Solar Energy xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

A reevaluation of the solar constant based on a 42-year total solar irradiance

time series and a reconciliation of spaceborne observations
Christian A. Gueymard1
Solar Consulting Services, Colebrook, NH, USA

A R T I C LE I N FO A B S T R A C T

Keywords: A reevaluation of the solar constant is undertaken here to take into account the progress in space radiometry that
Solar constant has occurred since the early 2000s. Various sources of spaceborne total solar irradiance (TSI) observations are
Total solar irradiance investigated here, including the long-term ACRIM and PMOD composites, as well as recent observations from the
Space radiometry SORCE-TIM, TCTE-TIM, and PICARD-PREMOS instruments. A proxy model is constructed using daily data of
Solar variability
sunspot number, radio flux at 10.7 cm, and MgII index, as predictors for TSI over the 42-year period 1976–2017.
These daily estimates are used to fill in 9.7% of missing TSI observations during that period. By comparison with
these proxy estimates, the PMOD composite appears generally more reliable than the ACRIM composite before
2003, and particularly before 1981. The 42-year time span is separated into nine periods, each defining the
revised TSI daily values from one or more sources that are selected based on the trend of their resemblance with
the proxy model. A final correction is added to emulate the highly accurate absolute calibration of PREMOS.
Based on the resulting TSI reconstruction, a revised solar constant value of 1361.1 W/m2 is obtained, with a
standard uncertainty of 0.5 W/m2. The revised solar constant is ≈5 W/m2 less than the previous values pro-
mulgated in ASTM and ISO standards. A revision of these standards is thus highly recommended.

1. Introduction Abbot (1911) mentions 13 SC estimates (between 1185 and 2371 W/

The magnitude of the solar output and its variability over various
time scales condition virtually all geobiological processes on Earth.
Among many other important phenomena, this planet’s weather and
climate are directly affected by the energy received from the Sun at
each instant, and in the long-term. Among the numerous terrestrial
energy-related applications that are dependent on the Sun’s output,
solar power is of course at the forefront. The exploration of space also
requires solar power as a reliable energy source for satellites and
spacecraft.
The early days of solar radiation research were directly related to
the quantification of the solar power output and its variations. This
solar output is still usually referred to as the “solar constant” (SC), even
though it is now known to actually fluctuate due to solar activity. In
1837, the first acknowledged pioneer, Pouillet, reported measurements
with an early type of pyrheliometer and obtained a value of 1230 W/m2
(after unit conversion using the thermochemical calorie definition,
1 cal = 4.184 J). In the early 1900s, Abbot continued the work of
Langley and developed the Smithsonian Institution’s Solar Constant
Program (1902–1957), aimed at finding a link between sun’s cyclical
activity and its power output (Hoyt, 1979; Hoyt and Schatten, 1997).
m2) that were published between 1837 and 1908. His own estimate at
that time was 1340 W/m2, which is remarkably close to the current
observations, despite the large experimental uncertainties that existed
then. Based on elaborate measurements from high mountains or bal-
loonsondes, the values he obtained over the years ranged between
about 1320 and 1550 W/m2. These variations were mostly attributed
(wrongly for the most part) to solar activity. Hoyt and Schatten (1997)
listed 36 SC values that were proposed between 1838 and 1993. A
shorter list, covering the period 1940–2004, can be found in Gueymard
(2006), with 16 SC values in the range 1322–1429.5 W/m2. The most
recent value, 1366.1 W/m2 was initially proposed by ASTM (2000),
thus updating the previous value of 1353 W/m2 originally proposed by
Thekaekara (ASTM, 1974), and was then confirmed by (Gueymard,
2004; hereafter G04).
To account for the temporal variability in the sun’s instantaneous
output, it is now termed total solar irradiance (TSI). In what follows, the
term ‘‘solar constant’’ is used only to describe the long-term mean value
of TSI. Using spaceborne instruments, the continuous monitoring of TSI
has started in 1978 and has shown that TSI was indeed somewhat
variable over time, as a consequence of the sun’s 27-day rotation cycle
and of complex mechanisms behind the 11-year solar cycle (Fröhlich

E-mail address: Chris@SolarConsultingServices.com.

1
ISES member.

https://doi.org/10.1016/j.solener.2018.04.001
Received 12 February 2018; Received in revised form 28 March 2018; Accepted 1 April 2018
0038-092X/ © 2018 Elsevier Ltd. All rights reserved.

