Crucial weather observations supplied by commercial aircraft have been dramatically cut after the grounding of most of global fleet. There are just a quarter of the average 800,000 daily readings of temperature, wind strength and direction being sent automatically from aircraft during flight.
The impact on weather forecasts is “significant”, says the Geneva-based World Meteorological Organisation (WMO). “Aircraft contribute a lot to the accuracy of forecasting of most weather systems and phenomena, including forecasts associated with systems like hurricanes,” says WMO Scientific Officer Dean Lockett.
Lockett coordinates international activities of the WMO Aircraft Meteorological Data Relay (AMDAR) system, which supplies weather data to the organization’s Global Observing System. It is the backbone for all weather and climate related information for the 193 WMO member states.
“Even though aircraft fly around hurricanes to avoid the worst of the weather, the data they collect is still critical because it feeds into forecasts for the intensity and future track of such systems,” he said. “It’s been shown that this data is very important, affecting computer-based modelling and the prediction of those types of systems.”
The Atlantic hurricane season runs from 1 June to 30 November, with the eastern Pacific season covering 15 May to 3 November.
Information from commercial planes is one of several sources that meteorological bodies use to predict the weather. Altogether millions of readings from aircraft, satellites, ships, ground stations and weather balloons go into modelling Numerical Weather Prediction (NWP) systems that provide the basis of global weather forecasts.
In normal times observations from aircraft increase the accuracy of weather forecasts about 10%. Without that information the model is significantly degraded, says Lockett.
The WMO is still assessing exactly how seriously the loss of data will be to the accuracy of weather modelling. Through AMDAR about 250 million observations a year are fed into computer models at the US National Weather Service. As at the end of March, data provided by US aircraft had dropped by half.
At the European Centre of Medium Range Weather Forecasts (ECMRWF), readings provided by aircraft across the continent has been cut by 80%.
The center estimates that if all data provided by aircraft was cut off the accuracy of weather forecasts would fall about 15%.
Airlines themselves also lose out. Information on wind strength and direction is used to calculate wind shear, the location and strength of the jet stream and conditions that can cause icing. It is used by pilots to plot the quickest and safest route around bad weather.
Less than 18 months ago, IATA launched the Turbulence Aware program using turbulence measurements from aircraft to improve the safety for crew and passengers and optimize fuel burn. Airlines use these reports collected and disseminated by IATA in real-time to make operational decisions about turbulence mitigation.
The program has gained strong industry support with 35 airlines participating in the operational pilot in 2019. The program became fully operational in 2020.
Meanwhile, IATA is also working with the WMO to expand the global coverage of AMDAR. Currently 43 airlines support the system. On average, an aircraft will send around 100 or more observations per flight, most coming during the ascent and descent.
“We are trying to capture a snapshot of the atmosphere in the vertical and the first five to six kilometers are particularly important,’ says Lockett. “And the more readings we have the more accurate the modelling is.”
The information is particularly important for a 24-hour snapshot of the weather but the WMO is also building a database that could be used in climate change modelling.
But for now, there will be significant hole in both short and long-term readings as COVID-19 shuts down the global airline industry. More concerning is that it could take years before the industry returns to its pre-coronavirus levels and same level of accuracy in weather forecasting.