When it comes to testing a climate model, the future is more important than the past. You can take your model and adjust its various parameters to fit observations already made, but the crucial test is how well that model matches the future. This involves making a prediction, but, as they say, prediction is difficult, especially about the future.
The important thing about a prediction is that once made you don’t modify it or start making excuses. It is prediction though that sets a model apart from mere curve fitting and tells you if your model is in touch with reality and is not just matching wriggles on graphs. The problem with computer models of the climate is that ideally several decades are needed to see if they are any good. This is tricky to fit into an academic life and frustrating for all concerned, but there are other ways.
Shorter-term predictions will contain more noise but if the model captures accurately the putatively increasing man-made warming signal in the data it should be possible to say something statistically about its presence. In the past few years the Met Office has been doing exactly that. What they have also been displaying is a public lack of self-evaluation, and hoping we have short memories.
In 2007 it became obvious that annual global temperatures were not increasing in the past few year, and a few people said so in public what scientists were talking about in private hoping it wouldn’t get out.
With A Vengeance
In August of 2007 the UK Met Office released a study that predicted that global warming will set in with a vengeance after 2009, after they implicitly acknowledged the temporary warming standstill, with at least half of the five following years expected to be hotter than 1998, which was then, and is still the warmest year on record.
The Guardian said:
British scientists are predicting a succession of record-breaking high temperatures in the most detailed forecast of global warming’s impact on weather around the world.
Powerful computer simulations used to create the world’s first global warming forecast suggests temperature rises will stall in the next two years, before rising sharply at the end of the decade.
From 2010, they warn, every year has at least a 50% chance of exceeding the record year of 1998 when average global temperatures reached 14.54C.
Reporters were told that in order to make such a prediction, researchers at the Met Office had made a computer model that takes into account such natural phenomena as the El Nino pattern in the Pacific Ocean, and other fluctuations in ocean circulation and heat content.
The Telegraph said:
This is the prediction of the first computer model of the global climate designed to make forecasts over a timescale of around a decade, developed by scientists at the Met Office.
“This is a very valuable step forward,” Science magazine was told by meteorologist Rowan Sutton of the University of Reading. “It’s precisely on the decadal time scale and on regional scales that natural variability and anthropogenic (man made) effects have comparable magnitudes.”
The climate model was described in Science. It states, “…at least half of the years after 2009 predicted to exceed the warmest year currently on record.”
How well did this computer model do by its own test?
We are now two years into their five year prediction, and as yet none of them has been anywhere near the temperature of 1998. This means that, if the prediction is correct, then all three of the subsequent years, including 2014, must be hotter than 1998. Looking at the data I don’t think this is likely to happen.
In the event of the prediction’s failure it would be disingenuous to say: oh well, there were more La Nina cooling events than we thought, but that the model is sound. Such events were incorporated into the model. Strictly speaking if it fails its self-appointed test then the model is unsound, it has failed its crucial test at the least with its specific input parameters.
The problem is however that the Met Office keeps moving the goalposts.
After the 2007 prediction, and with the addition of only one annual more temperature datapoint (that for 2008 which was statistically identical to the previous years since 2001), in 2009 Prof Phil Jones of the University of East Anglia wrote in an email:
Tim, Chris, I hope you’re not right about the lack of warming lasting till about 2020. I’d rather hoped to see the earlier Met Office press release with Doug’s paper that said something like – half the years to 2014 would exceed the warmest year currently on record, 1998!
Then at the end of 2009 the Met Office admitted that 2009 was not a record breaker and was in fact near the “lower end of expectations.”
The problem was, of course, the cooling effect of a La Nina, which even though incorporated into the models, had compensated for the underlying warming. As Phil Jones put it, the presence of La Nina during the last year partially masked this underlying rate.
Phenomena such as El Nino and La Nina have a significant influence on global surface temperature, said Dr. Chris Folland of the Met Office Hadley Centre. Further warming to record levels is likely once a moderate El Nino develops. The transition from a La Nina effect to an El Nino one is expected late next year.
Of course, even in the presence of no warming seen in the past decade a strong El Nino might boost temperatures to a record level, even though that would say more about the El Nino’s starting point and not the underlying decadal temperature increase. Such crucial scientific specifics were however lost, at least in public. So the hopes were high that 2010 would be a record year.
Climate could warm to record levels in 2010, the Met Office said:
The latest forecast from our climate scientists, shows the global temperature is forecast to be almost 0.6 °C above the 1961-90 long-term average. This means that it is more likely than not that 2010 will be the warmest year in the instrumental record, beating the previous record year which was 1998.
It wasn’t. The forecast was a spectacular failure.
Despite a strong El Nino it had a temperature anomaly of 0.470 (with respect to 1961 – 90 average) making it the third warmest year though when the errors are looked at it is in undistinguished member of the post-2001 standstill.
In another rewriting of history Prof Phil Jones said that due to natural variability we do not expect to see each year warmer than the last, but the long-term trend is clear.
The problem is that no year has been warmer than the last!
So, forever hopeful, onto 2011. If the model is correct it must surely be the warmest, after all if the model is right a new warmest year ever must be along any year soon.
The Met Office said that for 2011 they were expecting another very warm year, with a global anomaly forecast of +0.44C above the 1961-1990 average. That would make 2011 the equal 6th warmest year on record.
It isn’t. Whilst we wait for the November and December data 2011 has a temperature anomaly of 0.356 and is the 11th warmest year. Again the Met Office forecast was way off.
Remember the 2007 prediction (half of the post 2009 years would be warmer than 1998). Curiously, at the beginning of 2011 the Met Office revised their prediction and said they expect that ‘half the years between 2010 and 2015 to be hotter than the hottest year on record’.
The goalposts have moved! 2009 – 2014 has become 2010 – 2015. The cardinal rule in making predictions is not to fudge them, and be honest about the results before making a new prediction with a revised timescale.
If the Met Office climate model is correct then there should be more than 50% of the years 2010 – 2015 exceeding the 1998 record. In the Times of 29 Nov 2011 Peter Stott of the Met Office changed the goalposts again, saying that they stood by their previous prediction that half the years in this decade would be warmer globally than 1998, the warmest year to date.
None of the Met Office forecasts have been correct. The global annual average temperature has to increase soon or else the goalposts will have to be moved again. At what stage does one admit the predictions were wrong?
The latest news is that a second cooling La Nina is brewing in the later part of 2011 and may extend into 2012. Given that, I wonder what the predictions for 2012 will be?