Please cite this article as: Gueymard, C.A., Solar Energy (2018), https://doi.org/10.1016/j.solener.2018.04.001
C.A. Gueymard
and Lean, 2004; Kuhn and Armstrong, 2004). What was referred to as
the “solar constant” in the past will be evaluated below by calculating
the long-term average TSI at the Earth’s top of atmosphere for the
average sun-earth distance (1 ua).2
As G04 mentioned, a discrepancy existed between the 1366.1 W/m2
SC value just mentioned and the TSI measurements made with the Total
Irradiance Monitor (TIM) radiometer. That instrument had just started
collecting data, as part of the Solar Radiation and Climate Experiment
(SORCE) mission (http://lasp.colorado.edu/home/sorce/), and was
reporting values typically ≈5 W/m2 lower than older instruments in
space. Since the publication of G04, a small number of new estimates of
SC were proposed, sometimes based on data from new spaceborne in-
struments. In 2007, ISO Standards 14222 (ISO, 2013) and 21348 (ISO,
2007) recommended a value of 1366 W/m2, almost identical to ASTM’s.
ASTM E490 was reapproved in 2014, and ISO 21348 was reapproved in
2015, even though significantly lower SC values had been proposed in
the mean time. For instance, Kopp and Lean (2011) suggested a “most
probable value” of 1360.8 W/m2 for the representation of TSI at solar
minimum, based on a few years of measurement with the SORCE-TIM
instrument. Gueymard (2012) used the older data time series from G04
and scaled it to the newer observations from the PREMOS instrument
onboard the PICARD satellite, obtaining an SC value of 1361.2 W/m2.
In parallel, Fehlmann et al. (2012) indicated that PREMOS was the best
understood and calibrated radiometer of its category ever sent to space.
In 2015, the International Astronomical Union (IAU) adopted a re-
solution suggesting a solar constant value of 1361 W/m2 (IAU, 2015),
based on available measurements during solar cycle 23 only. This value
was confirmed by Prša et al. (2016), but may be understood as provi-
sional, since not based on an in-depth study.
The new developments in precision space radiometry that followed
the publication of G04 show that the newer radiometer design of TIM
removed sources of bias that were impacting the older ones—based on
which the 1366.1 W/m2 value was derived. Even with older radiometer
designs, such as that of PREMOS, advanced pre-flight calibration and
characterization techniques allowed a substantial decrease in un-
certainty. In an effort to improve accuracy in solar or other applica-
tions, these developments justify a revision of the standard SC values
that were stipulated in ASTM (2000) or ISO (2007, 2013), for instance.
Moreover, the solar energy literature still frequently refers to even older
(and obsolete) SC determinations, such as 1367 or 1373 W/m2. This
contribution aims at alerting the scientific community involved in ter-
restrial or space applications about the recent evolution in the ob-
servation of TSI, justifying a lower value for SC, and proposing a long-
term reconstruction of the TSI time series using the best possible ab-
solute calibration.
2. TSI observations
Following many decades of attempts at estimating SC from terres-
trial observatories (e.g., by Abbot) and high-altitude balloons, aircraft
or rockets (Thekaekara, 1965, 1973, 1976), a new era started in No-
vember 1978 with the launch of Nimbus-7, which allowed the con-
tinuous monitoring of TSI with spaceborne radiometers without any
interference from atmospheric constituents. During its first nine years
of measurement, the mean annual TSI varied between 1370.2 and
1371.3 W/m2 in good synchronicity with solar activity, which estab-
lished the reality of this connection (Fox, 2004; Hickey et al., 1988;
Willson et al., 1981). In turn, the Earth climate impacts caused by this
variability started to be studied extensively (Fröhlich and Lean, 1998;
Lean, 2010; NRC, 2012; Schatten and Arking, 1990; Solanki et al.,
2013). Spaceborne measurements made from the satellites that
2 the time series so that it starts at the onset of solar cycle 21.
The abbreviation to be used for the astronomical unit is somewhat confusing: ISO
stipulates “ua” whereas the International Astronomical Union (IAU) favors “au”, and
ASTM reports it as “AU”. The ISO nomenclature is used here.
2
Solar Energy xxx (xxxx) xxx–xxx
followed Nimbus-7 indicated significantly lower TSI values, even under
“calm sun” conditions. This was caused mainly by differing absolute
calibrations of the radiometers, which prompted the need to reconcile
the time series from different platforms, using elaborate corrections and
time-dependent scaling factors. This arduous process, which is docu-
mented in a number of publications (e.g., Dewitte et al., 2004; Fröhlich,
2004, 2006, 2012a,b; Mekaoui and Dewitte, 2008; Willson, 2014;
Willson and Mordvinov, 2003), led to the development of composite
time series based on combined TSI observations from different instru-
ments, depending on period and estimated data quality.
In what follows, the PMOD and ACRIM composites are used ex-
tensively. (An earlier version of the PMOD composite was used in G04.)
The latest versions of the extended PMOD composite,3 covering the
period to 1976-01-11 to 2017-09-20,4 and of the ACRIM composite,5
covering the period 1978-11-17 to 2013-09-17, are used here. These
two composites differ because they use different observations from
different platforms during some periods, and differing reduction pro-
cesses. Moreover, the latest versions used here include appropriate
corrections of about −5 W/m2 compared to their older versions, in
order to compensate for the systematic overestimation of the mea-
surements made with older radiometers, as a result of two sources of
error: (i) stray light impacting the radiometer’s reading (Kopp and Lean,
2011); and (ii) difference of ≈0.3% between the World Radiometric
Reference (WRR) used to calibrate older radiometers and the absolute
calibration method based on the SI realization of the watt (with cryo-
genic radiometers) used for the TIM and PREMOS instruments
(Fehlmann et al., 2012; Kopp et al., 2012; Walter et al., 2017). In the
case of the PMOD composite, the correction process included a down-
ward correction of –0.362%, as explained in its accompanying online
documentation.6 To complement these composites, three recent, single-
source time series of measurements made during recent, or still on-
going, missions are added for reference. These are SORCE-TIM (2003-
02-25 to 2017-12-31), TCTE-TIM (2013-12-16 to 2017-12-31) and PI-
CARD-PREMOS (2010-07-27 to 2014-02-11). (In the present nomen-
clature, the satellite name comes first, followed by the instrument ac-
ronym.)
A comparison of the latest versions of the PMOD and ACRIM com-
posites is shown in Fig. 1. There are obvious differences between the
two time series, most particularly during 1978–79, when ACRIM’s TSI
appears systematically larger than PMOD’s TSI (or vice versa, PMOD
appears lower than ACRIM). Another source of concern is their dis-
crepancy during the solar minima between cycles 22 and 23 and cycles
23 and 24, as underlined before (Kopp, 2014). More generally, the
differences between the two datasets have been the object of intense
debate in the literature (Fröhlich, 2006; Scafetta and Willson, 2014;
Willson and Mordvinov, 2003), due to differing views on the radio-
meters’ calibration, stability, degradation, and need for correction. The
disagreement also fueled controversy regarding the reality of long-term
trends in the solar output and their possible impact on climate change
(NRC, 2012; Scafetta, 2013). One of the objectives of the present study
is to reconsider these differences from another standpoint and find a
way to reconcile the datasets.
Similar to Fig. 1, Fig. 2 shows the temporal evolution of TSI as
observed by SORCE-TIM, TCTE-TIM and PICARD-PREMOS. The time
series are shorter here, particularly for PREMOS and TCTE-TIM. The
differences between them are also typically smaller than in Fig. 1. A
comparison between the two composites (PMOD and ACRIM) and
SORCE-TIM is shown in Fig. 3 during their common period
3
ftp://ftp.pmodwrc.ch/pub/data/irradiance/composite/ext_composite_42_64_1508.
dat.
4
According to the data documentation, the extended version used here contains ad-
ditional data obtained with a proxy model from 1976-01-11 until 1978-11-16 to prolong
5
http://www.acrim.com/RESULTS/data/composite/acrim_composite_131130_hdr.txt.
6
ftp://ftp.pmodwrc.ch/pub/Claus/VIRGO-TSI/VIRGO_Char2Space.pdf.
C.A. Gueymard Solar Energy xxx (xxxx) xxx–xxx

