Matt Huddleston explains how the Met Office has staked a claim to tropical storm forecasting

The key to the forecasting crystal ball until now has been being able to use the past to predict the future. Traditional techniques that use statistical correlations, analogues to current conditions and subjective forecaster insight are under pressure as our climate changes though. The global community has also not reached consensus as to the impact global warming will have on the world’s climate.

Climate models, which are based on the underlying physics of the climate and incorporate weather forecasting technology, are the new kid on the block. Some predictions use complete coupled ocean-land-atmosphere models cast dozens of times into the future to sample the risks. An estimate of the state of the complete climate system is generated taking both satellite data and robotic observations of the ocean’s depths and used to initialise the models.

The technology is new and somewhat limited in detail at the moment and even the scientists were surprised that the climate models can resolve tropical storms at all. But resolve them they can – and there are now high hopes of developing risk-based products across a range of timescales given that forecasts of tropical storm numbers have proven significantly better than statistical competitors

“Climate models, which are based on the underlying physics of the climate and incorporate weather forecasting technology, are the new kid on the block

Matt Huddleston Principal consultant on climate change, Met Office

The Met Office has performed well at risk-based forecasts in the past. Global weather forecasts in 2007 predicted the formation of sub-tropical storm Andrea in May 2007 and predicted Hurricane Katrina landfall a full 12 hours ahead of the competition.

For the 2007 season, we predict the North Atlantic is likely to see ten tropical storms. The figure is lower that the 1990-2005 long-term average of 12.4 storms and includes a 70% chance that the number of storms will be in the range of seven to 13.

The Met Office Hadley Centre for Climate Prediction has also developed new dynamical models for the Inter-governmental Panel on Climate Change process, and these are now being tested on timescales from seasonal to decadal ranges – all with a risk-based probabilistic and verifiable approach. The models in research for future systems already include new areas like dust production – a factor thought to have dampened the 2006 season.