the shorter the data the harder it’s distinguish trend from noise.

The innumerable skyrocket projections from the short warming 1978-1998 are warning examples and prove nothing.

About human influence Akasofu has interesting considerations:

“A roughly linear global temperature increase of about 0.5°C/100 years (~1°F/100 years) seems to have occurred from about 1800, or even much earlier, to the present…This trend (0.5°C/100 years) should be subtracted from the temperature data during the last 100 years when estimating the manmade contribution to the present global warming trend. Thus, there is a possibility that only a fraction of the present warming trend is attributable to the greenhouse effect resulting from human activities…”

Ref: Akasofu, Syun-Ichi, 2008. Is the Earth still recovering from the “Little Ice Age”? A possible cause of global warming. Study, revised January 23, 2008, online http://people.iarc.uaf.edu/~sakasofu/pdf/Earth_recovering_from_LIA_R.pdf

After all these longer natural variations, oscillations and cycles which are not in models, yet, we have the shortage of reliable data and in the end, the extremely vague ‘climate sensitivity’, when we try to estimate future climates and exclude impossibilities.

Statistics is used to prove something, but too ofter it is “lie, damned lie, statistics”.

Empirically, we do need 3 cycles of length “T” anything to estimate the energy spectral density associated with cycles of period “T”.

If the modelers saying we need “N” years would suggest that we need that number because there is a disproporionate amount of energy in some particular long cycle, then

a) the consumer of climate information would have been given a basis for the recommendation and

b) people from many, many fields could use that information to estimate the amount of uncertainty associated with using fewer years.

However, the sound bites in newspaper articles are utterly vague. If someone says one can’t test any hypothesis whatsoever at all using 10 years of data, but it becomes magically testable after 30 years, that is just nuts.

Our ability to distinguish between different hypotheses improves as we obtain more data. But certainly, most must recognize that empirical tests can be used to exclude, say 10C/century, or -10 c/century based on 90 months of data.

The question is: Can we exclude 2C/century? I say yes. Others say no. But simply saying the time is too short without discussing the properties of the data is foolish. It’s sad to read scientists resorting to puffery like claiming it’s reckless to examine claims with 7 years of data and then provide no solid criteria for how one determines the correct amount of time to test data.

FYI Gary D. Sharp once stated that we need reliable data of three 70-cycles i.e. 210 yr data to draw any conclusions. We have at least 300 yrs data about 70-yr cycles but the problem is how reliable they are. Perhaps we’ll have reliable projections in 50-100 yrs.

I have never encountered anyone who is unable to grasp the distinction between showing that 2 C/century lies outside the uncertainty of the data, but 0C/century lies inside it. I suspect if you explained it that way, you would have no difficulty getting them to grasp the notion either.

As for Roger’s statement that with short times, our uncertainty is large: That’s reflected in my uncertainty intervals. The 95% uncertainty intervals are currently about ±2C/century to ±3C/century depending on data set or method of estimating these intervals. That’s a large uncertainty interval. As we accumulate data, the uncertainty intervals will drop. Everyone agrees on this.

But there is no magic time when uncertainty intervals go from infinity to zero. And, if, hypothetically, a trend is sufficiently different from a prediction, we can say the prediction is incorrect– even with small amounts of data.

I strongly suspect that Gavin’s 20 years and Rahmsdorf’s 18 years comes from taking that contradiction into consideration; ie the period 1980 to 1998 is the only period that can be demonstrated to be a rapid warmth apparently not attributable to natural variation. So under 15 years is too short due to 10 years of plateauing temperatures and over 20 years is too long because that would rule out the unusual late 20th century warmth theory.

But the non-warming global temperature is actually only one of several indicators that the basic theory is wrong; probably partly due to the pessimistic decision not to include negative feedbacks and partly due to the simplistic conflation of electric control theory to a chaotic, non-linear climate. Though most explanations for this theory failure tend to conclude that the obs are wrong, ie always too cool, because they disagree with the models. This is the only area of physical science where it is argued that the obs are misleading and that we should trust the models. Of course it makes no sense at all because the whole idea of a model is to try to simulate reality and obs are our only indicator of reality and errors should be evenly distributed.

TT attempts to mix the Arctic warming with global warming but not only do we have valid, competing and yet unchallenged theories about Arctic warming, eg winds, soot, solar but we also have temperature records which say that it was just as warm there in the 30’s. The main oddity of the global warming chart seems to stem from a combination of apparently unreliable sea temperature data and a peculiar spike in Siberian temperatures which overlay an apparently unnatural climate shift from 1980 to 1998 onto an otherwise natural pattern. Time will tell but in the meantime can we stop this unscientific obsession with altering obs to match what is expected from the models?

50-100 yr projections are already abandoned, and shorter periods show primarily natural variations, oscillations and cycles of which the longer ones are not included in models, yet.

When this is a policy forum, I again emphasize that current actions to combat global warming are based on precaution.

I have nothing against precaution, but I call it precaution and not a proven scientific truth.

While I agree with you that “many people critical of Lucia Liljegren’s efforts (and to some extent my “consistent with” discussions”) confuse the evaluation of predictions with some effort to “disprove” AGW,” I’d say that this is also much in evidence among those who applaud Lucia’s efforts – viz., they confuse the evaluation of predictions as somehow showing that AGW has been disproved or that very obvious climate change isn’t occurring.

Do you have any suggestions on any layman-accessible readings that explain more about why “the evaluation of long-term predictions can be done with relatively short records, however your confidence in the results might not be very high”?

Further, isn’t the confidence in results/meaningfulness of the evaulation precisely the question? Should an evaluation period cover at least a full ENSO cycle? Why mock Lynch and Schmidt without explaining why they are wrong?

I am not sure what you mean by the notion of AGW being “disproven”. The theoretical and empirical bases behind the notion that humans influence the climate system is fairly robust in my (inexpert) opinion.

However, just because we have a robust set of understandings does not mean that we can offer accurate predictions of the evolution of that system. Economics provides a good example of a similar situation. We can observe that economic forecasts are sometimes less than accurate without “disproving” economic theory.

This is where I think many people critical of Lucia Liljegren’s efforts (and to some extent my “consistent with” discussions”) confuse the evaluation of predictions with some effort to “disprove” AGW. As I’ve said before (and I think Lucia has said the same thing as well), the evaluation of long-term predictions can be done with relatively short records, however your confidence in the results might not be very high. Lucia’s discussion is valuable because she explores and presents a range of ways to evaluate this confidence. It is not a yes or no question, which is why I highlight statements by Lynch that we have to wait 40 years or Schmidt (20 years) to evaluate climate forecasts. We don’t.

In addition to commenting on Amanda Lynch’s apparent “record”, do you care to venture either your own opinion on how soon AGW will be “disproven”, or your basis for implying that Lynch’s suggestion is way too long? Is Gavin Schmidt, at 20 years, simply more credible?

How about running a poll?!

By the way, has anyone told the Arctic that the world has been cooling since 1998?

http://climaticidechronicles.org/2008/08/11/temperatures-hit-80-degrees-in-the-arctic-2008-may-see-a-record-sea-ice-melt-after-all/

http://nsidc.org/news/press/20080610_Slater.html

Does anyone really believe that somehow all climate noise (and natural cycles) has disappeared, so that a period of relative stasis in the global average means that man is NOT affecting a climate that is still very obviously changing?