- JoNova - http://joannenova.com.au -

Are transfer functions meaningless (the “white noise” point)? Beware your assumptions!

Posted By Joanne Nova On June 29, 2014 @ 8:33 pm In Global Warming | Comments Disabled

Some people are claiming that the transfer function is meaningless because you could use white noise instead of temperature data and get the same notch. It’s true, you could. But the argument is itself a surprisingly banal fallacy. It looks seductive, but it’s like saying that it is meaningless to add 3 oranges to 3 oranges because you could add 3 oranges to 3 apples and you’d still get six!

It is trivially obvious that the transfer function will find a relationship between entirely unrelated time series, as any mathematical tool will when it’s misapplied. The question that matters — as with any mathematical tool — is has it been misapplied? What matters is whether the base assumption is valid, and whether the results will be a useful answer to the question you’ve asked. If the assumption is that apples and oranges are both pieces of fruit, and the question you ask is “how many pieces of fruit do we have”, then it is useful to add apples and oranges. But if you are trying to compare changes in fruit consumption, adding the two is mindless. So let’s look at the assumptions and the question being asked.

Assumptions first

Two assumptions were made before computing the transfer function. And before anyone complains that the whole project was a circular tautology — pay attention — the assumptions are temporary. They are a “what if” used to see if we get a meaningful answer. Later the assumptions are dropped and tested.

Assuming that:

1. Recent global warming was associated almost entirely with TSI.

2. The climate system is linear and invariant

…then, the transfer function from TSI to temperature is of great interest and sinusoidal analysis is appropriate.

David Evans has been explicit about both right from the start, but not all commenters seem to realize the implications.

The transfer function between TSI and Earth’s (surface) temperature will be meaningless if there is no causal link between TSI and Earth’s temperature. (Some people may need to read that twice).

This “link” could be an indirect one. It doesn’t mean that TSI itself is causing the change in temperature. It could, for example, mean that TSI is a leading indicator of other solar events that lag it by 11 years. It could be that those other events — say magnetic fields, solar wind, UV or other spectrum changes — are the ones actually causing the albedo changes that cause the temperature to change 11 years after the TSI changes.

By all means, if you have definitive evidence that changes in TSI cannot possibly be directly or indirectly associated with changes in Earth’s temperature, do let us know. It will save us a lot of time. Likewise, if you know of any reason why TSI can not possibly be a leading indicator for some other solar factor which acts with an 11 year cycle, please let us know. Some people are willing to declare they know that TSI cannot be associated with changes in Earth’s temperature.  Some of us have an open mind. The solar dynamo is not completely worked out. Fair?

What about the question we are trying to answer?

As to the second part, what question was David Evans asking, and are the results useful? He made it explicit.

The initial aim of this project is to answer this question: If the recent global warming was associated almost entirely with solar radiation, and had no dependence on CO2, what solar model would account for it?

So is the discovery of a notch filter useful, and does it help to create a solar model? It certainly looks that way so far.

The model was constructed in the frequency domain. The main feature in the transfer function is the notch, so we tried building a similar notch in the model. The existence of the notch implies there has to be an accompanying  delay (the timing seems unnaturally perfect, people are understandbly having trouble wrapping their heads around that). The delay was later found to likely be 11 years, which is not only the length of the major cycle of the solar dynamo but is borne out by other independent studies– such as Usoskin, Soon, Archibald, Friis-Christensen and Lassen, Solheim, Moffa-Sanchez, etc. (see  post III). Later, a model based on an 11 year delay was found to produce reasonable results (see that hindcasting).

So the notch turned out to be very useful in building the model, giving two of the  five elements in the model (the other three are the low pass filter, the RATS multiplier, and the immediate path for TSI, which were deduced by physical reasoning).

It would be better if there was a known mechanism. Of course, but Rome wasn’t built in a day, steady on. We are working on it. If people already knew what force X was then presumably they would have noticed its correlation with temperature and the climate problem would already have been solved, wouldn’t it? Some commenters, (those not focused on fallacies like argument from incredulity or the mechanics of publication time-tables) are being very helpful in gathering clues on force X — thank you!.

PS: The release?

And for those who are impatiently waiting the full working model, we’re working on it. There are a few last-minute things  to sort out. The spreadsheet used data to August 2013 in the investigation, and was frozen months ago with that data. That’s the copy that is available to people who got advance notice. Now that we are releasing it, it would be nice to update the data, while preserving the original calculations. David is copying the Aug 2013 data and updating all the data. We are also figuring out the creative commons conditions that would be workable, and deciding how to manage suggestions, adaptations, and modifications. We suspect the normal open source software sites don’t deal with 20Mb Excel files which people can modify, but which are very difficult to track changes on (does anyone know of a similar project?). Right now the sciencespeak legal department, open science support team, human relations division and marketing arm are working flat tack. (That’s both of us. ;-) )

The biggest impediment at the moment is that some people still haven’t read the first posts we put up carefully enough. Even though we answered their questions personally in comments they still keep repeating the same points. Should we have kept the whole project secret until we had solved all these questions? Perhaps, but it’s been immensely helpful to get some feedback and help from some readers, and we didn’t know who would be the most useful beforehand. They have made themselves known.

On the other hand we’re being compared to Phil Jones and Michael Mann by one commenter, which we think is a tiny bit over-the-top, given that Jones and Mann are funded by the taxpayer and they spent years and used legal means to prevent their data being made public. To put a fine point on it, we got no income from taxes, and we owe the critics nothing. We also ask nothing of them (except, implicitly, patience and manners). Maybe that looks equivalent to a few people –  we can’t see it. All the fuss, seriously, is flattering (if counter-productive).

Manners makes no difference to the scientific method, but ultimately the human practice of the Scientific Method is only ever advanced by … humans, and manners do matter. Science is never advanced by namecalling, misquoting, strawmen and personal attacks. Please quote us exactly, eh?

VN:F [1.9.22_1171]
Rating: 9.3/10 (138 votes cast)

Article printed from JoNova: http://joannenova.com.au

URL to article: http://joannenova.com.au/2014/06/are-transfer-functions-meaningless-the-white-noise-point-beware-your-assumptions/

Copyright © 2008 JoNova. All rights reserved.