiction not only with other, partly contemporary, observations, but also with the up-to-date models, which are at the same time Natalie A. Krivova and Ilaria Ermolli: Solar radiative forcing 71

Fig. 3. Top: Reference solar spectrum recorded in April 2008 within the Whole Heliosphere Interval (WHI) project. Bottom: Relative SSI variability as observed by UARS/SUSIM (red curve) between the maximum of cycle 23 (March 2000) and the preceding minimum (May 1996), as well as by SORCE/SOLSTICE (light blue) and SORCE/SIM (dark blue) (Harder et al., 2009) between April 2004 and December 2008. Also shown is the variability between 2000 and 1996 predicted by the SATIRE-S model (green). For each period, averages over one month are used. Negative values are indi- cated by dotted segments. Following Solanki et al.(2013). Abbreviations: UARS - Upper Atmosphere Research Satellite; SUSIM - Solar Ultraviolet Spectral Irradiance Monitor; SORCE - Solar Radiation and Climate Experiment; SOLSTICE - Solar Stellar Irradiance Comparison Experiment; SIM - Spectral IrradianceMonitor; SATIRE-S - Spectral and Total Irradiance Reconstruction for the Satellite era. 72 Earth’s climate response to a changing Sun

Fig. 4. Relative contribution of the UV (200–400 nm), visible (400–700 nm), near-IR (700–1000 nm) and IR (1000–2430 nm) ranges to the TSI change over the same period. For SORCE/SIM, only the period between 2004 and 2008 can be considered. For other data and models, the plotted relative differences are between solar maximum an minimum. Within each wavelength bin, from left to right: UARS/SUSIM (green), NRLSSI (black), SATIRE-S (blue), COSI (purple), OAR (light blue), SCHIAMACHY/ Mg index empirical model (orange), SORCE/SIM (red). Following Ermolli et al.(2013). Abbrevia- tions: UARS - Upper Atmosphere Research Satellite; SUSIM - Solar Ultraviolet Spectral Irradiance Monitor; NRLSSI - Naval Research Laboratory Solar Spectral Irradiance; SATIRE-S - Spectral and Total Irradiance Reconstruction for the Satellite era; COSI - COde for Solar Irradiance; OAR - Osservatorio Astronomico di Roma; SCIAMACHY - SCanning Imaging Absorption spectroMeter for Atmospheric CHartographY); SORCE - Solar Radiation and Climate Experiment; SIM - Spectral IrradianceMonitor. very successful in reproducing all other available TSI and SSI observations (see Figure4). A growing body of evidence has been found that SIM/SORCE measurements might still suffer from unaccounted instrumental effects but the resolution of the debate is still pending.

4 Origin of solar irradiance variations

Variations of solar irradiance on very short (minutes to hours) time scales are irrelevant to Earth’s climate and are not discussed here. Variations on time scales longer than one day are driven by changes in the amount and distribution of the solar surface magnetic field. The magnetic field emerging on the surface manifests itself in form of different brightness features, such as sunspots, faculae, plage or network. Natalie A. Krivova and Ilaria Ermolli: Solar radiative forcing 73

Fig. 5. Left: Continuum image of the Sun taken by the SoHO/MDI instrument on 30 March 2001. Courtesy of SoHO/MDI consortium. SOHO is a project of international cooperation between ESA and NASA (http://sohowww.nascom.nasa.gov/gallery/ bestofsoho.html). Right: Active regions with sunspots and faculae near solar limb. Im- age by Dan Kiselman and Mats Löfdahlin G-band (430.5 nm) taken on June 29, 2003 with the Swedish 1-m Solar Telescope (SST). The SST is operated on the island of La Palma by the Institute for Solar Physics of the Royal Swedish Academy of Sciences in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofsica de Canarias (http://www.solarphysics.kva.se/gallery/images/2003/). Sunspots are strong concentrations of the magnetic field, in which the heat transport from the interior is inhibited. Therefore they are cooler than the sur- rounding photosphere and appear dark (Figure5). Sunspots crossing the visible solar hemisphere cause a darkening of the Sun by up to 0.3% on time scales of hours to about a week (Figure1). The magnitude of the brightness dip is roughly proportional to the projected area of the spot. Smaller magnetic field concentrations cause brightenings on the solar surface termed faculae, plage and the network. Faculae and plage are usually observed in active regions, in the vicinity of sunspots (Figure5). The weaker network elements are found in the quiet Sun all over the disc. Faculae and the network are responsible for the overall brightening by about 0.1% (for recent cycles) from activity minimum to maximum (Figure1).

