Monday, March 24, 2008

Climate science the right way: An example of the Sun-climate connection

Many people have noted and investigated the connection between the Sun and the Earth's climate. Since the late 19th century, most of the scrutiny has focused on the Sun's roughly 11-year magnetic cycle.* Its connection to the Earth's climate is small but significant, important enough to be one of the prime determinants of Earth climate during the current interglacial period. We'll return to this solar influence later, as it's not fully understood.

But there is another influence. The Sun "flickers," over its whole radiation frequency range, on time scales running from days up to a few years. (The technical name is total solar irradiance (TSI) fluctuations.) This flickering leaves an "imprint" on the Earth's climate which will show up in many measured time-series indexes. Because the Earth's climate is nonlinear, we should expect subharmonics to form, at lower frequencies or longer periods - months to decades. The IPCC and many scientists tend to dismiss such intermittency as noise - as if chaos had never been discovered and thrown into question whether "noise" even exists at all.

A newly published empirical study of the Sun-Earth climate connection demonstrates that indeed this does happen. Not only does it happen, this flickering alone explains roughly half of the variation of a "global temperature" statistical index over the last 60 years.** The authors, Scafetta and West, summarize their work here (PDF). They extended it back over the last four centuries by combining the shorter-term flickering with the 11-year solar magnetic cycle variations and explained roughly three-quarters of the variations of the "global temperature" index.

Scafetta and West's work is a beautiful piece of science done right. They identified the correlation between the flickering signal and the temperature index signal in an airtight way, checking their result by randomly scrambling the data's time ordering to see if their result would change. It didn't. As they point out, this result indicates that, while the coupling between the "space weather" caused by solar changes and the Earth weather is weak from the point of view of energy transferred, it does conserve information: the same structure of events and times in one signal shows up in the other, like a faint echo mimicking some distinctive signal.

"Consensus" science fails again. So what gives in the world of "official" climate science? While the scientific reports of the IPCC do acknowledge that the Sun-climate connection is important, they waffle on its exact nature and significance. The IPCC summary reports arrogantly dismiss it altogether. Previous attempts to nail it down quantitatively came up with ambiguous results. But Scafetta and West's methods show why: those negative results came from statistical techniques based on Gaussian "Mediocristan" methods and assume the central limit theorem. We've seen how it can and does break down; chaos and Extremistan behavior is all around us. It's taken science and mathematics several centuries to be able to cope with such phenomena, but modern methods up to the task are available.†

Why aren't they being used by the IPCC? Beats me. But one obvious result of the politically-driven, journalistically-obsessive climate hysteria is junk results based on unphysical concepts and bad methods - methods and concepts that have been proven wrong, don't make sense, or have been superseded. This cost of climate hysteria - the cost of bad theory, scientific regression, and intellectual corruption - is a serious topic for another day, one that deserves its own consideration.

My only real beef with Scafetta and West's summary of their technical results is that Physics Today published it as "opinion." It's not: it's real science. What can be said about the IPCC and its executive summaries is another matter.



In a short while, we'll see other, similar, but older results that seem to get up to 80% or so of the temperature index variation by comparing it to the Sun's variability. I'm willing to bet money on the proposition that, with all these solar influences taken into account and combined with the Earth's internal climate cycles, all or almost all (90% or more) of the variability can be correlated, not just back four centuries, but all the way back to the end of the last Ice Age.††

POSTSCRIPT: Here's another nail in the "global warming" coffin, from NPR: the case of the missing ocean heat that "should" be there. (Hat tip to Instapundit.)

As the coffin nails pile up in the case of "global warming," some are even talking about the pending collapse of the whole climate scare: see this from The Australian.
---
* Now you know enough to understand that "roughly periodic" means its Fourier spectrum isn't peaked at one frequency, but has a well-defined, but not infinitely narrow, peak at around 1/11 inverse years of frequency. The "flickering" is a broader, flatter part of the spectrum at higher frequencies running from inverse months to inverse days.

The solar irradiance is itself modulated by strongly non-linear processes near the Sun's surface, including the Sun's magnetism. Some of these processes themselves seem to be chaotic. So we have one chaotic system (the Sun's surface and atmosphere) coupled to another (the Earth's climate), the former "pumping" the latter with energy and leaving behind distinctive "information fingerprints."

** Such an average is physically meaningless in and of itself. But all points on the Earth's surface illuminated by the Sun experience this flickering, and the effect under discussion here necessarily has to show up in some way in any statistical index built up from local temperature measurements. The index used doesn't have to have a physical meaning.

† Scafetta and West state their result in terms of the statistical distribution P(t) of the time t between "events" (flickerings), which is different from, but equivalent to, the chaos and non-Gaussian discussions earlier. They find P(t) ~ (λ/t)α, with α about 2.1 to 2.2.

In the Gaussian case - which would hold if, for example, the Earth were a closed thermodynamic system at a single temperature - this distribution would be exponential, P(t) ~ exp(-t/λ), with some relaxation time λ characteristic of the whole system. This is roughly the time that it takes a system to "settle down" after a single external disturbance.

Power-law, instead of exponential, decay in time of a disturbance is a different sort of "long tail" phenomenon. It means the disturbance takes much longer to really disappear than you might naively expect. Such behavior has been seen in laboratory measurements of controlled systems since the 1930s. Recall that a chaotic system is one that never settles down. The disturbances never stop; one leads to another, and so on.

†† I'm excluding the exactly periodic daily and annual cycles. It's sometimes good to state the obvious.

Labels: , , , , , , ,

0 Comments:

Post a Comment

<< Home