1366 And All That – The Secret History of Total Solar Irradiance

Introduction

The concept of a “solar constant” is firmly embedded [and frequently repeated] in the mainstream media:

The solar constant includes all types of solar radiation, not just the visible light.
http://en.wikipedia.org/wiki/Solar_constant

Direct overhead sunlight at the top of the atmosphere provides 1366 W/m2
http://en.wikipedia.org/wiki/File:Greenhouse_Effect.svg

Since 1979 the “solar constant” has been measured by satellite and the resulting Total Solar Irradiance [TSI] dataset is firmly embedded [and frequently reproduced] in the mainstream media:

Satellite observations of Total Solar Irradiance from 1979–2006.
http://en.wikipedia.org/wiki/Global_warming

The Wikipedia graphic very successfully communicates these concepts.

The imagery conveys scientific certainty and continuity.
The detail projects a history of precise measurement.
The caption provides associations with space technology and scientific achievement.

However, being curious, I wanted to look behind this superficial imagery so I could understand the science and fully appreciate this space age achievement.

Jolly Hockey Sticks

Before the curtain rises lets perform a quick reality check.

When did the official TSI satellite age start?

The Wikipedia caption says:

Satellite observations of Total Solar Irradiance from 1979

While the graph displays solar irradiance starting in 1976.

Strange that.

Feels like familiar territory.

Looks like another “hockey stick” trick.

Has someone seamlessly stitched a TSI reconstruction onto the satellite data?

Let’s go back in time.

Let’s look at the beginning of this story.

Before the Satellite Age

Before the satellite age TSI calculations were based upon balloon and sounding rocket measurements.

Sounding rocket
A sounding rocket, sometimes called a research rocket, is an instrument-carrying rocket designed to take measurements and perform scientific experiments during its sub-orbital flight.

The origin of the term comes from nautical vocabulary, where to sound is to throw a weighted line from a ship into the water to gauge the water’s depth. Sounding in the rocket context is equivalent to taking a measurement.

The rockets are used to carry instruments from 50 to 1,500 kilometres (31 to 930 mi) above the surface of the Earth, the altitude generally between weather balloons and satellites (the maximum altitude for balloons is about 40 kilometres (25 mi) and the minimum for satellites is approximately 120 kilometres (75 mi)).

Advantages
Sounding rockets are advantageous for some research due to their low cost, short lead time (sometimes less than six months) and their ability to conduct research in areas inaccessible to either balloons or satellites.

They are also used as test beds for equipment that will be used in more expensive and risky orbital spaceflight missions.

The smaller size of a sounding rocket also makes launching from temporary sites possible allowing for field studies at remote locations, even in the middle of the ocean, if fired from a ship.

http://en.wikipedia.org/wiki/Sounding_rocket

The Great Leap Forward

The official TSI satellite record starts in 1979 and [luckily] a paper by Richard C. Willson [published in 1981] precisely quantifies the scientific impact of the TSI calculations based upon ACRIM satellite observations:

1969 – 1,366 W/m2 – Balloon experiment

1976 – 1,368.1 W/m2 – Sounding rocket experiment

1978 – 1,367.6 W/m2 – Sounding rocket experiment

1980 – 1,367.8 W/m2 – Sounding rocket experiment

1980 – 1,367.7 W/m2 – ACRIM Satellite

Solar total irradiance observations by Active Cavity Radiometers
Richard C. Willson
Solar Physics – November 1981, Volume 74, Issue 1, pp 217-229
http://link.springer.com/article/10.1007%2FBF00151292

This 1981 paper by Richard C. Willson provides several curious insights:

TSI is calculated. TSI is not “observations” as captioned by Wikipedia.

The TSI calculation made using the first five month average from the ACRIM satellite was virtually identical to the previous sounding rocket calculations.

The abstract clearly talks about monitoring the variability of solar total irradiance and provides upper limits on solar total irradiance variability:
1969-1980 +/- 0.2 %
1976-1980 +/- 0.1 %

The Wikipedia TSI graph does not accurately portray the 1980 ACRIM result.

