AI Capable of Predicting Deadly Storms Days In Advance

With our understanding of artificial intelligence growing everyday we have begun to understand not only its benefits but also its applications in different aspects. Other than the usual application of machine learning algorithms in smartphones and other portable devices, we have managed to utilise artificial intelligence in a way that not only makes peoples lives easier but also saves them. In this day and age meteorologists have many tools, equipment and technologies at their disposal to help them detect and predict sever weather conditions in advance. However, despite the tools at hand, there are far too many variables that have to be taken into account in order to accurately predict the occurrence of thunderstorms, hurricanes, blizzards, etc. The result of our limited understanding of these natural occurrences means that when these deadly storms to occur we end up with many casualties, injuries, property damage and costs, etc.

To solve this problem, a research project set up by scientists at AccuWeather, Pennsylvania State University was launched with the end goal of successfully being able to develop an AI model capable of predicting storms more efficiently and to a more accurate degree.

The program can identify 40 clouds per second with a success rate of 99%

The program can identify 40 clouds per second with a success rate of 99%

The research team successfully built the AI program to look for a single sign of an early-stage of a storm, known as comma-shaped clouds. These comma-shaped clouds are strong indicators that the occurrence of a cyclone is likely in the near future. In order to train the artificially intelligent system, the research team took a sample of over 50,000 satellite images taken from a variety of different atmospheric conditions. By combining this data set and incorporating it within the machine learning algorithm, the AI was successfully able to detect these comma-shaped clouds.

The AI program is also capable of notifying meteorologists of danger zones. This could allow meteorologists to make efficient decisions, cut to the chase and focus their attention on an approaching cyclone which could potentialy pose a threat. The researchers managed find the model that could can pick out comma-shaped clouds with a success rate of 99 percent with a speed of 40 seconds per clouds identified. This ensures that the AI program is designed to be constantly on the lookout for any dangerous cyclones by identifying a large number of clouds in a small time interval.

With artificial intelligence proving its usefulness in areas where peoples lives are on the line, the possibilities for AI seems to be endless. We may see a day where we are able to successfully predict to an accurate degree the specific weather conditions of the coming weeks based on advanced machine learning algorithms which could take into account a number of factors and deduce a trend.

Zacharia Sharif