Fig. 1. PMOD and ACRIM composite TSI time series during 1976–2017. Numbers in yellow circles indicate the solar cycle. The horizontal line at 1361.1 W/m2
indicates the revised solar constant. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 2. TSI observations from SORCE-TIM, TCTE-TIM and PICARD-PREMOS during 2003–2017.

Fig. 3. TSI time series from SORCE-TIM and the PMOD and ACRIM composites during 2003–2013.

(2003–2013), which coincided with the solar minimum between cycles
23 and 24. A difference of ≈0.4 W/m2 exists between the ACRIM time
series and those from PMOD or SORCE-TIM, whereas the latter two are
in close agreement (within 0.1 W/m2 most of the time).
3. Solar activity effects on TSI
The essential reason for TSI’s temporal variability is the solar activity
(Fröhlich and Lean, 2004). During periods of high activity, two different main
processes are exacerbated and interact with TSI: sunspots and faculae.
Whereas the former have a darkening effect that lowers TSI, the latter have a
brightening effect that increases it. (Other minor TSI-enhancing effects are
plages and flares.) Because the faculae cover the whole solar sphere, whereas
sunspots have only limited areal coverage, the net effect is a general increase
in TSI during high-activity periods, with occasional low points during the
passage of a large sunspot, as can be seen in Fig. 1. For more details, the
reader is referred to Chapman (1987), Fröhlich and Lean (2004), Godoli and
Allen (1964), Shapiro et al. (2016) and Solanki et al. (2013). There are
various ways to estimate the sun activity’s intensity, but the older proxy is the
sunspot number, RZ. Compilations of RZ at a daily resolution are available
since 1818 from the Sunspot Index Data Center (SIDC). (Monthly values are
also available since 1749.) Starting in 2015, a corrected methodology has
been used by SIDC to derive RZ (Clette and Lefèvre, 2016; Clette et al., 2016;
2014), whose “version 2” time series is now available.7 As suggested before
(e.g., G04, Solanki and Fligge, 1999), it is desirable to study the possible
correlation between TSI and RZ. This is done here with two different time
series. The first one combines PMOD and SORCE-TIM, whereas the second
one combines ACRIM and SORCE-TIM. The reliance on SORCE-TIM from the
start of its observational record (2003-02-25) until 2017-12-31 is justified by
the assessment of its stability, which is significantly better than that of older
instruments (NRC, 2012). For the two time series, the period before 1981 is
not included because of the noticeable discrepancy between ACRIM and
PMOD discussed above. A plot showing the variation of TSI vs. RZ appears in
Fig. 4, using the PMOD/SORCE-TIM combined time series. A quadratic
function appears to exist between TSI from RZ, thus confirming earlier results
(G04; Solanki and Fligge, 1999):
TSI = a0 + a1 RZ + a2 RZ2 (1)
Coefficients ai are given in Table 1 separately for the two time
series, along with the correlation coefficient and root mean square error
of the estimate (rmse). The latter is lower for the PMOD/SORCE time
series than for ACRIM/SORCE.
Two other usual proxies for solar activity (De Toma et al., 2004) are
the radio flux at 10.7 cm (hereafter, F) and the MgII doublet (at 280 nm)
index (hereafter, M). F is observed from terrestrial observatories and is
available daily since 1947 from the Canadian Solar Monitoring Pro-
gram.8 M is measured from space and is available only since 1978-11-
07. There are various sources of data for M, but most of them are in-
complete. The only two sources that could be found with a complete
time series between 1978 and end of 2017 are those provided by Space

