Lightning is one of the most destructive natural disasters and the leading cause of weather-related fatalities. The Centers for Disease Control and Prevention (CDC) said that 1 in 500,000 people gets struck by lightning every year, depending on the risk factors at play.

However, accurately predicting lightning is hard, so there are still casualties every year. A new study from the researchers at the University of Washington showed that artificial intelligence (AI), which uses computer algorithms to improve themselves, can enhance forecasting of lightning to improve safety warnings and develop more accurate long-range climate models.

TOPSHOT-URUGUAY-WEATHER-STORM-LIGHTNING BOLT
(Photo: Getty Images)
TOPSHOT - A lightning bolt strikes near the Uruguayan Yacht Club during a thunderstorm in Montevideo early on November 16, 2021. (Photo by Mariana SUAREZ / AFP) (Photo by MARIANA SUAREZ/AFP via Getty Images)


Improving Lightning Forecast Using AI

In their study, titled "Improving Data-Driven Global Weather Prediction Using Deep Convolutional Neural Networks on a Cubed Sphere," published in the Journal of Advances in Modeling Earth Systems, researchers created a global weather model based on the four-decade weather data rather than on detailed physics calculations.

According to the press release of the university, researchers from UW collaborated with Microsoft Research to show how AI can analyze past weather patterns to predict future weather. The simple, AI-based technology can simulate a year's global weather quicker and almost as accurately as traditional weather models.

Study lead author Jonathan Weyn said that machine learning is essential in pattern recognition because it sees and recognizes a typical pattern and identifies how it usually evolves. From there, it decides the best course of action based on the examples seen in the past 40 years.

The team trained the AI with the lightning data from 2010 to 2016 to let the technology discover relationships between variables that affect the weather and lightning strokes, Phys.org reported. Then they tested the technology on predicting the weather from 2017 to 2018 by comparing it to an existing physics-based method.

It showed that the AI could forecast lightning with the same skill two days earlier than the current leading techniques because it was trained to analyze weather patterns in the US. However, it is less accurate in predicting weather patterns in other places where lightning is less common since it is only trained in one area.

ALSO READ: Six Tools Meteorologists Use to Predict the Weather


AI is Becoming an Important Tool in Weather Prediction

Researchers said that machine learning requires a lot of data to predict lightning and other weather events. A few years ago, this technology will not be possible because of the lack of necessary data. As machine learning devices advance, accurate and reliable lightning observations are necessary for correct weather forecasting.

According to an article in Hewlett Packard Enterprise, scientists at NCAR, National Oceanic and Atmospheric Administration (NOAA), and the National Weather Service have been using AI to improve forecasts and provide more accurate, long-range information about the climate, ocean, and space weather.

The accuracy of weather forecasts and warnings depends on smart algorithms and supercomputers that involve managing, analyzing, and visualizing the data that has depended on AI for many years now. Experts said that the combination of physics-based modeling and AI would enable better predictions.

RELATED ARTICLE: Space Weather is Now Easier to Predict, and New Australian Satellite is Here to Help

Check out more news and information on Artificial Intelligence in Science Times.