Recalculating TSI

In 1988 Hoffert-Frei-Narayanan reviewed the first five years of ACRIM data.

Interestingly, the first five monthly averages [in 1980] have been revised upwards so that they no longer agree with the 1,367.7 W/m2 average reported by Willson in 1981.

Figure 2 shows monthly averages of the first five years of ACRIM irradiance data. In contrast to the short-term anticorrelation of irradiance with the area of sunspot groups crossing its surface, the monthly mean ACRIM data trends downward along with sunspots since the sunspot maximum in 1980.

The negative, deviant irradiance point at about April 1984 is probably insignificant – it occurred at the time of SMM repair and represents little real data.

Application of solar max ACRIM data to analysis of solar-driven climatic variability on Earth
Martin I. Hoffert, Allan Frei, Vijay K. Narayanan
Climatic Change – December 1988, Volume 13, Issue 3, pp 267-285
http://link.springer.com/article/10.1007%2FBF00139810?LI=true

This small example of a retrospective change highlights two very important points:

TSI is calculated.
Therefore, a retrospective change can be applied [to any part] of the historical TSI record [at any time] by simply using a revised TSI algorithm.
This does not change the raw data but it does change the [derived] TSI history.

Unfortunately, historical TSI documents may appear anomalous because the TSI history has been subsequently recalculated [and the TSI history rewritten].

Clearly, the early years of the satellite age didn’t provide any startling TSI revelations.

Therefore, let’s fast forward so we can review the achievements of the last 33 years.

The Official Satellite Data Series

There are seven satellites data series currently included in the “official” TSI dataset.

http://www.acrim.com

Plus a few others that have been excluded from the “official” TSI dataset.

Something doesn’t quite tally:

Has solar irradiance actually declined that much in the last 33 years?

Has the accuracy of the instrumentation simply improved in the last 33 years?

Don’t panic!

We have the technology.

We have a solution.

Let me introduce you to Data Compositing.

TSI Composites

Unsurprisingly, the harmonisation [stitching together] of the “official” data series has become a recognised technique [although “art form” is a more appropriate term].

There are several flavours of TSI composite: PMOD, ACRIM, IRMB.

Solar Radiative Output and its Variability: Evidence and Mechanisms
Claus Frohlich, Judith Lean
August 2004
ftp://ftp.pmodwrc.ch/pub/Claus/Publications/A%26ARev_12_273_2004.pdf

The most remarkable [but not surprising] aspect of these three TSI Composites is how they all [somehow] manage to converge upon the published “solar constant” of 1,366 W/m2 which is so effectively embedded within the original Wikipedia graphic.

Arguably, data compositing is the greatest achievement of the TSI satellite age.

Let’s try to understand this mess a little better. Let’s scrape back another layer.

Incompatible with the Scientific Method

Richard Willson, in a letter written to Nicola Scafetta in September 2008, noted:

To ‘adjust’ satellite data to agree with such models is incompatible with the scientific method.

The “Solar Constant”

In 1981 Richard Willson wrote about “the variability of solar total irradiance”.

Before 2012 the story changed and Wikipedia promotes the “solar constant” concept.

Evidently, somewhere a long the line, “settled science” decided the Sun is constant.

Obviously, this decision is not supported by the TSI dataset.

Therefore, supporting evidence was manufactured by Data Compositing.

However, “settled science” is only part of the story.

The real problems are far more fundamental.

The real problems originate in the very definition of the solar constant:

The solar constant, a measure of flux density, is the amount of incoming solar electromagnetic radiation per unit area that would be incident on a plane perpendicular to the rays, at a distance of one astronomical unit.
http://en.wikipedia.org/wiki/Sunlight

The first problem is that the Solar Constant has to be measured “at a distance of one astronomical unit” from the Sun.