7 8
http://www.sidc.be/silso/newdataset. http://www.spaceweather.gc.ca.

3
C.A. Gueymard Solar Energy xxx (xxxx) xxx–xxx

Fig. 4. Daily sunspot number vs. TSI from the PMOD composite combined with
SORCE-TIM observations during 1981–2017.

Table 1
Coefficients of Eqs. (1)–(3), correlation coefficient R, and root mean square
error of the estimate (rmse, in W/m2) for the ACRIM/SORCE-TIM and PMOD/
SORCE-TIM combined TSI time series since 1981.
Coefficient ACRIM/SORCE-TIM PMOD/SORCE-TIM

a0 1360.7073 1360.596
a1 0.0104882 0.008748
a2 −0.00002773 −0.000025
R 0.5335 0.5072
rmse 0.5426 0.4517

b0 1359.1022 1359.0877
b1 −0.001321 −0.002116
b2 0.0287896 0.0269138
b3 −0.00007135 −0.00006641
R 0.5775 0.5749 Fig. 5. Daily radio flux at 10.7 cm (top) and MgII index (bottom) vs. TSI from
rmse 0.5237 0.4289 the PMOD composite combined with SORCE-TIM observations during
1981–2017.
c0 1274.5352 1281.9099
c1 −0.0029 −0.003747
c2 −0.010111 −0.010316
c3 932.89608 839.68749
c4 −2366.911 −2077.815
R 0.7097 0.7692
rmse 0.4520 0.3350

Environment Technologies9 and the University of Bremen.10 Since the

latter uses observations from various satellites, which cover different
periods, their composite index is used here. Unfortunately, the scales of
these two sources of data are not compatible. The best correlations were
obtained with the Bremen version of M, which is thus used exclusively
in all what follows. The relationship between the PMOD/SORCE TSI
time series and the F and M proxies is shown in Fig. 5. As in Fig. 4, a
quadratic function can describe a large part of the variability in TSI.
Better correlations can be found by using a linear model of more
than one variable. Before November 1978, when the MgII index was not
available, the following equation provides the best fit: Fig. 6. Scatterplot comparing TSI from the ACRIM and PMOD composites to the
proxy model estimates during 1978–2017.
TSI = b0 + b1 RZ + b2 F + b3 F 2. (2)

Finally, an even better correlation is found when M is available:
TSI = c0 + c1 RZ + c2 F + c3 M + c4 M 2.
All coefficients for the proxy model developed here (Eqs. (1)–(3))
are provided in Table 1. It is clear that the best results are obtained with
Eq. (3) and the PMOD-SORCE time series, with substantially lower
9
http://www.spacewx.com/About_MgII.html.
10
http://www.iup.uni-bremen.de/UVSAT/Datasets/mgii.
noise than with ACRIM-SORCE. The TSI estimates from Eq. (3) are
compared to the measurements from the PMOD and ACRIM composites
(3) in Fig. 6, now including all their data since 1978. The lower noise of
PMOD and apparent overestimation of ACRIM confirm what could be
inferred from Fig. 1. The equations above—Eq. (3) since 11/1978 and
Eq. (2) before—can be used to fill in missing data in the original ob-
servational time series. The lower accuracy of Eq. (2) has a limited
impact in what follows because it only applies to the (short) period that
precedes TSI observations.

