Estimating Solar Irradiance Since 850 CE
Solar total and spectral irradiance are estimated from 850 to 1610 by regressing cosmogenic irradiance indices against the National Oceanic and Atmospheric Administration Solar Irradiance Climate Data Record after 1610. The new estimates differ from those recommended for use in the Paleoclimate Model Intercomparison Project (PMIP4) in the magnitude of multidecadal irradiance changes, spectral distribution of the changes, and amplitude and phasing of the 11‐year activity cycle. The new estimates suggest that total solar irradiance increased 0.036 ± 0.009% from the Maunder Minimum (1645–1715) to the Medieval Maximum (1100 to 1250), compared with 0.068% from the Maunder Minimum to the Modern Maximum (1950–2009). PMIP4’s corresponding increases are 0.026% and 0.055%, respectively. Multidecadal irradiance changes in the new estimates are comparable in magnitude to the PMIP4 recommendations in the ultraviolet spectrum (100–400 nm) but somewhat larger at visible (400–700 nm) and near‐infrared (700–1,000 nm) wavelengths; the new estimates suggest increases from the Maunder Minimum to the Medieval Maximum of 0.17 ± 0.04%, 0.030 ± 0.008%, and 0.036 ± 0.009% in the ultraviolet, visible, and near‐infrared spectral regions, respectively, compared with PMIP4 increases of 0.17%, 0.021%, and 0.016%. The uncertainties are 1σestimates accruing from the statistical procedures that reconstruct irradiance in the Medieval Maximum relative to the Modern Maximum, not from the specification of Modern Maximum irradiances per se. In the new estimates, solar irradiance cycle amplitudes in the Medieval Maximum are comparable to those in the Modern Maximum, whereas in the PMIP4 reconstruction they are at times almost a factor of 2 larger at some wavelengths and differ also in phase.
Solar irradiance is Earth’s primary energy input. It establishes the thermal and dynamical structure of the terrestrial environment and is the primary external cause of terrestrial variability. The specification of solar irradiance over multiple centuries is requisite input for numerical simulations of climate variability prior to the industrial epoch that provide a baseline against which to evaluate contemporary anthropogenic influences. For this purpose, the Paleoclimate Model Intercomparison Projects PMIP3 (Schmidt et al., 2011, 2012) and PMIP4 (Jungclaus et al., 2017) developed reconstructions of the Sun’s total and spectral irradiance since 850 CE that are compatible with the absolute scale and variability of irradiance inputs recommended for climate change simulations in the subsequent industrial epoch (Lean, 2009; Matthes et al., 2017).
Because reliable, accessible, ongoing solar irradiance specifications are necessary for a range of Earth science research and applications, the U.S. National Oceanic and Atmospheric Administration (NOAA) implemented the Solar Irradiance Climate Data Record (CDR, Coddington et al., 2016) in 2015. The Solar Irradiance CDR includes estimates of total and spectral solar irradiance made using models constructed to replicate variations in contemporary space‐based observations. Currently, the NOAA CDR irradiance specifications (v02r01) extend from 1610 to the present but not, as yet, from 850 to 1610.
Continuous space‐based observations of total solar irradiance (TSI) began in late 1978, when the Nimbus 9 satellite carried the Hickey‐Freidan solar radiometer into Earth orbit, followed in 1980 by the launch of the Active Cavity Radiometer Irradiance Monitor on the Solar Maximum Mission. Thereafter, a dozen or more solar radiometers on space‐based platforms have continued the record, including the Total Irradiance Monitor (TIM) on the Solar Radiation and Climate Experiment (SORCE, Rottman, 2005) whose observations enable the model that specifies TSI for the NOAA CDR. Lean (2017) summarizes the space‐based historical irradiance observations; the record continues with the recently launched state‐of‐the‐art TIM of the Total and Spectral Solar Irradiance Sensor (TSIS) on the International Space Station (Richard et al., 2011).
Compared with the database of TSI observations, that of spectral irradiance observations is more limited in temporal coverage, has less certain absolute calibration, and reduced repeatability, especially on decadal time scales. Thus far, spectral irradiance observations over multiple cycles exist only at ultraviolet wavelengths less than 400 nm albeit discontinuously. The launch of the Solar Mesosphere Explorer (Rottman, 2006) in 1980 initiated systematic ultraviolet irradiance observations for a decade. Solar spectroradiometers on the Upper Atmosphere Research Satellite continued the record from 1992 to 2003 (Dessler et al., 1998), on SORCE from 2003 to the present (see Lean, 2017, for overview) and on the International Space Station into the future (Richard et al., 2011). Additional observations of solar spectral irradiance made in pursuit of ozone concentration measurements, such as by the Ozone Monitoring Instrument, also contribute to the solar spectral irradiance database (Marchenko et al., 2016). Systematic, continuous observations of solar spectral irradiance at wavelengths from 400 to 2000 nm exist only since 2003, made by the Solar Irradiance Monitor (SIM, Harder et al., 2009) on SORCE.