5 Models and reconstructions of solar irradiance variations

Records of TSI and SSI observations, covering less than four decades, are too short to allow reliable linking of climate change on Earth with solar irradiance variability. It is therefore necessary to extend the irradiance time series back in time with the help of suitable models. The aim of such models is two-fold: (1) to understand and reproduce the directly measured irradiance variations and (2) to reconstruct past (and ideally, also predict future) irradiance changes. 74 Earth’s climate response to a changing Sun

A number of models have been constructed that can be divided into two classes: the so-called proxy and semi-empirical models. Proxy models regress different indices of solar magnetic activity (two or more to describe sunspot darkening and facular brightening) to the measured irradiance variation. The need to have a record of direct irradiance observations that is stable over a sufficient period of time makes in particular reconstructions of SSI with this technique circuitous. The other type of models follows a more physics-based approach. The brightness spectra of individual surface components (sunspots, faculae, plage, network and the quiet Sun) are calculated using radiative transfer codes from corresponding semi-empirical model atmospheres. The latter describe how the temperature and density of the solar atmosphere change with height. The corresponding features are then identified on the solar visible surface and their fractional area disc coverage is quantified. Typically high resolution full-disc solar images are employed for this, but disc-integrated indices are also often used, e.g. when going back in time to periods when such maps were not available. The irradiance, as a function of time, is then calculated by weighting the intensity spectra of each surface component with the corresponding area coverage and summing up all contributions. State-of- the-art models reproduce TSI variations in great detail (see, e.g., Figure2). Reconstructions of solar irradiance into the past require suitable proxies of solar activity. Most of the reconstructions on time scales of centuries rely on the historical record of the sunspot number, which is the longest record of direct solar observations. It goes back to 1610 and describes the cyclic evolution of the active regions fairly accurately. However, the emergence rate of the weaker ephemeral magnetic regions does not exactly follow the evolution of active regions. The activity cycles of these weak elements are significantly weaker in amplitude (if any at all) and are most probably somewhat shifted and stretched in time (i.e., their activity cycles are longer than the corresponding sunspot cycles, which have an average period of roughly 11 years). It has been proposed that long-term changes in the emergence rate of such weak features is the main source of the secular change in irradiance. As no direct long-term proxy of this component exists, the magnitude of the secular variation remains heavily debated (see Figure6 for different TSI reconstructions since 1610). The situation is further complicated by the fact that the secular change over the period of spaceborne irradiance observations was weak and rather uncertain (Section3). Thus most models agree with the direct measurements within the measurement uncertainty, while diverging on longer scales. The esti- mated magnitude of the change in TSI since 1700 ranges between 0.6 and 3 W m−2. Estimates of the irradiance on yet longer time scales rely on the records of the cosmogenic isotopes 10Be and 14C (Chapter 2.5).

6 Summary

Measurements of solar irradiance are now available for around 3.5 solar cycles. These data are invaluable. Neverthless, the debate about the presence or absence of a secular trend is ongoing. To understand the role of the Sun on climate, longer Natalie A. Krivova and Ilaria Ermolli: Solar radiative forcing 75

Fig. 6. Various reconstructions of TSI since 1610. Vertical bars denote uncertainties of the models plotted in the same colours and identified in the plot. The dark blue bar shows the estimate of the possible TSI change following Schrijver et al.(2011); no reconstruction done. The uncertainty of the model by Shapiro et al.(2011), ±3 W m−2, extends downward outside the plot. The revised estimate for this model by Judge et al. (2012) is marked by the blue bar and arrow. The black dotted line marks the TSI value at the solar minimum in 2008 according to SORCE/TIM measurements. Other references are (Steinhilber et al.(2009); Krivova et al.(2010); Dasi-Espuig et al.(2014), priv. comm.) time series are imperative. These can only be obtained using suitable models. To ensure the correct interpretation of the physical processes, successful models have, in the first step, to reproduce the observed changes. State-of-the-art models have progressed to a high level of accuracy on time scales up to the solar cycle, although further progress is still needed and possible. On longer (and for climate studies more important) time scales, a number of reconstructions going back to 1610 or even longer exist. But the magnitude of any secular trend remains a matter of intense debate. Recent esimates converge to the value of about 1 to 1.5 W m−2, but range from 0.6 to 3 W m−2, thus differing by a factor of five. Further work in this direction is of great importance.