Should we believe the cosmic ray flux theory (CRF)? Here I attempt to answer this question quantitatively, by calculating the strength of evidence so-far presented for CRF as a major forcing factor in climate change. Specifically we need to ask, what is the probability of being wrong about CRF? This can be calculated by combining the significance values of independent lines of evidence.

Below I have started calculating and tabulating the P values. The first 8 rows were worked out from the difference of means from Shaviv’s paper, with and without CRF. I have a sense that independence of evidence can be judged by the manner in which CRF or its response is measured, so I have listed that in the table. At the long time scales I think CRF is estimated using an isotope of iron in meteorites (Fe). Over medium periods 10Be was used, while at shorter time scale the climate sensitivity was calibrated on the solar output TSI.

I take this roughly as three independent sources of evidence, as follows:

Period P=0 Time yrs Indicator
Phanerozoic 0.34 500,000,000 Fe/temp
Cretaceous 0.19 50,000,000 Fe/temp
Eocene 0.56 20,000,000 Fe/temp
LGM 0.25 10,000 10Be/temp
20thCentury 0.08 100 10Be/temp
SolarCycle 0.08 ~11yrs TSI/temp
CombinedC 0.09
CombinedLM 0.08
UK20thCent 0.01 50 Neutron/cloudiness
Forbush 0.05 0.1 Neutron/cloudiness

The last two lines are the effect of CRF as measured by neutron flux, on cloudiness in Empirical evidence for a nonlinear effect of galactic cosmic rays on clouds (2006) by R. Giles Harrison and David B Stephenson. This paper finds a variations of 20% in cloudiness between the max and mins of CRF. The effect is detectable even at the shortest timescale of a Forbush event, a sudden and transient reduction in cosmic rays lasting a few days.

The combination of independent probabilities is simply their product. The probability of the theory that CRF affects climate is given by four probabilities, multiplied together:

0.33*0.15*0.08*0.01 = 4xe-5 = 4 sigma

Even a conservative estimate where some results are ignored provides a 4 sigma significance for the CRF theory. This level of significance is typical, nay expected in physics, while climate science is lucky to achieve 2 sigma, or around 95% confidence. The numbers show that the probability the CRF theory is wrong is very low indeed. In other words, the CRF theory has a 0.004 % or 1 in 25,000 chance of being wrong, so far.

The evidence shows CRF forcing climate change, at most time scales. In contrast, CO2 is uncorrelated at both the long and short time scales, and at the medium scales the direction of causation is uncertain. Only the PDO/NAO would there seem to be another major factor. Shaviv estimates that only 20% of the last centuries warming is possibly attributable to green house gases.

A lot of posts here have been negative — highlighting the sloppiness of climate change statistics and the self-serving exaggerations of climate effects scientists. For the first time I am becoming convinced that the evidence is really there to show CO2 is just a bit player in climate change, and there is another factor that can explain a large chunk of the wiggles that we see in global temperature changes.

Here is the data from Table 1. sens

Here is the turnkey R code.

d<-read.table("http://landshape.org/enm/wp-content/uploads/2009/03/sens.txt")

prob<-function(m1,s1,m2,s2) {
s<-sqrt(s1^2+s2^2)
z<-(m2-m1)/s
pnorm(z)
}

run<-function() {
for (i in 1:8) {
m1<-d[i,4]
s1<-d[i,4]-d[i,3]
d2<-d[i+8,4]
s2<-d[i+8,4]-d[i,3]
print(prob(m1,s1,d2,s2))
}
}

run()