Coronavirus outbreak leads to less data for weather forecasters
The weather community, like all scientific groups, is certainly affected by the COVID-19 virus.
Meteorologists are studying what bearing the weather has on the pandemic, and how the epidemic may result in the degradation of weather forecasting capabilities.
A recent publication by the World Meteorological Organization (WMO) indicates the worldwide virus could decrease the quantity of weather observations that feed weather forecast models. The April 1 memo describes concerns that travel restrictions caused by COVID-19 are responsible for a decrease in the amount of data that feed weather forecast models.
A variety of instrument platforms
Computer models that generate weather forecasts are run several times per day. A plethora of data sources provide atmospheric measurements needed to initialize each model run.
Instrument platforms that feed weather forecast models can be stationed on the earth’s surface, within the earth’s atmosphere and out in space. COVID-19’s influence on weather data from a couple of platforms in the earth’s atmosphere becomes evident when assessing data streams that originate from domestic and international aircraft.
For more than two decades, the WMO has been the lead agency responsible for the Aircraft Meteorological DAta Relay program (AMDAR), in cooperation with the International Air Transport Association. The program is an example of one of many government/industry cooperative arrangements found in the world of weather.
This program utilizes aircraft communications systems to transmit meteorological data from onboard sensors. Satellites transmit inflight weather data, with precise location and time stamps, to WMO offices that aggregate and quality control each observation.
They combine the AMDAR data with observations from all other measurement platforms and use it as input to global and regional weather models. These models produce forecasts used by the public agencies, governments and industries, including the airline industry itself.
Fewer flights spells trouble
It can be safely said that model output accuracy is dependent on the quality and quantity of data that are used to calculate the initial state of the atmosphere for each model run. A June 2018 WMO report estimated that the AMDAR program provided more than 700,000 observations per day to the weather modeling community.
Since the onset of this crisis, airline travel has been disrupted and the number of daily domestic and international flights drastically reduced. Hence, there has been a decrease in the amount of weather data available to the AMDAR system.
Meteorologists are data lovers. On a hunch, I dug deep into the National Weather Service website and found details that paints a clear picture of the data loss. My data gold mine was a listing of aircraft AMDAR soundings (packets of atmospheric data transmitted during a flight) for all global airports.
These data show that AMDAR soundings have dropped by about one half since early March. The average number of weekly soundings reported between November 2019 and March 2020 was 85,000-105,000. For the week ending April 4, 2020, the number of soundings was less than 50,000.
The loss of model input data volume is not the only issue. The same database also lists the number of airports where incoming flights provide soundings. Prior to March, nearly 600 airports contributed to the AMDAR database. By the first week of April, that number dropped by about 200.
In summary, the loss of AMDAR data along with decreased areal representativeness of the soundings have potentially resulted in increased forecast model errors.
Since many meteorologists are one part physical scientist, one part data junkie and one part statistics geek, I’m sure that several post-pandemic studies will attempt to quantify forecast model error trends due to the COVID-19 pandemic effect on AMDAR data.
When it’s over …
Woody Whitlatch is a meteorologist retired from PG&E. Email your questions or comments to
clayton_909@yahoo.com