Technology vs Nature: IA to predict storms

The history of mankind has shown that man is capable of resisting what nature throws at them, by surviving natural disasters since the time of the caves. Among the worst calamities that have ever existed are the diversity of storms, cyclones and tornadoes around the world, sweeping through.

Unfortunately these always attack unexpectedly, given their complexity of prediction. However, as it was previously mentioned, man has evolved to overcome himself. In this case using Artificial Intelligence as an algorithm of a study attempts to predict such climatic attacks.

The study, published in the journal IEEE Transactions on Geoscience and Remote Sensing, shows a model that can help meteorologists to identify potential serious storms faster and more accurately.

Scientists have created a framework based on automatic learning that detects rotational movements in the clouds from satellite imagery that could otherwise go unnoticed.

“The best forecast contains as much data as possible, there is a lot to assimilate, because the atmosphere is infinitely complex. Using our models and data, we’re taking a picture of the most complete atmosphere,” said Steve Wistar, Senior Forensic Meteorologist at AccuWeather, USA”

For the purposes of the study, scientists analyzed more than 50,000 American satellite images, identified and marked the shape and movement of decimal clouds.

These cloud patterns are strongly associated with cyclone formations that can lead to severe weather events including hail, thunderstorms, strong winds and blizzards, they said.

Then, using artificial vision techniques, the researchers taught the computers to automatically recognize and detect clouds in satellite images.

Computers could then help experts by indicating in real time where in the ocean of data they could focus their attention on detecting the emergence of a serious climate.

“Because the cloud is a visual indicator of severe weather events, our plan can help meteorologists forecast such events,” said the study’s lead author, Rachel Zheng, of Penn State University in the United States.

Scientists have discovered that their method is effective in detecting comma clouds with 99% accuracy, averaging 40 seconds for prediction.

It was also able to predict 64% of severe weather events, surpassing other existing severe weather detection methods.

Media: criptotendencia.com