An astronomical unit (abbreviated as AU, au, a.u., or ua) is a unit of length equal to exactly 149,597,870,700 metres (92,955,807.273 mi) or approximately the mean Earth–Sun distance.

http://en.wikipedia.org/wiki/Astronomical_unit

The second problem is that the Solar Constant is “a measure” of flux density.

The third problem is that the Solar Constant measurement has to include all “solar electromagnetic radiation”.

Upon first inspection those problems don’t appear that serious.

That’s was my initial reaction.

However, when I dug deeper they became fundamental problems.

Earth’s Orbit

Do you remember that first problem?

The Solar Constant has to be measured “at a distance of one astronomical unit”.

Let’s find a solution to that problem.

It’s not that complicated.

1 AU 149,597,871 km

The Earth has an elliptical orbit around the Sun.

Aphelion: 152,098,232 km
Perihelion: 147,098,290 km
Range: 4,999,942 km

Obviously a TSI satellite must orbit the Sun at a distance of 1 AU.

Unfortunately, no TSI satellite has ever been placed in a 1 AU orbit around the Sun.

Note:
The SOHO-VIRGO satellite orbits the Sun-Earth L1 Lagrangian point which is about 1.5 million km from the Earth.

In theory the SOHO/VIRGO data series should be an outlier data series because:
a) It is the only TSI mission that has measured TSI beyond the Geosphere.
b) It has measured the outlier Solar Cycle 23-24 quiet sun era.

Therefore, it is truly remarkable that the SOHO/VIRGO data series harmonises so very well with the Low Earth Orbit ACRIM II data series in the “official” dataset.

Low Earth Orbit Satellites

TSI satellites have been predominately placed into Low Earth Orbit:

Low Earth Orbit (LEO) – Geocentric orbits ranging in altitude from 160 kilometeres (100 statue miles) to 2,000 kilometres (1,200 mi) above mean sea level. At 160 km, one revolution takes approximately 90 minutes, and the circular orbital speed is 8,000 metres per second (26,000 ft/s).

http://en.wikipedia.org/wiki/Geocentric_orbit

Do you remember that second problem?

The Solar Constant is “a measure” of flux density.

Well that has now become a really big problem because the TSI satellite is not orbiting “at a distance of one astronomical unit”.

The TSI satellite cannot take “a measure” of flux density at “at a distance of one astronomical unit” because it’s stuck in Low Earth Orbit.

Don’t panic!

We have the technology

We have a solution.

Let me introduce you to Computer Modelling.

Will the model be accurate?

Don’t worry!

No one will be able to tell the difference.

After all, nobody is actually measuring TSI as 1 AU.

TSI Computer Modelling

In the early days the computer models were fairly straightforward.

The irradiance measured by ACRIM is corrected for the following effects (in order of significance):

(a) Normalization to 1 AU distance, based on linear interpolation from the ephemeris distance plus the projection of the satellite orbit on the radial direction to the Sun;

(b) correction for the slow decrease in channel A’s sensitivity between days 62 and 163;

(c) temperature-dependent corrections for radiation lost through the aperture and for the temperature coefficient of resistance of the cavity heating elements;

(d) correction for relativistic radiative effects due to the Sunsatellite relative velocity;

(e) correction for the cosine of the angle between ACRIM’s line-of-sight and the Sun’s center.

Solar total irradiance observations by Active Cavity Radiometers
Richard C. Willson
Solar Physics – November 1981, Volume 74, Issue 1, pp 217-229
http://link.springer.com/article/10.1007%2FBF00151292

However, as the years ticked by things got more complicated.

People asked questions.
Operational events occurred.
Solar cycles changed.
People made new discoveries.

Greg Kopp, lead scientist for the SOURCE-TIM satellite, describes some of the complexity embedded in the latest generation of TSI computer models:

We produce the 1-AU corrected TSI for people studying the Sun’s output; and we produce the ‘tsi_true_earth’ value for those, such as climate modelers, wanting direct radiative inputs to the Earth’s system.

Thus the ‘tsi_true_earth’ appropriately does not remove the effect of the lunar tug on the Earth, since that does affect the at-Earth radiative inputs.

