How are weather forecasts created?

There is no better time to look deeper into meteorology than on British Science Week. Starting with the world of forecasts.

Meteorologists use some of the world’s most powerful super-computers to produce our weather forecasts.

These incorporate a vast amount of data and perform trillions of calculations per second to give us an idea of how the weather will change in the coming days.

Due to the complexity of the atmosphere however, even these weather forecasts tend to only be reliable for up to five days or so.

If the weather conditions are particularly complex, there can still be a degree of error even within the five-day period.

Any forecast that offers an outlook greater than ten days’ time is not considered a forecast but a trend.

To produce these forecasts, meteorologists usually run their forecast models a number of times and group the forecasts into what are known as clusters.

This gives probabilities of what may happen in the future, rather than one single solution.

Short-term forecasts cover a period of 24 hours up to three days. We see these on the TV every day. They are generated using synoptic meteorology – current atmospheric observations – and numerical models.

Medium-range forecasts cover between three and 10 days with long-term forecasts covering periods over 10 days away. Both are generated by using weather ensemble models to create a weather trend.

For both ranges, conditions tend to be described as wetter or drier, warmer or cooler with no specific information.