4
C.A. Gueymard
Fig. 7. TSI time series from the proxy model and ACRIM composite during 1976–2017. The horizontal line at 1361.1 W/m2 indicates the revised solar constant.
4. Reconstructed TSI time series and solar constant
The goal of this section is to develop a complete time series covering
the most part of four solar cycles (21 to 24) between 1976 and 2017,
and based on the best possible TSI observations. (Solar cycle 24 started
in December 2008; as of this writing it has waned substantially, but
might still be ≈12 months away from end.) Fig. 1 above has shown that
there are periods when the two longest composites disagree sub-
stantially. Hence, a piecewise approach is devised here to select the best
periods in each time series, and correct them whenever necessary. The
proxy model developed above is used here for that purpose. Figs. 7–9
show a comparison between the times series from the proxy model and
those from the ACRIM composite, PMOD composite and SORCE-TIM,
respectively. In general, the proxy model cannot predict large sudden
excursions in TSI due to, e.g., sunspots that are larger or denser than
usual. (These circumstances explain some of the largest deviations in
Fig. 6.)
To gain insight, this type of plot must be examined at higher re-
solution. An example is shown in Fig. 10 for the period July 2012 to end
of 2013. Since PICARD-PREMOS was operational then, its observations
are shown too. Following an almost perfect agreement between the four
data sources during December 2012, slight differences (in the
0.1–0.3 W/m2 range) started to emerge after that. Starting in March
2013, SORCE-TIM’s reading became higher than that of PMOD or
PREMOS, whereas that of ACRIM became lower. It is possible that the
TIM instrument became affected by an issue with the battery, which
eventually failed in July 2013. The SORCE-TIM observations were in-
terrupted between then and March 2014. Similarly, the ACRIM com-
posite stopped in September 2013 after three months of sporadic ob-
servations that were typically higher than those of PMOD or PREMOS.
Another period of interest is 2015–2017, when TCTE-TIM was op-
erational (Fig. 11). On average, its output is found ≈0.34 W/m2, or
0.025%, larger than that of SORCE-TIM, apparently due to some ab-
solute calibration issue. Simultaneously, SORCE-TIM and PMOD agree
to within ≈0.2 W/m2.
Based on this kind of analysis over the full 42-year span
(1976–2017), and still assuming that the proxy model provides a good
trend reference, the whole database is decomposed into nine successive
periods. Each period corresponds to a specific way of using one or more
sources of data, also involving correction factors in two cases. Those
cases are (i) the period from January 1976 to 1978-11-16, during which
Fig. 8. TSI time series from the proxy model and PMOD composite during 1976–2017. The horizontal line at 1361.1 W/m2 indicates the revised solar constant.
5
Solar Energy xxx (xxxx) xxx–xxx
the PMOD extended composite relied on an old proxy model providing
TSI values apparently too low by 0.01516% on average; and (ii) the
period from 2014-03-01 to 2017-12-31 during which TCTE-TIM is used
along with SORCE-TIM but requires a downward correction by 0.025%,
as mentioned above. When two or more sources of data appear equally
reliable during a single period, their average is used. This kind of re-
dundancy is useful because there are always a few data breaks in any
observational time series. It is also assumed that, when observations are
close and within their mutual uncertainty limits, they are essentially of
equivalent quality and accuracy. A description of the construction of
the nine periods appears in Table 2. The PREMOS dataset is in-
tentionally not used for this reconstruction, but is kept for a final ca-
libration, considering its recognized status as the radiometer with the
lowest in-flight uncertainty—0.027% according to Kopp and Lean
(2011). Comparing the average TSI from the proxy model to that of
PREMOS over the period from 2011-01-01 to 2013-09-22, a final
scaling factor of 1.000036735 is obtained, thus slightly increasing the
whole TSI reconstruction described in Table 2 by ≈0.05 W/m2. A time
series of the difference between the reconstructed TSI and each of the
sources of data evaluated here appears in Fig. 12. The daily differences
are normally within 0.5 W/m2, except with the ACRIM composite
during some periods.
The daily TSI values obtained over 1976–2017 with the re-
construction just described are available in the Supplementary
Material. That tabulation indicates the specific period (hence the im-
plied reconstruction method per Table 2), and whether each daily value
is derived from direct observation or from a proxy model. The proxy
estimates account for only 1483 days (or 9.7%) of the 15341-day total
period. The time series of daily TSI values is also shown in Fig. 13,
along with a 27-day moving average to smooth the rapid variations of
TSI during a rotation cycle. This figure can be compared to Fig. 1 in
G04, for instance.
The reconstructed TSI time series just described averages to a mean
value of 1361.1 W/m2 over the 3-cycle period from 1976 to 2008, with
a standard deviation of 0.52 W/m2. This mean value is conserved when
rather extending the period to the whole 42-year reconstruction,
1976–2017. The extreme minimum and maximum TSI values occurred
on 2003-10-29 (1357.1 W/m2) and 1989-05-12 (1363.3 W/m2), re-
spectively. The 42-year frequency distribution of TSI values, in bins of
0.1 W/m2, appears in Fig. 14. It is clearly skewed toward values
somewhat lower than the average, due to the impact of sunspots.
C.A. Gueymard Solar Energy xxx (xxxx) xxx–xxx

Fig. 9. TSI time series from the proxy model and SORCE-TIM observations during 1976–2017.

Fig. 10. Daily difference between various TSI data sources and the proxy model during 2012–2013.

Fig. 11. Daily difference between various TSI data sources and the proxy model during 2015–2017.

