At Heathrow Airport, for example, snow arrived within ten minutes of when Met Office forecasters had predicted – giving vital guidance for those managing the situation.

This level of forecasting accuracy is far from easy to achieve, however. Snow is an example of a small-scale weather feature, affected by a number of variables and notoriously difficult to forecast.

The Met Office is using cutting-edge developments to improve the accuracy of forecasts in these challenging situations which deal with so-called ‘small scale’ weather.

This includes things like intense rain showers or thunderstorms – which can be just a few hundred metres across, or weather which depends on fine details of the land surface, such as snow or valley fog.

These types of weather can be very difficult to represent in forecasting models, which are the computer generated simulations of what the atmosphere – and weather – will do next.

Because the weather at a particular location is influenced by much larger scale weather patterns, models need to be run on a global scale even just to forecast for the UK.

Forecast models require a very large number of calculations and, with the computing power available, the global model the Met office runs uses a grid-scale of 25km (i.e. every grid-box is 25km x 25km).

At this resolution, large scale weather patterns will be well reproduced but the model will be unable to capture the detail of small scale weather. To tackle this, the Met Office has developed UKV.

This involves running a version of the model which focuses on the UK, allowing a much smaller 1.5km scale to be used. Information is fed in to the edges of UKV from the 25km global model.

The 1.5km grid-boxes enable UKV to capture things like snow much better, leading to improved forecasts in many situations.

In most situations, even with a 1.5km grid, current science and technology does not enable the prediction of the exact location and timing of each shower that passes over the UK. However, the increased detail gives a better indication of the character of the weather and could be useful for giving probabilistic forecasts – which give the chances of, for example, rainfall in a given place at a given time.

As well as the the 1.5km weather model helping with our forecasts in the last couple of days, they also helped with the accuracy of our snow forecasts in the very cold and snowy weather at the end of 2010. Back in November of that year, we saw numerous heavy snow-showers being carried inland from the sea in a NE wind caused significant disruption in the north east of England. The picture below shows that for the coarser 12 km model (NAE) showers stall over the coast causing a major underestimate of snow inland. This is a well known problem with models of this grid length. In contrast, the UKV is able to represent the showers more realistically and brings the showers inland, producing a much better forecast. The UKV better represents what actually happened as shown by the radar image to the left.

24 hour accumulations for 25 Nov 2010 from UKV and 12 km (NAE) models compared to that actually observed by radar. This shows an example of the advantages of a high resolution models

We are continuing to look at ways of even further improving the accuracy and detail of our forecasts. You can find a more in-depth article about UKV in our Research News section.