Skip to content

Met Office Decadal Forecast–2007 Version

February 6, 2013

By Paul Homewood


The latest decadal forecast, above, of global temperatures, issued by the UK Met Office in December, created quite a stir for several reasons, not least the fact that it predicts an effective flatlining for the next few years.

It is worth, though, reminding ourselves that the first decadal forecast was published by them in 2007, as per the graph below.

Globally averaged annual mean surface temperature anomaly (relative to 1979–2001) forecast by DePreSys starting from June 2005. The CI (red shading) is diagnosed from the standard deviation of the DePreSys ensemble, assuming a t distribution centered on the ensemble mean (white curve). Also shown are DePreSys and ensemble mean NoAssim (blue curves) hindcasts starting from June 1985 and June 1995, together with observations from HadCRUT2vOA (black curve).


According to this decadal forecast

Both NoAssim and DePreSys, however, predict further warming during the coming decade, with the year 2014 predicted to be 0.30° ± 0.21°C [5 to 95% confidence interval (CI)] warmer than the observed value for 2004. Furthermore, at least half of the years after 2009 are predicted to be warmer than 1998, the warmest year currently on record.

The Met Office kindly sent me a list of the actual temperatures that they had forecast, so that we can see a direct comparison with actuals, and I show this below. You will note that the forecast anomalies, in the above graph, are against a baseline of 1979-2001. The figures supplied by the Met are based against 1971-2000, whilst the actual HADCRUT numbers produced are based on 1961-90! Therefore some conversion is in order.( The period 1961-90 was 0.118C colder than 1971-2000, according to HADCRUT3).


Year Forecast v 1971-2000 baseline Forecast v 1961-90 baseline HADCRUT3 Actual Temperatures
2004     0.447
2005 0.415 0.533 0.482
2006 0.406 0.524 0.425
2007 0.426 0.544 0.402
2008 0.486 0.604 0.325
2009 0.534 0.652 0.443
2010 0.558 0.676 0.478
2011 0.637 0.755 0.341
2012 0.617 0.735 0.402
2013 0.627 0.745  


I have included the actuals for 2004, so we can measure the prediction that 2014 would be 0.30C warmer, i.e. 0.747C. The temperature for 1998 was 0.548 in comparison.

You will also note that, although the decadal forecast was issued in 2007, it does include “forecasts” for 2005 and 2006. It is not clear whether these are hindcasts, or forecasts that were made while the work was ongoing prior to publication. Nevertheless :-

  • Temperatures are still below 2004 levels, with no sign of approaching the 0.745C forecast for 2013.
  • None of the years have been hotter than 1998.





I criticised the Met Office last month, for not being open about the errors they had made in their earlier forecasts. At the time, what they now say are “retrospective predictions”, (i.e. hindcasts), were labelled as “previous predictions”. This gave the misleading impression, whether deliberate or not, that “previous predictions” had been quite accurate.

However, they still fail to compare the latest forecasts, with previous ones. Indeed, none of the earlier ones seem to be in the Met Archives. I only got access to the 2007 version by specifically requesting a copy.

If a private organisation had got its forecasts so utterly wrong, heads would be rolling. In the Met Office, they simply cover them up.

  1. Andy DC permalink
    February 6, 2013 10:35 pm

    It shows that any long range weather forecast is at best a stab in the dark. These guys are way in over their heads trying to do claim they can make such predictions.

  2. grayman permalink
    February 7, 2013 5:07 am

    No, heads do not roll. They get a BONUS instead!

  3. February 7, 2013 9:44 am

    From the Met Office

    “Our models cannot predict volcanic eruptions, the Mount Pinatubo eruption appears as a feature in the graph you refer to because these are retrospective forecasts, or hindcasts, which are used to assess the model performance. These should not be interpreted as actual forecasts. A hindcast is a way of testing a mathematical model. Known or closely estimated inputs for past events are entered into the model to see how well the output matches the known results. In retrospective forecasts volcanic forcing was specified, according to the CMIP5 protocol, such that eruptions occurring during the forecast were included even though they could not have been predicted at the time. “

  4. Jim Cripwell permalink
    February 7, 2013 12:05 pm

    You write “You will also note that, although the decadal forecast was issued in 2007, it does include “forecasts” for 2005 and 2006. “

    The original paper was written using 2004 data, and there were forecasts for 2005 and 2006. It took until 2007 before it was published in Science…

    • February 7, 2013 12:11 pm

      Thanks Jim.

      I guessed that might be the case, but could find no reference.


  1. Weekly Climate and Energy News Roundup | Watts Up With That?
  2. “Global warming predictions prove accurate”– Guardian | Watts Up With That?

Comments are closed.

%d bloggers like this: