As facial recognition technology becomes more popular on smartphones in attempts to eradicate passwords, U.S. scientists have found a way to use this same AI tech to predict hailstorms and their severity.
A new study published to Monthly Weather Review details how scientists have trained a deep learning model to recognise features of individual storms that lead to the formation of hail and how large the hailstones will be.
The learning model is known as a convolutional neural network and the research comes from the U.S. National Center for Atmospheric Research (NCAR). This technology takes into account a storm’s entire structure, which has previously been difficult to do with existing weather forecasting techniques.
There is a multitude of factors which contribute to the formation of hail. For example, the air needs to be humid when close to the surface, but dry higher up. The freezing levels within the clouds needs to be closer to the ground and strong updrafts that keep the hail aloft to grow large are also essential. Even when these conditions are met, the size of hailstones can vary dramatically.
The team of scientists, led by David John Gagne, input simulated images of storms, combined with information relating to temperature, pressure, wind speed and direction, coupled with simulations of the kind of hail produced by a combination of these factors. This then trained the AI to identify storm features which correlated with the hail and what size it would be.
The findings identified that storms with a lower-than-average pressure near the surface of the land were more likely to produce severe hail when coupled with higher-than-average pressure near the top of the storm.
Storms with more of a circular shape, and storms with winds blowing from the southeast close to the land surface and from the west at the top were also more likely to produce hail.
This AI technology is able to ingest large amounts of data, search for patterns and teach itself which storm features are crucial to accurately predict hail, as well as how much damage said hail can cause. It differs from the existing weather technology which is limited as it cannot distinguish the broader form and structure of the storm.
Gagne said: “I think this new method has a lot of promise to help forecasters better predict a weather phenomenon capable of causing severe damage,” Gagne said. “We are excited to continue testing and refining the model with observations of real storms.”
Source: Silicon Republic