In our orbital corrections, we use the JPL ephemeris VSOP87, which accounts for the positions of all the planets in the solar system as well as the Moon to make our Sun-Earth distance corrections to a fixed 1-AU; so you shouldn’t see any lunar signal in the ‘tsi_1au’, but you should (as you do) in the ‘tsi_true_earth’.

This ephemeris also includes effects such as that the Sun itself rotates around the center of mass of the solar system, which, thanks to Jupiter, is close to the Sun’s surface and has a ~12-year period.

We also correct for spacecraft effects, which include Sun-instrument distance changes due to the spacecraft’s low Earth orbit. These are comparable to the lunar effects (+/- 14000 km) and occur on 95-minute orbital time scales. And we apply Doppler corrections, as the instrument collects blue-shifted photons depending on its radial velocity toward the Sun, whether due to the spacecraft’s or the Earth’s orbital motions. These are ~50 ppm corrections over the spacecraft’s 95-minute orbital period.

http://tallbloke.wordpress.com/2012/12/02/tsi-sorce-and-a-signature/#comment-37123

Unsurprisingly, I have seen no mention of adjustments for:

Absorption and emissions in the upper atmosphere.
Absorption and emissions in the plasmasphere and magnetosphere.
Solar wind density between the satellite and 1 AU.

Perhaps someone will eventually conclude that “TSI at 1 AU” is just a crazy dream.

Perhaps someone will eventually realise they could learn a whole lot more by comparing the SOHO L1 raw data with the SORCE/TIM Low Earth Orbit raw data.

I am not holding my breath.

Mind the Gap

You do remember that third problem?

The Solar Constant measurement has to include all “solar electromagnetic radiation”.

Well, guess what.

Total spectrum coverage was first achieved in 2003 with SORCE/TIM.

Only 24 years late.

What happened before 2003?

Good question.

They might have simply ignored the problem.
They might have invented the missing data in their Computer Models.

I haven’t discovered the answer to that question yet.

Either way, it’s a big problem because the largest gaps were in the high energy ultraviolet spectrum [which climatologists are just beginning to realise is important].

Remarkable Stability

A remarkable aspect of the individual satellite data series [and the composite datasets] is that they display a remarkable level of stability when compared to the other solar metrics such as: solar wind, sunspots, magnetic polarity etc.

The general assumption is that the reported TSI simply reflects the “solar constant”.

Is this a safe assumption?

No.

Firstly:
TSI at 1 AU is generated by Computer Model.
Computer Models are notorious for reflecting the preconceived ideas and prejudices of their originators. Computer Models are not reality

Secondly:
If the high energy ultraviolet spectrum gaps, prior to 2003, were simply ignored then the TSI history underestimates solar volatility. If the high energy ultraviolet spectrum gaps, prior to 2003, were manufactured by the computer model then [most likely] the TSI history underestimates solar volatility.

Thirdly:
Paradoxically, the TSI data series have incrementally declined [over the last 33 years] whilst the coverage of the ultraviolet spectrum has incrementally expanded.

Failed Primary Objectives

The history of TSI satellite missions is a singular history of failed primary objectives.

In the words of Greg Kopp, lead scientist for the SOURCE-TIM satellite:

We produce the 1-AU corrected TSI for people studying the Sun’s output; and we produce the ‘tsi_true_earth’ value for those, such as climate modelers, wanting direct radiative inputs to the Earth’s system.

http://tallbloke.wordpress.com/2012/12/02/tsi-sorce-and-a-signature/#comment-37123

The astronomical objective has always been:

Consistently measure “TSI at 1 AU”.

Unfortunately, this objective has never been realised.

TSI satellites have never orbited the Sun at a constant 1 AU.
The “TSI at 1 AU” results have always been derived using computer models.

Equally valid modelling results could be been achieved using inputs from high altitude observatories, high altitude flights, balloons and sounding rockets.

The climate modelling objective has been:

Establish a consistent measure of TSI.

Unfortunately, this objective has never been achieved because no “orbital altitude” standard has ever been set for TSI satellites.

