Sunday, September 12, 2010

GISS, HadCRUT3: August 2010 was 7th warmest August

For years, we would think of the surface record as being the main empirical pillar supporting the idea of a global warming trend. The satellite RSS AMSU and especially UAH AMSU records showed a significantly smaller warming trend than GISS and HadCRUT3, the surface station-based datasets.

Some of us - those who were "cynical" or "suspicious", to put it mildly - used to think that it was natural because some of the worst alarmists were behind GISS and HadCRUT while some fine skeptics were working on UAH AMSU. However, times are changing. The recent months look much cooler through the GISS's and HadCRUT3's glasses than from the UAH's or RSS's perspective.

This video was shot elsewhere but it's still true that Tom's Diner from Suzanne Vega's song is right below NASA's GISS. I included this video because I prefer the instrumental version. ;-) You may also try the Václav Havel 20th Velvet Anniversary 2009 version of the first MP3 song by the mother of MP3. ;-)

You certainly expect some details. Here they are.

According to UAH AMSU, connected to the names of John Christy and Roy Spencer, August 2010 was the
UAH: 2nd
warmest August on their record. Its global anomaly of +0.51 °C was just 0.01 °C cooler than the anomaly +0.52 °C measured in August 1998. However, their colleagues at RSS AMSU got an even warmer result. The most recent available figure is from July 2010 and among Julies, it was:
The anomaly of +0.608 °C in July 2010 was 0.002 °C (two millikelvins) warmer than the anomaly of +0.606 °C in July 1998, its nearest competitor. (Update: at 0.58 °C, August 2010 was also the hottest August in RSS records.) Given this data and the past experience with the difference between satellite and surface records, you could expect GISS and HadCRUT3 to report an "ultrahot" July and August 2010.

But you would be wrong.
Both teams, GISS and HadCRUT3, think that August 2010 and July 2010, respectively, was the 7th warmest August or July on their record. The rank is 5 or 6 steps lower than the rank according to the satellite records! The rank 7 puts 2010 exactly to the middle of the 13 years between 1998 and 2010. Imagine: after this another overhyped "ultrahot" summer (or, in Australian English, winter), GISS and HadCRUT3 conclude that July or August 2010 was an average month in the recent 13 years.

Update: HadCRUT3 later reported August 2010 as the 7th warmest August, too.

For example, here is the ranking of Augusts in these 13 years according to GISS:
01. 1998, 0.63 °C
02. 2003, 0.62 °C
03. 2006, 0.60 °C
04. 2005, 0.56 °C
05. 2009, 0.55 °C
06. 2007, 0.54 °C
07. 2010, 0.53 °C
08. 2001, 0.45 °C
09. 2002, 0.45 °C
10. 2004, 0.42 °C
11. 2000, 0.37 °C
12. 2008, 0.35 °C
13. 1999, 0.27 °C
This is just the ranking of the recent 13 years' eighth months. There was a dozen or so of earlier years that were warmer than e.g. August 1999 but all of them were cooler than August 2010.

Similar results hold for the newest month from HadCRUT3, which is July 2010.

Michael Mann, James Hansen, and Gavin Schmidt wrote a new paper in which they argue that the Earth used to have a flat climate (the shaft of the hockey stick) because it was a flat Earth. And they have even explained what lies at the boundary of the pancake.

If this "reverted" behavior will continue, the surface records may produce a more substantial numerical cooling - or smaller warming - than the satellite records in the years to come. It may even happen that some "professional alarmists" will switch to the satellites as their preferred source of data while some "professional skeptics" will start to prefer the data from GISS or HadCRUT3. ;-)

When it comes to the interpretations, there are many.

First, obviously, the trends measured by satellites and weather stations will converge towards one another a little bit so you may view this as one of the diminishing inconsistencies. The satellite-measured warming is still smaller but it doesn't have to be this way forever. Both trends are going to be small. (GISS still produces a positive trend, 0.68 °C per century, from 1998 to 2010 but it is probably statistically insignificant.)

Note that this comparison of the satellite and surface teams is something else than the comparison of the near-surface temperatures and the mid-troposphere temperatures; both of them may be measured by the satellites if you want to achieve the maximum accuracy.

Second, you could have thought that the higher trend of GISS and HadCRUT3 was here with us to stay. However, when the numbers have apparently reverted, you may say that the warm early 2000s according to GISS or HadCRUT3 were a fluke that was going to be largely cancelled by the opposite, imminent fluke.

A priori, you may have held several "obvious" beliefs about the future.

One of them would say that the surface-team trends would be higher forever. The other would be that the different teams will tend to produce similar trends in the long run - so that faster trends of one team and in one period would be compensated by slower (or cooling) trends of the same team in other periods.

The second answer seems to be more correct now, after August 2010, but you can never be sure what will happen in the future. The actual reasons why the different teams show pretty different detailed numbers are not understood. You may therefore parameterize the differences as "noise". Well, there are still different kinds of noise - depending on what you assume about their statistical distribution and autocorrelation.

When it comes to the near future, the daily UAH AMSU data finally indicate that the most recent days are cooler than the days that took place exactly one year ago - after months when almost every day would pretty much set the hot record for the day.

The cooling may continue in the months to come because the La Nina that was getting started in recent months seems to be a truly giant one. We will see. The most recent ONI 3.4 anomaly was -1.6 °C. I expect the number released tomorrow will be even more extreme. After 1988, there's been no three-month period when the average anomaly was below -1.6 °C.

Update: A reader has correctly observed that the HadCRUT3 and RSS figures are from July 2010, not yet August 2010 which are not yet out. The comparison between UAH and GISS may still be made and I don't expect the final August rankings to be too different. Thanks to Mr Pavel Panenka for the correction. (Update: indeed, the at least the RSS ranking stayed the same for August 2010 - first hottest August.)