Image source: https://news.ucar.edu.
Researches are using the same technology as in face recognition methods to create hail storm forecasting models.
The solution created by the National Center for Atmospheric Research (NCAR) and by the University Corporation for Atmospheric Research (UCAR) make it possible to forecast hail storms and their seriousness more accurately than ever. The used artificial intelligence was previously used for face recognition systems, but now the researches are taught it to recognize the ice rain from the various weather phenomenons.
With the help of the new system, even the size of the hail pellets are predictable. This was practically impossible with the technologies used by meteorologists so far. David John Gagne, the scientist of NCAR, and the leader of the research said that they accurate know that the structure of an approaching storm affects whether an hail storm is expectable or not. A super cell is more likely to create an ice rain than a storm chain. The emergence of an ice rain is depending on various meteorological factors, but even if they considered every aspect, they still could not say what will be the size of the ice pellets.
Because the computer models were not capable of dealing with the required amount of data to calculate this. But the artificial intelligence, and the machine learning within it are able to do this. The developed software calculates the structure of the storm and process the most important weather data, for example the force of wind, the temperature, and the atmospheric pressure. The forecast is based on these data, which will be matched to the features of the occurred storm. Thanks to the matching, the method will be more and more sophisticated, and the results will be more and more reliable.