The majority of TSI satellites have been in a variety of Low Earth Orbits.
The one exception is the SOHO-VIRGO satellite which orbits the Sun-Earth L1 Lagrangian point [about 1.5 million km from the Earth].

Given the limitations of the climate models [and given the fact that they have been happily using the same TSI number for years] a TSI measurement taken by high altitude aircraft would have been more appropriate [as would a decent set of terrestrial observations].

But forget the past.

The New Normal

We live in “interesting times”.

The SORCE/TIM satellite has suspended daily performances:

Updates to the SORCE data record are currently not being produced for some instruments while the SORCE spacecraft recovers from a battery anomaly.

Updates will recommence when the instruments are acquiring science data again and data processing changes are implemented to accomodate more limited observing modes.

http://lasp.colorado.edu/sorce/data/tsi_data.htm

NASA has unilaterally declared an “official” reduction in the “solar constant” of 5 W/m2 [after looking at the SOURCE satellite data for nine years] because it “is critical in examining the energy budget”.

Total (TSI) and spectral solar irradiance (SSI) upon Earth
Total Solar Irradiance upon Earth (TSI) was earlier measured by satellite to be roughly 1.366 kilowatts per square meter (kW/m²), but most recently NASA cites TSI as “1361 W/m² as compared to ~1366 W/m² from earlier observations [Kopp et al., 2005]”, based on regular readings from NASA’s Solar Radiation and Climate Experiment(SORCE) satellite, active since 2003, noting that this “discovery is critical in examining the energy budget of the planet Earth and isolating the climate change due to human activities.”

http://en.wikipedia.org/wiki/Solar_radiation

SORCE/TIM [with a little help from Kopp & Lean] has determined the previously calculated TSI values are “erroneously high” because of “internal instrument scatter”.

The 4.5 W/m^2 by which the TIM reads lower than prior instruments has been resolved as being largely due to internal instrument scatter in those prior instruments causing erroneously high readings (see Kopp & Lean, GRL, 38, L01706, 2011).

http://lasp.colorado.edu/sorce/data/tsi_data.htm#historical

Uncorrected scattering and diffraction are shown to cause erroneously high readings in non-TIM instruments.

A new, lower value of total solar irradiance: Evidence and climate significance
Greg Kopp and Judith L. Lean
http://www.agu.org/pubs/crossref/2011/2010GL045777.shtml

This Time Its Different

SORCE/TIM is totally convinced they are calculating TSI correctly.
Long-term relative uncertainties are estimated to be less than 0.014 W/m2/yr.

Data Quality Description
On-orbit instrument characterization is an on-going effort, as the TIM team regularly tracks instrument degradation and calibrates the instrument servo system on-orbit, periodically updating the data processing system with new calibration values. Only minor corrections are anticipated at this phase in the SORCE/TIM mission.

To date the TIM is proving very stable with usage and solar exposure, and long-term relative uncertainties are estimated to be less than 0.014 W/m2/yr (10 ppm/yr).

Present absolute accuracy is estimated to be 0.48 W/m^2 (350 ppm), largely determined by the agreement between all four TIM radiometers.

http://lasp.colorado.edu/sorce/data/tsi_data.htm#historical

http://lasp.colorado.edu/sorce/total_solar_irradiance_plots/images/tim_level3_tsi_24hour_640x480.png

Surreal Science

One Ring to rule them all, One Ring to find them,
One Ring to bring them all and in the darkness bind them

J. R. R. Tolkien

After declaring the old “official” history of TSI calculations to be “erroneously high readings” the next step is to declare the old “official” TSI composites invalid because they “lack coherent temporal structure”.

In addition to the offsets, published irradiance observations composing the 32-year TSI database lack coherent temporal structure because of inconsistent trends that indicate the presence of uncorrected instrumental drift and are not explained by known sources of solar irradiance variability.

