This is a supplement to the story Cyclones in the Forecast.
It is a common mystery that elicits complaints and questions — why does the weather forecaster in the media sometimes get it wrong? Maybe “chance of showers” turned into a sunny day or a predicted thunderstorm rolled in with much stronger winds than predicted and unexpected hail.
Incorrect forecasts tend to be remembered better than correct ones, but weather forecasting is actually a lot more accurate than it gets credit for — and continually improving. Thanks to research developments leading to steadily increasing forecast skill, 72-hour forecasts today have the same level of accuracy that 36-hour forecasts had in 1990. But, some error inevitably remains.
“Weather is complex,” said Adam Clark, Iowa State alumnus ('04 BS, ’06 MS, ’09 Ph.D. atmospheric science) and research meteorologist at the National Severe Storms Laboratory. “Perfect weather predictions would require accurately observing numerous variables over every square inch of the atmosphere.”
The National Weather Service cannot cover every square inch — but they cover as much as they can through observations from weather stations, ocean buoys, satellites and balloons. They also log these observations over time, incorporating billions of past observations into their prediction models.
“These observations still have errors because the instruments themselves are not perfectly accurate,” Clark said. “Plus, the prediction models themselves have errors because the equations that explain atmospheric motion do not have exact solutions.”
The compounding of tiny errors can grow exponentially. This is commonly known as “chaos theory” or the “butterfly effect,” the idea that something very small in one part of the world, such as the flap of a butterfly’s wings, can have a large impact elsewhere in the world.
Research, said Clark, focuses on three key areas to continue to improve predictions:
(1) obtaining better observations and developing new ways to observe more of the Earth
(2) developing better technologies to put different observational datasets together for input to a weather prediction model
(3) improving the weather prediction models themselves
“Research is needed because there is a lot of room for improvement in our current weather models,” Clark said. “These improvements can have huge economic impacts since weather affects such a large part of the U.S. economy.”