Weather forecasting has become more accurate than ever, yet mobile apps can still show wildly different temperatures. This discrepancy was highlighted during a spring bank holiday in Bournemouth, where the BBC weather app forecast rain and thunderstorms, but the actual weather turned out to be sunny with blue skies. The incident sparked media outrage, with some outlets calling it a "blunderstorm" and criticising the forecast as "crap forecasting" and "total incompetence."
The foundation of modern weather forecasting involves gathering vast amounts of data on atmospheric conditions, such as temperature, humidity, and wind. Gaps in data are filled by extrapolating from observations and past forecasts. Forecast models use sets of equations to calculate future conditions. The number of weather observations has increased dramatically, with improvements like the Met Office integrating wind-speed data from aircraft transponders and Nasa's GOES-16 satellite providing higher-resolution scans.
Supercomputers capable of a thousand trillion calculations per second process this data, running complex models that approximate atmospheric processes. "Ensemble forecasting" runs models multiple times with slightly different starting data to explore probabilities. This has led to significant improvements: a 2015 study found that the ability to predict atmospheric pressures three to 10 days ahead has improved by about one day per decade since 1981. The Met Office says its four-day air pressure forecasts are now as accurate as its one-day forecasts were 30 years ago.
Despite these advances, smartphone apps can still disagree. There are 8,000 weather apps for Android and 2,400 for iPhone. Measuring forecasting accuracy is complex, as different providers perform better on different measures, timeframes, or regions. The key takeaway is that while overall accuracy is improving, variations between apps stem from differences in data sources, models, and interpretation methods.