A new, lower value of total solar irradiance: Evidence and climate significance
Greg Kopp and Judith L. Lean
http://www.agu.org/pubs/crossref/2011/2010GL045777.shtml

Luckily, science has developed a magic “Irradiance Variability Model” than can bring them all together and bind the old TSI values [produced by various computer models] into one new reality [produced by one computer model].

Irradiance variations estimated from an empirical model that combines the two primary influences of facular brightening and sunspot darkening with their relative proportions determined via regression from direct observations made by SORCE/TIM.

A new, lower value of total solar irradiance: Evidence and climate significance
Greg Kopp and Judith L. Lean
http://www.agu.org/pubs/crossref/2011/2010GL045777.shtml

The New Past

A new four hundred year history of TSI has been conjured up by simply subtracting the “erroneously high” -4.8741 W/m^2 “from their models” of the past.

The values from their model have been offset -4.8741 W/m^2 to match the SORCE/TIM measurements during years of overlap and then extended or replaced using SORCE/TIM annual averages from 2003 onward.

http://lasp.colorado.edu/sorce/data/tsi_data.htm#historical

Presumably this new model TSI is designed to fit the new model Temperature Index.

Global mean land-ocean temperature change from 1880–2011, relative to the 1951–1980 mean.
The black line is the annual mean and the red line is the 5-year running mean.
The green bars show uncertainty estimates. Source: NASA GISS.
http://en.wikipedia.org/wiki/Global_warming

The New Climate Science

NASA has finally acknowledged that the “solar constant” might vary significantly in the ultraviolet spectrum [after looking at the SOURCE satellite data for nine years] and the associated climate response is probably more complicated than originally envisaged [by “settled science” and their climate models].

Furthermore the Spectral Irradiance Monitor (SIM) has found in the same period that spectral solar irradiance (SSI) at UV (ultraviolet) wavelength corresponds in a less clear, and probably more complicated fashion, with earth’s climate responses than earlier assumed, fueling broad avenues of new research in “the connection of the Sun and stratosphere, troposphere, biosphere, ocean, and Earth’s climate”

http://en.wikipedia.org/wiki/Solar_radiation

Long-Term Variations in UV and EUV Solar Spectral Irradiance
http://solar.physics.montana.edu/SVECSE2008/pdf/floyd_svecse.pdf

Presumably, we can now look forward to a new series of prognostications from a new generation of [re-calibrated and ultraviolet aware] “climate change” models.

The Review: 1366 And All That

Reviewing TSI satellite data is very subjective because [like any other satellite project] only the results have been published. The raw satellite data and necessary data processing tools have not been published. Therefore, the satellite results are not strictly scientific because the results cannot be independently verified.

A charitable review [assuming good faith and best endeavours] could only ever conclude [from the actual satellite data series] that:

The “solar constant” appears to be a constant that varies.

Computer modelling of the “solar variable” [based upon the measurements collected from various satellite instruments over the last 33 years] has yielded an array of modelled values for the “solar variable” [at 1 AU] that have incrementally declined from 1,368 W/m2 [in 1980] to 1,361 W/m2 [in 2012].

Unfortunately, it is impossible to quantitatively attribute specific causes to the perceived decline in the derived “solar variable”.

Some of the decline is probably attributable to natural solar variability.

However, it is suspected that the majority of the perceived decline in the “solar variable” is associated with:

Inconsistent satellite orbits
Changes in instrumentation and solar spectrum coverage
Variable data processing procedures
Evolving computer models
Morphing organisational roles, responsibilities and priorities
Changes in personnel
Technical and budgetary constraints

A less charitable review would conclude that the collection of raw solar irradiance data [via satellite] has been an elaborate [and very expensive] charade.

NASA seems [at the moment] to be adopting a more pragmatic approach because they really have no other option than to “run” with the current SORCE/TIM results.

The new star of the show SORCE/TIM is potentially good news.

Unfortunately, SORCE/TIM has temporarily suspended daily performances.

Perhaps this is the final curtain for a farce that has been running for 33 years.

Only time will tell.

Tim Cullen
Malaga
December 2012