Pilots routinely fly over remote stretches of ocean with almost no weather information, apart from a pre-flight briefing and updates every four hours. Yet this is where some of the worst turbulence and, as Air France Flight 447 tragically encountered on June 1, most intense weather occurs. The National Centre for Atmospheric Research (NCAR) is developing a new system that combines satellite data, weather models and A.I. to identify and predict rapidly evolving storms, enabling pilots to avoid areas of potentially severe turbulence.
The existing problem for pilots is twofold. First, weather satellites are the only source of information over these remote stretches and, even then, provide less frequent images than over land. Secondly, thunderstorms over water can develop and move very quickly, rendering briefings obsolete. Onboard radars are able to detect clouds and precipitation, but can provide no clue as to the location of turbulence, which may be far from the most intense precipitation
The prototype NCAR system, however, will eventually provide pilots and ground-based controllers with real-time maps of turbulence at various altitudes. Based on existing products which NCAR has developed to alert pilots and controllers about storm activity over the continental U.S., the new system will also incorporate global mapping of clear air turbulence and an artificial intelligence technique known as “random forests” for short-term turbulence forecasts.
Developed with funding from NASA, the NCAR system is on-track for testing next year on selected transoceanic routes. Feedback will allow further adjustments to the system, and then the system should be finalized in about two years.