People send me things. Today it is a curious graph of the number of supernovae (dying stars) discovered versus the HadCRUT temperature data since 1960. There’s a good correlation. So at first glance you might conclude two things, 1) GCR’s, which are known to be the result of supernovae thanks to data gathered by the Chandra Space Telescope, are indeed influencing Earth’s temperature or 2) Earth’s AGW is killing stars, and aliens are correct to be concerned about Earth and may need to wipe us out to protect the Universe.
Our contributor at an observatory sheds more light on the subject. He writes:
I am a senior research fellow at ICRAR (International Centre for Radio Astronomy Research) in Perth, Australia. I was studying the sample of supernovae (SNe) discovered in the last 50 years (source: Harvard-Smithsonian CfA List of SNe), and I discovered that the number of SNe discovered per year correlates pretty well with the temperature anomaly. I produced a plot, placed at the URL below. Clearly the temperature anomaly has a better correlation with the observed number of dead stars than with dead polar bears, tree rings, CO2 or number of pirates. This is proof that global warming is causing more stars to explode. It’s worse than we thought. We are killing the universe. We need more funding.
Dr Rob Soria
International Centre for Radio Astronomy Research
This person is all legit, he’s real and at ICRAR. The data appear so well correlated, it would seem to be a cinch to use this to apply for a research grant, no matter which premise you want to prove. The possibilities are tantalizing. But, let’s analyse the data first.
The first thing I asked for is the data source for Supernovae (I know where to get HadCRUT data), which he provided here:
Sure enough, his work was replicable.
I spotted a couple of curious things though. Why the logarithmic graph on the right Y axis, and why only use data back to 1960, that favorite cutoff date for “hide the decline”?
Well there’s data, and then there’s data reporting bias. While it would be easy to conclude on this sample that there’s something worth further (funded) study, especially given the recent first results of the CERN CLOUD experiment, there’s a bit of a rub in the data. That rub has to do with the recent explosion of amateur astronomy and technology.
You see, around 1980 or so, affordable CCD detectors started to become available to the amateur astronomer, and in the decades that followed up to the present sensitivity increased 10x thanks to Peltier cooled CCD chips and other improvements in CCD imaging technology. Costs came down and you can now buy a good CCD detector for under $2000, often less than the cost of a good telescope.
So as a result, the number of detectors trained on the sky blossomed, and the number of supernovae detected by amateur astronomers soared. Hence the need for the logarithmic axis in the graph above. As for the cutoff date of 1960, well, um, the correlation doesn’t hold well before that. Thus, the decision was made to truncate the data prior to 1960. We figure if it was good enough for the hockey stick (which has been recently vindicated again) then it is good enough to do here to write a grant proposal.
Neither Rob nor I plan to write that proposal, but if any WUWT readers succeed in getting funded, I’ll happily publish a notice here.
So the moral of this story is: you can find short correlations in many things, such as correlating El Niño and Civil Wars, and truncating data is OK to make your point for the grant application and study, because you’ll be vindicated later if the study becomes popular and/or included in the IPCC AR5.
It also underscores the issue of reporting bias, which I’ve talked about again and again relating to the issue of bogus severe weather and AGW correlations, which simply don’t exist. They are a byproduct of improved radar systems, storm chasers, improved communications, and global 24/7 news gathering.
Caveat: For anyone reading with the composition of a neutron star, this essay is satirical, but with a real lesson: correlation is not causation.