Developing the science for long range weather forecasts

Photo courtesy of NASA, located at Visible Earth (http://visibleearth.nasa.gov/)

7 November 2013 - There's often a lot of talk in the media about long-range weather forecasts predicting summer heat waves or record winter snowfall, but what's the scientific truth behind these detailed predictions for the weather months ahead?

Here we look at why providing detailed forecasts even out to five days ahead is a huge scientific challenge, the complexities of looking at detail further ahead, and the work the Met Office is undertaking to improve longer-range weather outlooks.

Modern day forecasting out to 5-days

The processes that determine the weather we get are a classic example of what is known as a chaotic system - which means tiny factors today could have a huge impact on the weather to come.

One scientist put it like this: 'one flap of a seagull's wings could change the course of weather forever'. These days the idea is more poetically referred to as the 'butterfly effect'.

Forecasting weather involves predicting future variations in this complex system, so it's no surprise that it is a huge undertaking.

Millions of observations of what is happening now are fed into supercomputers capable of trillions of calculations a second. These computers feed the observations into a virtual 'model' of the atmosphere based on the laws of physics - made up of more than a million lines of computer code which runs forward in time to produce predictions of the weather we expect to see.

In a chaotic system, the further ahead you try to predict, the less likely it is that your forecast for the detail in the weather (where clouds will be, where rain will fall, etc) will be accurate.

Advances in technology and understanding have helped us to meet these challenges to deliver huge improvements in forecasting accuracy over recent decades.

For example, today's Met Office forecasts out to four days are as accurate as our one-day forecasts were 30 years ago. You can see more about the accuracy of our detailed five-day forecasts on our web pages.

Looking further ahead

While there has been a huge amount of progress in long range forecasting, atmospheric chaos means it is not possible to give accurate definitive predictions for the UK weather months ahead - for example saying how much snow might fall in a particular week several weeks ahead. Speculative forecasts appearing in the media and claiming to accurately and definitively forecast UK weather months ahead are therefore doomed to failure when analysed over a long period of time.

However, there is the potential to develop longer range forecasting. The Met Office is developing experimental forecasts out to months and years ahead. These show promise in predicting the risk of different seasonal average conditions a few months ahead, particularly in winter. For example it may be possible to give the risk of harsh winter overall from some weeks in advance.

To do this we have to add even more complexity than is handled by weather forecasts. We now have to take into account not only the current state of the atmosphere used to make weather forecasts, but also other factors which could affect UK climate such as the current state of the global oceans.

Ocean conditions in the North Atlantic and even the Pacific Ocean can influence British winters.  Although the Pacific may seem remote, our scientists have uncovered clear links between the El Nino/La Nina cycle and winter weather across northern Europe. Our latest computer models, based on laws of physics, accurately reproduce these and other links important for long range forecasting.

Risk Assessment

These experimental systems will still not be able to predict detailed weather on a given day or week months ahead - instead they predict levels of risk. For example, you could assess the level of risk of an extreme winter. 

For example, if we predict a 75% chance of a severe winter then on three out of four occasions we make such a forecast then we should indeed see a cold winter.  Conversely on one in four cases we should not. With that in mind, you'd need to look at these forecasts over a long period of time to assess their accuracy.

The risk approach is agreed upon by international experts in the field. The Met Office is working with other national weather forecast agencies and the World Meteorological Organisation to support an exchange of long range forecasts of risk for many regions across the globe. 

Over time, it's hoped that the Met Office - working with research institutes around the world - can bring about large improvements in long range forecasting as we have seen in our short-range predictions.

Last updated: 8 November 2013