Airlines have had to dramatically cut flight schedules due to the coronavirus pandemic, and some experts believe this has begun to hurt weather forecasting.
What?!
It turns out that forecasting models depend heavily on data collected by aircraft. The European Centre for Medium-Range Weather Forecasts (ECMWF) said this week that the number of aircraft reports received worldwide declined 42 percent from March 1 to 23. In less than a month, the number of aircraft reports over Europe received and used by the ECMWF fell 65 percent.
Weather forecasting models depend heavily on data collected by aircraft.
A 2017 American Meteorological Society study found that using aircraft observations reduced six-hour forecast errors in wind, humidity, and temperature by 15 percent to 30 percent across the United States.
This is no small matter. The more accurately experts can predict impending weather, the better prepared individuals, communities, and businesses can be. Less accurate forecasts can lead to a lack of preparation and bad weather-related decisions. From an insurance perspective, this can result in larger claims and losses.
So, late last night, worried about yet another negative implication of coronavirus, I fired off an e-mail to Triple-I non-resident scholar Phil Klotzbach. Dr. Klotzbach is a research scientist in the Department of Atmospheric Science at Colorado State University. He has published over two dozen articles in peer-reviewed journals and is quoted regularly by the Weather Channel, Forbes, The New York Times, USA Today, and The Wall Street Journal. He and his team also publish an annual forecast for the Atlantic hurricane season.
True to form – and thanks, in part, to the two-hour time difference – he responded almost immediately:
“I don't think it's going to be a huge reduction in model skill, but the ECMWF estimates that removal of all aircraft can reduce prediction ability at upper levels in the atmosphere (~30000 feet) by around 10-15% for 12-hour predictions. Subtracting aircraft-provided information from historical model forecasts increased errors by about 3% for surface pressure. The lack of aircraft data has a greater impact on shorter-term forecasts (e.g., <1 day) than it does on longer-term forecasts (e.g., 5-7 days), although some degradation of the forecasts continues even at longer-range timescales.
Of course, some aircraft will still be flying, and some of the loss may be mitigated by other data sources, such as additional launches of weather balloons.”
In other words, the reduction in aircraft data is likely to degrade accuracy of same-day and longer-term forecasts a bit, and some of that degradation will likely be offset by other data resources the forecasting community brings to bear.
Amid everything we need to be concerned about while dealing with the impacts of COVID-19, the reliability of weather forecasting isn’t yet at the top of the list.