Table 2 cycle are summarized in Table 3. The onset of each cycle is defined in
Reconstruction periods during 1976–2017 and their sources of data. the literature only loosely, at a monthly resolution. Here, the onset’s
Period # Start date End date Source(s) of data
day of that month is defined as the day following the time of the lowest
M value during a period of sunspot absence. For Cycle 21, the lowest F
1 1976-01-01 1978-11-16 PMOD * 1.0001516 value was used instead, since the MgII index was not measured yet.
2 1978-11-17 1981-02-28 PMOD Table 3 shows that Cycle 23 was longer than the previous ones, and
3 1981-03-01 1989-09-30 PMOD, ACRIM
4 1989-10-01 2003-03-06 PMOD
recorded the lowest minimum TSI of the whole period. In parallel, it is
5 2003-03-07 2012-02-29 PMOD, SORCE noteworthy that the current Cycle 24 tends to experience much less
6 2012-03-01 2013-01-31 PMOD, ACRIM, SORCE daily variability (lower standard deviation) than the three previous
7 2013-02-01 2013-06-30 PMOD, ACRIM cycles.
8 2013-07-01 2014-02-28 PMOD
The standard uncertainty of the TSI values proposed here is esti-
9 2014-03-01 2017-12-31 SORCE, TCTE * 0.99975
mated at 0.5 W/m2 (coverage factor k = 1, for a 68% confidence in-
terval), which corresponds to an expanded uncertainty of 1.0 W/m2
Despite this, the median (1361.0 W/m2) is only marginally lower than (coverage factor k = 2, for a 95% confidence interval), based on the
the mean. latest evaluations from experts (Pers. comm. with Drs. Greg Kopp and
Since each solar cycle is different, statistics pertaining to each solar Werner Schmutz, 2016). The present solar constant determination is

Fig. 12. Daily difference between various TSI data sources and the reconstructed time series during 1976–2017.

6
C.A. Gueymard Solar Energy xxx (xxxx) xxx–xxx

with a temporal resolution of one day or better. This monitoring has

Fig. 13. Daily values of the reconstructed TSI time series during 1976–2017.
The red curve is the corresponding 27-day moving average, and the horizontal
line indicates the revised solar constant (1361.1 W/m2).
provided confirmation that TSI was indeed impacted by solar activity.
Nevertheless, the different radiometers that have made this monitoring
possible from successive satellite platforms revealed disagreements in
their results, which prompted the creation of TSI composites. Two of
them, the PMOD and ACRIM composites, are used here, along with
more recent sources of data, to reconstruct a continuous TSI time series
over a 42-year period that encompasses the most part of four solar
cycles (21 to 24, from 1976 to 2017).
To evaluate TSI during days of missing observations (which account
for 9.7% of the 42-year period), a proxy model has been obtained, by
which TSI is estimated from the daily sunspot number (using the new
version of its time series), the radio flux at 10.7 cm, and the MgII index.
This proxy model has also been used to select the best source(s) of TSI
data during different periods. In particular, it was found that the PMOD
composite performed better than the ACRIM composite before March
1981 and between the end of 1989 and early 2012. After 2003, the TSI
reconstruction also uses data from radiometers based on a newer de-
sign, SORCE-TIM and TCTE-TIM. A slight scale correction is finally
applied to benefit from the short-lived, but high-accuracy measure-
ments from PICARD-PREMOS. The daily TSI time series thus re-
constructed is available in the Supplementary Material.
The solar constant, obtained as the average of TSI over solar cycles
21–23, is determined as 1361.1 W/m2. It is ≈5 W/m2 lower than pre-
vious accepted values dating back from the early 2000s, and it has now
an estimated standard uncertainty of 0.5 W/m2. In conclusion, it is
strongly recommended that the older determinations promulgated by
standard bodies like ASTM or ISO be revised downward to reflect the
most recent developments described here.

Acknowledgments

Fig. 14. Relative frequency of TSI during 1976–2017, established using bins of
0.1 W/m2. The vertical line indicates the revised solar constant (1361.1 W/m2).
Table 3
TSI statistics for each solar cycle since 1976. SD: Standard deviation.
The author thanks Drs. André Fehlmann and Werner Schmutz for
the PREMOS data and helpful discussions. The dedicated work of the
personnel from these institutions is also gratefully acknowledged: The
Physikalisch-Meteorologisches Observatorium Davos team for the
PMOD composite data; the Active Cavity Radiometer Irradiance
Monitor team for the ACRIM composite data; the SORCE and TCTE
teams of the Laboratory for Atmospheric and Space Physics for the TIM
data; the WDC-SILSO, Royal Observatory of Belgium, Brussels, for the
sunspot data; the Solar Monitoring Program of Space Weather Canada
for the radio flux data; and the University of Bremen for the MgII data.
Appendix A. Supplementary material

