Annual global mean temperature forecasts
Each year since 2000, the Met Office has made an experimental forecast of the global mean surface temperature (GMST) for the year ahead. The forecasts rely on a statistical prediction system and, since 2008, additional decadal predictions made with a climate model.
We now have well over a decade of forecast data and this shows there is no obvious warm or cold bias in the forecasts, while the overall accuracy is around 80%. Full details of the method used for the forecasts and their accuracy are fully outlined in the research paper Folland et al 2013 (Geophysical Research Letters).
The table below shows the previous Met Office GMST forecasts as both a range and a central estimate (note that the range takes a wide range of likelihood into account - so the middle part of the range is considerably more likely to happen than the upper and lower extremes). The forecasts are verified against the latest World Meteorological Organization (WMO) estimate of observed GMST for each year. The final column shows the difference between the observed temperature and the central estimate from the forecast. The figure below the table shows the forecast range, with darker shaded areas around the central estimate representing a higher probability, along with the WMO observed temperature estimates.
|Year||Range of forecast at 95% confidence level (in °C)||Central estimate of forecast range (in °C)||WMO Global average temperature anomaly (95% confidence range of ± 0.1°C)||Difference between central estimate of forecast range and WMO temperature anomaly (in °C)|
|2000||0.25 - 0.57||0.41||0.30||0.11|
|2001||0.33 - 0.61||0.47||0.44||0.03|
|2002||0.33 - 0.61||0.47||0.50||-0.03|
|2003||0.41 - 0.69||0.55||0.51||0.04|
|2004||0.38 - 0.62||0.50||0.45||0.05|
|2005||0.39 - 0.63||0.51||0.56||-0.05|
|2006||0.33 - 0.57||0.45||0.51||-0.06|
|2007||0.38 - 0.71||0.54||0.51||0.03|
|2008||0.23 - 0.50||0.37||0.42||-0.05|
|2009||0.30 - 0.58||0.44||0.52||-0.08|
|2010||0.43 - 0.73||0.58||0.58||0.00|
|2011||0.28 - 0.60||0.44||0.46||-0.02|
|2012||0.34 - 0.62||0.48||0.50||-0.02|
|2013||0.43 - 0.71||0.57||0.54||0.03|
|2014||0.43 - 0.71||0.57||0.61||-0.04|
|2015||0.52 - 0.76||0.64||0.77||-0.13|
|2016||0.72 - 0.96||0.84||0.86 (Jan-Oct)||-0.02|
- Chris K. Folland, Andrew W. Colman, Doug M. Smith, Olivier Boucher, David E. Parker and Jean-Paul Vernier (2013). High predictive skill of global surface temperature a year ahead, Geophysical Research Letters, Vol. 40, pp1-7, doi:10.1002/grl.50169.
- Doug M. Smith, Stephen Cusack, Andrew W. Colman, Chris K. Folland, Glen R. Harris and James M. Murphy (2007). Improved Surface Temperature Prediction for the Coming Decade from a Global Climate Model, Science, Vol. 317, no. 5839, pp. 796-799. doi: 10.1126/science.1139540.
- Jeff R. Knight, Martin B. Andrews, Doug M. Smith, Alberto Arribas, Andrew W. Colman, Nick J. Dunstone, Rosie Eade, Leon Hermanson, Craig MacLachlan, K. Andrew Peterson, Adam A. Scaife, and Andrew Williams (2014). Predictions of climate several years ahead using an improved decadal prediction system. Journal of Climate. Vol. 27, no. 20. DOI: org/10.1175/JCLI-D-14-00069.1