Is It the Sun?

(c) 2009 by Barton Paul Levenson

Some folks who admit that global warming is real say it's all caused by the sun--not carbon dioxide. If it's the sun doing it, clearly there's no need to stop burning fossil fuels or cutting down forests, and certainly no need for cap-and-trade bills or carbon taxes. Thus the denier cry, "It's the sun, stupid!"

Is it the sun? Well, no. There are five good reasons why it can't be.


The Sun hasn’t gotten noticeably brighter in 50 years. We’ve been measuring it from satellites like Nimbus-6 and -7 and the Solar Maximum Mission. If solar output has been flat for the past 50 years, it’s hard to see how it could have caused the sharp upturn in warming of the last 30 years. Here are the TSI figures ("Total Solar Illumination" is the solar constant):

Lean's TSI reconstruction for the past 400 years

Svalgaard's TSI reconstruction for the past 300 years


Increased sunlight would heat the stratosphere first, where the ozone layer absorbs solar ultraviolet light. But the stratosphere is cooling.

The energy balance in the stratosphere is between UV heating of ozone and infrared emission from carbon dioxide. Ozone depeletion accounts for some of the cooling, but not enough. Most of the cooling is caused by increased carbon dioxide. In fact, the stratosphere cooling while the troposphere warms is the characteristic signal of global warming caused by increased carbon dioxide.


Increased sunlight would heat the equator more than the poles (Lambert’s cosine law). Instead we see “polar amplification,” another thing predicted by the climate modelers. For those who don't know Lambert's cosine law, it's:

I = Io cos θ

where Io is the radiation measured perpendicular to the surface in question and θ is the angle the surface is turned away from that radiation. For the Earth, θ is the latitude. So you can see that at latitude 0 (the equator), the solar constant hits full on, whereas at latitude 90 (the poles), there's no illumination at all. Clearly this is an ideal case and applies strictly to a planet with no axial tilt which only gets illumination from the sun, but on average it's still true for the Earth.

So why polar amplification? There are two reasons. One is "ice-albedo feedback"--as more bright polar ice melts, more dark land or seawater is revealed to the sun, and dark things absorb light better than bright things. Another is that colder air holds less water vapor. Water vapor is the dominant greenhouse gas in Earth's atmosphere. In cold regions with less water vapor in the atmosphere, carbon dioxide accounts for proportionally more of the greenhouse heating. So with carbon dioxide rising, you get faster global warming the closer you get to the poles.


Increased sunlight would increase daytime temperatures more than nighttime temperatures (duh!). But nighttime temperatures have risen more, which is consistent with an atmosphere of increased IR opacity holding in surface warmth better. Think of the way a cloudy night is warmer than a clear night.


Increased sunlight would increase summer temperatures more than winter temperatures (duh again!). But winter temperatures have risen more. Again, more greenhouse gases hold in the heat better when sunlight is decreased.

So in short, folks, not only is it not the sun, but it can't be.

Let's examine the problem one last way: A statistical analysis. Let's take the NASA GISS temperature anomalies for 1880 to 2007 and regress them on ln CO2 (the natural log of the carbon dioxide level, since radiative forcing from CO2 is proportionate to the log of the concentration) and TSI. We'll use Svalgaard's TSI, but anyone is welcome to use Lean's figures for 1880-2000 instead (see above for the link). The loss of seven points (2001-2007) doesn't much affect the outcome, and you get about the same results. Figures for temperature anomaly and CO2 can be found here:

Temperature and CO2

The regression line is:

Anom = -5749 + 324.2 ln CO2 + 2.842 TSI

With N = 128 observations and R2 = 76.5% of the variance accounted for. The t-statistics on the coefficients are -1.06 for the intercept (p < 0.289), 19.8 for carbon dioxide (p < 1.93 x 10-40), and 0.717 for TSI (p < 0.474). In other words, the carbon dioxide term is highly statistically significant, and the TSI term is not significant at all--no better than random chance. If the regression is run on ln CO2 alone, R2 is still 76.4%, and since adding any extra variable always increases R2 a little, the TSI term is effectively making no difference at all. Running the analysis with other measures of solar activity, such as sunspot number, years since solar minimum, or years since solar maximum, makes no difference to the results.

It's not the sun. It's carbon dioxide. And if we don't do something to control our output of that greenhouse gas, we're going to be in serious trouble.

Page created:07/11/2009
Last modified:  02/13/2011