Cycle Onset date Duration Mean TSI ± SD Maximum TSI Minimum TSI
Supplementary data associated with this article can be found, in the
(days)
online version, at http://dx.doi.org/10.1016/j.solener.2018.04.001.
21 1976-03-01 3855 1361.11 ± 0.55 1362.96 1358.12
22 1986-09-20 3671 1361.04 ± 0.55 1363.33 1358.86 References
23 1996-10-08 4459 1361.08 ± 0.55 1362.74 1357.10
24 2008-12-23 — 1361.07 ± 0.42a 1362.35 1358.57a
Abbot, C.G., 1911. The solar constant of radiation. Proc. Am. Philos. Soc. 50, 235–245.
a ASTM, 1974. Standard specifications for solar constant and air mass zero solar spectral
Estimated. irradiance. Standard E490-73a, ASTM International.
ASTM, 2000. Solar Constant and Zero Air Mass Solar Spectral Irradiance Tables. Standard
E490-00, ASTM International.
exactly 5.0 W/m2 (or 0.366%) lower than the previous value of Chapman, G.A., 1987. Solar variability due to sunspots and faculae. J. Geophys. Res. 92D,
809–812.
1366.1 W/m2 derived in G04 and ASTM (2000), which must both be Clette, F., Lefèvre, L., 2016. The New Sunspot Number: assembling all corrections. Solar
considered obsolete. In contrast, this determination is halfway between Phys. 291, 2629–2651. http://dx.doi.org/10.1007/s11207-016-1014-y.
the values proposed by IAU (2015) and Gueymard (2012), and within Clette, F., Lefèvre, L., Cagnotti, M., Cortesi, S., Bulling, A., 2016. The Revised Brussels-
Locarno Sunspot Number (1981–2015). Solar Phys. 291, 2733–2761. http://dx.doi.
only 0.1 W/m2 of them.
org/10.1007/s11207-016-0875-4.
Clette, F., Svalgaard, L., Vaquero, J.M., Cliver, E.W., 2014. Revisiting the Sunspot
5. Conclusion Number. A 400-Year Perspective on the Solar Cycle. Space Sci. Rev. 186, 35–103.
http://dx.doi.org/10.1007/s11214-014-0074-2.
De Toma, G., White, O.R., Chapman, G.A., Walton, S.R., Preminger, D.G., Cookson, A.M.,
The last 200 years have seen relentless attempts at determining the 2004. Solar cycle 23: an anomalous cycle? Astrophys. J. 609, 1140–1152.
solar constant and investigating its relationship with solar activity. Dewitte, S., Crommelynck, D., Mekaoui, S., Joukoff, A., 2004. Measurement and un-
Since 1978, spaceborne observations of the total solar irradiance (TSI) certainty of the long term total solar irradiance trend. Solar Phys. 224, 209–216.
Fehlmann, A., Kopp, G., Schmutz, W., Winkler, R., Finsterle, W., Fox, N., 2012. Fourth
have started to provide a nearly continuous record of the Sun’s output,

