US meteorologists apply machine learning to detect fog
Fog and low stratus clouds over airports can create dangerous travel conditions that lead to costly delays and disrupted plans.
Offices of the U.S. National Weather Service monitor and issue warnings when conditions are favorable for fog and low cloud formation. Airlines use these warnings to anticipate conditions, prevent delays and reroute flights if necessary.
Now the NWS (National Weather Service) is using new fog detection software developed by scientists at the Cooperative Institute for Meteorological Satellite Studies at the University of Wisconsin-Madison and NOAA (National Oceanic and Atmospheric Administration) to help with these warnings.
The software, known as a satellite product, is the result of more than a decade of collaboration between CIMSS and NOAA. It uses machine learning methods with near-real-time data from weather satellites, such as NOAA's GOES-East and GOES-West, to monitor conditions around the clock and issue alerts about possible fog.
“As of late 2020, the GOES-R Fog and Low Stratus product was fully operational in NWS offices across the country. The product provides round-the-clock monitoring of hazardous fog conditions from Utqiagvik (Alaska) to Key West (Florida)”
Fog forms when a cloud forms at ground level, reducing visibility to less than 1 kilometer. While fog can be easy to spot during a morning commute or by looking out the window, determining the size of a fog patch over a large area remains a challenge, especially for satellite observations.
Earlier fog detection systems compared satellite data in visible and infrared bands, but were largely limited to conditions where there were no other clouds above the fog. In addition, previous methods were limited to nighttime use due to interference from reflected sunlight. And unlike thunderstorms, which contain larger raindrops and ice crystals, fog consists of much smaller droplets that are not visible on radar.
“To improve the system, we incorporated other atmospheric data used in numerical weather prediction algorithms. This helps determine the type of fog over an area by evaluating each image pixel and generating a probability of fog presence and its intensity”
Beyond detecting the presence of fog, this product helps airlines avoid delays that cost millions of dollars each year. According to an analysis by the Federal Aviation Administration, weather caused nearly 70% of flight delays in 2008–2013, resulting in real costs. Each hour of delay costs airlines between $1,000 and $4,000 per flight.
The GOES-R Fog product was developed based on the needs of the aviation industry, which has different requirements depending on the aircraft. Small planes, for example, rely on a different set of rules when it comes to weather, because they lack the advanced equipment found on larger airliners. Fog identifies hazardous conditions for both types of aircraft and can be used to issue warnings nationwide, avoiding costly delays and making air travel safer.
“Travelers may not know the origin of this product or what happens behind the scenes. But it is a great example of how the public benefits from this kind of collaboration between universities and the government”
While weather is the main cause of delays due to excessive demand on affected resources, volume caused by excessive demand even with unlimited resource capacity also accounts for 19% of delays. Equipment failures account for 1%, runway unavailability for 6%, and "other" factors for the remaining 5% of delays. These delay statistics include air carriers, air taxis, general aviation and military aircraft classes.
The type of weather causing air traffic delays varies throughout the year and also depends on the geographic region of the country. The peak of weather delays and total delays occurs in May, June and July (weather delays range from nearly 6,700 to more than 7,800, total delays range from 9,000 to 9,500), but weather delays are also significant in March, April, October, November and December (from 3,500 to 5,400).
NextGen Weather (a Federal Aviation Administration program) provides aviation weather products that support tactical and strategic air traffic management during weather events, helping minimize passenger delays and improve flight safety.
Traffic managers can take various steps, such as locally rerouting traffic depending on weather conditions, directing airborne traffic to a weather-free arrival path when it is near the destination airport, and holding arrivals in stacks until the weather clears.