Models that combine the influences of the two primary solar sources of irradiance variability, namely dark sunspots and bright faculae, reproduce the observed space‐based TSI variations with high fidelity (Fröhlich & Lean, 2004; Kopp & Lean, 2011). For example, the Naval Research Laboratory Total Solar Irradiance (NRLTSI2) model, which the NOAA CDR utilizes to estimate both present and historical irradiance variations (Coddington et al., 2016), inputs a sunspot darkening function calculated from direct observations of sunspot areas and locations on the Sun’s surface and the Mg irradiance index as a facular proxy; the correlation of this model with daily averaged TIM observations (from 2003 to 2016) is 0.96. The Spectral and Total Irradiance Reconstructions (SATIRE) model derives its two sunspot (dark sunspot umbra and penumbra) and two facular (bright faculae and network) inputs from solar magnetograms (Krivova et al., 2010); the correlation of the SATIRE model of TSI with the TIM observations is also 0.96.
The same sunspot and facular solar features that cause total irradiance to vary also influence the spectral irradiance, their net effects being strongly wavelength dependent. The Naval Research Laboratory Solar Spectral Irradiance (NRlSSI2) model specifies solar spectral irradiance for the NOAA CDR with wavelength‐dependent combinations of sunspot and facular indices. The relative strengths of the sunspot and facular influences at different wavelengths are estimated from direct observations made by the Solar Stellar Irradiance Comparison Experiment (SOLSTICE) and SIM on SORCE (Snow et al., 2010). The SATIRE model uses a theoretical model of stellar atmospheres (Unruh et al., 1999) to specify the wavelength dependence of its sunspot and facular inputs. Unresolved instrumental trends thus far preclude observational determination of solar cycle spectral irradiance changes (Lean & DeLand, 2012), except, arguably, for SOLSTICE observations of the brightest and most variable HI Lyman α emission at 121.5 nm. The correlation of the NRLSSI2 model with the daily SOLSTICE SORCE Lyman α irradiance (from 2003 to 2016) is 0.99.
Models of solar irradiance variability such as NRLSSI2 and SATIRE expand and normalize the limited spectral and time domains of the observations. They provide regularly gridded specifications of solar spectral irradiance from the far ultraviolet to the far infrared, and in epochs prior to 1978, in formats suitable for input to climate and atmospheric model simulations (Matthes et al., 2017). To reconstruct historical solar irradiance variations, the models incorporate proxy indicators of the sunspot and facular sources, synergistically for total and spectral irradiance. Direct observations of the areas and locations of sunspots are available since 1882, but sunspot numbers are the only direct indicator of solar activity from 1610 to 1882. The NRLTSI2 and NRLSSI2 models estimate annual irradiance variations from 1610 to 1882 using direct correlations of annual mean sunspot numbers with total and spectral irradiance estimated after 1882. SATIRE algorithmically transforms the sunspot number to estimates of the model’s four separate inputs (dark sunspot umbra and penumbra and bright faculae and network; Kopp et al., 2016; Krivova et al., 2010).
Estimates of solar irradiance prior to 1610 rely on the 10Be and 14C cosmogenic indicators of solar activity extracted from ice cores and tree rings (Delaygue & Bard, 2011; Roth & Joos, 2013; Steinhilber et al., 2012, 2009). Cosmogenic isotopes contain information about solar activity because the Sun is the source of the heliospheric magnetic flux that modulates the flow of galactic cosmic rays that produce these isotopes of gases in Earth’s atmosphere (McCracken et al., 2004, 2013; McCracken & Beer, 2007). Figure 1 shows two different reconstructions of TSI since 850 developed as part of PMIP4, using more recent cosmogenic isotope indices and irradiance variability models than were available at the time of PMIP3. The PMIP4 irradiance reconstructions are synergistic with the absolute scale and variability of the irradiances that Matthes et al. (2017) recommend for use in Intergovernmental Panel on Climate Change’s Sixth Assessment Report simulations, namely, the average of irradiance modeled by NRLTSI2 (total) and NRLSSI2 (spectral) and SATIRE. Of the two different PMIP4 irradiance reconstructions shown in Figure 1, that based on 14C (rather than on 10Be) is specifically recommended for use in the Coupled Model Intercomparison Project (Phase 6) numerical model simulations.
This paper estimates total and spectral solar irradiance from 850 to 1610 consistent in magnitude and variability with the NOAA Solar Irradiance CDR from 1610 to 2016. In addition to extending the Solar Irradiance CDR prior to 1610, the goal is to provide independent, alternative, irradiance reconstructions for comparison with, and assessment of, the PMIP4 recommendations. The PMIP4 approach converts cosmogenic isotopes to sunspot numbers then calculates solar irradiance prior to 1850 using a SATIRE‐type numerical transformation of this single solar activity index to estimate the model’s four separate inputs. Vieira et al. (2011) report that this simplification of the magnetic flux inputs to the SATIRE model is a major source of uncertainty in its Holocene irradiance reconstructions. In contrast, the current approach estimates solar irradiance prior to 1610 using direct parameterizations of cosmogenic indices with the NOAA Solar Irradiance CDR after 1610.
full paper here: