Wednesday, December 22, 2010 ... Deutsch/Español/Related posts from blogosphere

HadCRUT3: 2010 will be 2nd-5th warmest year

Phil Jones' mailbox has already been unmasked while James Hansen's mailbox has not.

It just happens that since the ClimateGate, Jones' team is indicating a much lower warming trend than Hansen's team: Phil Jones' HadCRUT3 dataset is attributing November 2010 the coolest rating among Novembers and among the four datasets - with a 0.43 °C global anomaly, it was the 7th warmest November - while the GISS dataset says that November 2010 was the warmest November on record.

The satellite datasets sit in between: November 2010 was 3rd for UAH and 6th for RSS.




The same significant discrepancy exists for the whole year 2010.

While GISS is going to say that 2010 has surpassed 2005 by 0.04 °C or so (assuming that the Dec 2010 anomaly will coincide with the Nov 2010 anomaly, which is a good approximation), becoming the GISS' new warmest year, HadCRUT3 will release a substantially different ranking: 2010 won't beat 1998, it may easily happen that it won't beat 2005, and it is not impossible that 2010 will be cooler than 2003 and maybe even 2002.

Setting the Dec 2010 anomaly to the Nov 2010 anomaly for HadCRUT3 as well, the top five ranking - with years and anomalies - will be

  • 1998: 0.548 °C
  • 2010: 0.488 °C
  • 2005: 0.482 °C
  • 2003: 0.475 °C
  • 2002: 0.465 °C
If the December's anomaly will be higher/lower than the November anomaly by "D", you have to add/subtract "D/12" from the 2010 figure above. So you see it's not quite impossible that 2010 may drop up to the fifth place - assuming month-on-month cooling by  0.2 °C. However, it is already de facto impossible to beat 1998 - the month-on-month warming would have to be 0.7 °C or so while the biggest month-on-month changes they have seen since 1880 were +0.44 and -0.54 °C.



In the newest conversation with fellow activist Bill McKibben, James Hansen uses the term "deniers" nine times. You're so courageous, Mr Hansen, to have skipped several doses of your medication! ;-)

But we surely appreciate your opinion that the skeptics should receive a Nobel prize for having showed that the data supporting "climate disruption" were wrong.

A new development is that two days ago, the New Zealand climate agency (NIWA), after a patient battle by the Kiwi climate realists and auditors in general, has abandoned their previously distorted temperature data that showed a degree of warming, and returned to the data that show no warming. Via Daily Bayonet and Karlos.



Met Office denies that it has predicted anything about the winter

According to its Chief Press Officer, the British Met Office categorically denies that it has made any predictions about the weather in this winter.

However, they didn't manage to burn and delete all U.K. newspapers from October 28th, 2010.



If they manage to delete the web page above, here's a copy to prove that the Met Office is composed of deniers or, more precisely, shameless liars and crooks.

Add to del.icio.us Digg this Add to reddit

snail feedback (2) :


reader DizzyRingo said...

For the man in the street, it would be helpful if:
1. the basis of the anomaly is defined
2. The size of the anomaly was given
when the MSM reported the purported "hottest" year.

It was quite by chance that I discovered in my research that the rating is based on a comparison with the end of the 19th Century - i.e. 1899 or so. And then, as you so correctly quote, the anomaly is about .5 to .9 of one degree, depending on the situation.
If this is made clear to the public then this would put the whole warming stuff into perspective.
Actually, I would have thought the comparison with the CET - which runs from the early 1600s - is much more useful.


reader Luboš Motl said...

Dear Dizzyringo,

I don't really think that the numerical size of the anomaly is something that a scientist - or even a man in the street - should be interested it. It depends on conventions.

As you correctly point out in 1), the anomaly only makes sense relatively to a base point. But there is clearly no "God-given" base point for the calculation of an anomaly.

So at different moments, different teams take the base point value to be the average of all their data, e.g. 1850-2010, or just 1960-1990 data, or 1950-2000, or any interval you can think of.

Generally, the surface station records go to more distant past than the satellite record, so their base point contains many more "colder" years, and the anomalies are therefore higher than they are for the satellite record. This is on top of the fact that the satellites see a slower warming trend.

Obviously, if we calculated the anomaly relatively to the average of a period that includes the last ice age but not the previous interglacial, the anomaly would be higher by several degrees. That would imply exactly nothing - it's a pure convention. There is clearly no "right" temperature.

The additive uncertainty doesn't pose any problem as long as you calculate the differences of the anomalies only. In particular, if you decide which year on the observational record is the warmest one, it's not affected by the additive shift - by the choice of the base point.

So I kind of disagree with your priorities. The information about the ranking of the years is much more objective, convention-independent, and therefore physically more interesting information than the numerical size of the anomaly.

For Central England, even if you included the whole record since 1659, the base point wouldn't shift much - the anomaly would still be well below 2 °C, to say the least.

At any rate, the differences in the temperatures of individual years during the last 150 years are tiny - all of them fit within 1 °C - and this is how Nature naturally behaves. Even a healthy human body doesn't keep a constant temperature with a better accuracy.

Best wishes
Lubos