7
C.A. Gueymard
World Radiometric Reference to SI radiometric scale comparison and implications for
on-orbit measurements of the total solar irradiance. Metrologia 49, S34–S38. http://
dx.doi.org/10.1088/0026-1394/49/2/S34.
Fox, P., 2004. Solar Activity and Irradiance Variations. In: Pap, J.M., Fox, P. (Eds.), Solar
Variability and Its Effects on Climate. American Geophysical Union, Washington, DC.
Fröhlich, C., 2004. Solar irradiance variability. In: Pap, J.M., Fox, P. (Eds.), Solar
Variability and Its Effects on Climate. American Geophysical Union, Washington, DC.
Fröhlich, C., 2006. Solar irradiance variability since 1978. Space Sci. Rev. 125, 53–65.
http://dx.doi.org/10.1007/s11214-006-9046-5.
Fröhlich, C., 2012a. Solar constant and total solar irradiance variations. In: Meyers, R.A.
(Ed.), Encyclopedia of Sustainability Science and Technology. Springer, pp.
9469–9486.
Fröhlich, C., 2012b. Total solar irradiance observations. Surv. Geophys. 33, 453–473.
http://dx.doi.org/10.1007/s10712-011-9168-5.
Fröhlich, C., Lean, J., 1998. The Sun's total irradiance: cycles, trends and related climate
change uncertainties since 1976. Geophys. Res. Lett. 25, 4377–4380.
Fröhlich, C., Lean, J., 2004. Solar radiative output and its variability: evidence and me-
chanisms. Astron. Astrophys. Rev. 12, 273–320. http://dx.doi.org/10.1007/s00159-
004-0024-1.
Godoli, G., Allen, C.W., 1964. The indices of solar activity. Planet. Space Sci. 12,
349–354.
Gueymard, C.A., 2004. The sun’s total and spectral irradiance for solar energy applica-
tions and solar radiation models. Sol. Energy 76, 423–452.
Gueymard, C.A., 2006. Reference solar spectra: Their evolution, standardization issues,
and comparison to recent measurements. Adv. Space Res. 37, 323–340.
Gueymard, C.A., 2012. Solar radiation—Introduction. In: Meyers, R.A. (Ed.),
Encyclopedia of Sustainability Science and Technology. Springer.
Hickey, J.R., Alton, B.M., Kyle, H.L., Hoyt, D., 1988. Total solar irradiance measurements
by ERB/Nimbus-7, a review of nine years. Space Sci. Rev. 48, 321–342.
Hoyt, D.V., 1979. The Smithsonian Astrophysical Observatory solar constant program.
Rev. Geophys. Space Phys. 17, 427–458.
Hoyt, D.V., Schatten, K.H., 1997. The Role of the Sun in Climate Change. Oxford Univ.
Press, New York.
IAU, 2015. Resolution B3 on Recommended Nominal Conversion Constants for Selected
Solar and Planetary Properties. International Astronomical Union, Paris, France.
ISO, 2007. Space environment (natural and artificial) — Process for determining solar
irradiances. ISO 21348 standard, International Organization for Standardization.
ISO, 2013. Space environment (natural and artificial) — Earth upper atmosphere. ISO
14222 standard, International Organization for Standardization.
Kopp, G., 2014. An assessment of the solar irradiance record for climate studies. J. Space
Weather Space Clim. 4, A14. http://dx.doi.org/10.1051/swsc/2014012.
Kopp, G., Fehlmann, A., Finsterle, W., Harber, D., Heuerman, K., Willson, R.C., 2012.
Total solar irradiance data record accuracy and consistency improvements.
Metrologia 49, S29.
Kopp, G., Lean, J.L., 2011. A new, lower value of total solar irradiance: Evidence and
climate significance. Geophys. Res. Lett. 38. http://dx.doi.org/10.1029/
Solar Energy xxx (xxxx) xxx–xxx
2010GL045777.
Kuhn, J.R., Armstrong, J.D., 2004. Mechanisms of solar irradiance variations. In: Pap,
J.M., Fox, P. (Eds.), Solar Variability and Its Effects on Climate. American
Geophysical Union, Washington, DC.
Lean, J.L., 2010. Cycles and trends in solar irradiance and climate. WIREs Clim. Change 1,
111–122.
Mekaoui, S., Dewitte, S., 2008. Total solar irradiance measurement and modelling during
Cycle 23. Sol. Phys. 247, 203–216.
NRC, 2012. The Effects of Solar Variability on Earth’s Climate, a Workshop Report. The
National Academies Press, Washington, DC.
Prša, A., Harmanec, P., Torres, G., Mamajek, E., Asplund, M., Capitaine, N., Christensen-
Dalsgaard, J., Depagne, E., Haberreiter, M., Hekker, S., Hilton, J., Kopp, G., Kostov,
V., Kurtz, D.W., Laskar, J., Mason, B.D., Milone, E.F., Montgomery, M., Richards, M.,
Schmutz, W., Schou, J., Stewart, S.G., 2016. Nominal values for selected solar and
planetary quantities: IAU 2015 resolution B3. Astronomical J. 152, 41.
Scafetta, N., 2013. Solar and planetary oscillation control on climate change: hind-cast,
forecast and a comparison with the CMIP5 GCMs. Energy Environ. 24, 455–496.
http://dx.doi.org/10.1260/0958-305X.24.3-4.455.
Scafetta, N., Willson, R.C., 2014. ACRIM total solar irradiance satellite composite vali-
dation versus TSI proxy models. Astrophys. Space Sci. 350, 421–442.
Schatten, K.H., Arking, A., 1990. Climate impact of solar variability, Conf. Publ. 3086.
NASA, Greenbelt, MD.
Shapiro, A.I., Solanki, S.K., Krivova, N.A., Yeo, K.L., Schmutz, W.K., 2016. Are solar
brightness variations faculae- or spot-dominated? Astron. Astrophys. 589, A46.
http://dx.doi.org/10.1051/0004-6361/201527527.
Solanki, S.K., Fligge, M., 1999. A reconstruction of total solar irradiance since 1700.
Geophys. Res. Lett. 26, 2465–2468.
Solanki, S.K., Krivova, N.A., Haigh, J.D., 2013. Solar irradiance variability and climate.
Annu. Rev. Astro. Astrophys. 51, 311–351.
Thekaekara, M.P., 1965. The solar constant and spectral distribution of solar radiant flux.
Sol. Energy 9, 7–20.
Thekaekara, M.P., 1973. Solar energy outside the Earth's atmosphere. Sol. Energy 14,
109–127.
Thekaekara, M.P., 1976. Solar radiation measurement: techniques and instrumentation.
Sol. Energy 18, 309–325.
Walter, B., Winkler, R., Graber, F., Finsterle, W., Fox, N., Li, V., Schmutz, W., 2017. Direct
solar irradiance measurements with a cryogenic solar absolute radiometer. AIP Conf.
Proc. 1810, 080007. http://dx.doi.org/10.1063/1.4975538.
Willson, R.C., 2014. ACRIM3 and the Total Solar Irradiance database. Astrophys. Space
Sci. 352, 341–352.
Willson, R.C., Gulkis, S., Janssen, M., Hudson, H.S., Chapman, G.A., 1981. Observations
of solar irradiance variability. Science 211, 700–702.
Willson, R.C., Mordvinov, A.V., 2003. Secular total solar irradiance trend during solar
cycles 21–23. Geophys. Res. Lett. 30, 1199. http://dx.doi.org/10.1029/
2002GL016038.