If the weather forecast seems a bit more unreliable lately, the coronavirus might be partly to blame.
Air travel was one of the most significantly impacted sectors when the world locked down in March. According to a new study, that’s having an impact on the accuracy of forecasts globally.
When air traffic dropped, “the accuracy of surface meteorology forecast in March-May 2020 decreased remarkably,” according to Dr. Ying Chen, a senior research associate at Lancaster University’s Environment Centre.
Chen examined weather forecasts from these months and compared them to actual weather observed during the same time frame. The researcher found the accuracy in short-term forecasts for temperature, pressure and wind speed deteriorated through these months, sometimes more significantly in one place over another. This finding came as forecast accuracy was expected to improve in 2020, he added.
“Every year forecasts have been improving. Models have been getting better and better,” said Anthony Farnell, Global News’ chief meteorologist.
“This is the first year where it’s actually taking a step back. We’re going back basically five years as far as where our forecasting skills are. We haven’t had this in decades.”
What’s happening?
The data that helps meteorologists create forecasts come from a variety of tools.
Data is collected by satellites, buoys in open water, weather balloons, scattered weather stations and radar. It also comes from aircraft and cruise ships — two things the pandemic has reduced.
The computer models rely heavily on aircraft data, Farnell said. Without it, measurements for temperature, humidity, wind speed in areas not densely populated — like the Artic — are hard to collect.
“Flight data is definitely high on the scale of importance,” Farnell said.
“The more data you have the better your computer models are going to be. So, weather balloons are launched daily, but those are only from a few set locations. You can’t replace aircraft data with balloon data. That’s the problem. When you lose those planes, you lose a lot of those numbers.”
Commercial air traffic shrunk significantly during the March-May period examined in the study.
“It’s about March 11 when the bottom really started to fall out,” said Ian Petchenik from Flightradar24, an online global flight tracking service.
While there was a “slight downtrend” in the last week of January when China started to pull back on air traffic, February and the beginning of March “didn’t see much change in anything.”
FlightRadar’s data shows commercial flights were down merely 3.9 per cent in the first week of March compared to the same time in 2019. By the last week of the month, it had dropped to 60.9 per cent.
April and May saw the most considerable drops. In April, commercial flight totals shrunk 73.7 per cent compared to 2019, and 70.8 per cent in May.
Changes were also seen in flight routes, said Petchenik, as flights that were operating could take advantage of more “direct, economical routing” due to the lack of traffic.
Farnell said that while he’s seen the effects, it’s hard to pinpoint blame.
“Over the course of spring, I did see some computer models that were out of whack, especially in the tail-end of our seven-day forecast. Days four through seven and beyond, we started to see some big variations,” he said.
“But it’s tough to quantify whether a forecast missed the mark because of that or something else.”
Long-term impacts
Remote areas saw the biggest issues with forecast accuracy, according to the study.
Flight data are far more crucial in areas like Greenland and Siberia, Chen notes, as other conventional measurement tools are “very limited.”
For similar reasons, the loss in data is also more substantial for the northern hemisphere than the southern.
Those concerns were echoed by the World Meteorological Organization (WMO). In May, it estimated that measurements taken from aircraft plummeted by an average of 75 to 80 per cent compared to normal. In the southern hemisphere, the loss is closer to 90 per cent, it said.
While not detailed in the study, Canada will also feel the brunt of the data loss, said Farnell.
“We’re a smaller population. We’re scattered over such a large landmass and we don’t have weather stations everywhere. Plus, we’re a popular flight path. A lot of flights go up and over Canada’s far north as they’re travelling around the planet,” he said.
“Nobody’s flying over Greenland or Canada’s far north anymore. We don’t have any other data. That’s a big gap.”
The concerns are greater now as the U.S. and others head into peak hurricane season.
With flight traffic still at unusually low levels, the data that makes up hurricane forecast models will take a hit. Farnell said there are plans to increase the number of hurricane hunter aircraft, or surveillance missions, to try and populate computer models with valuable data.
“It’s expected to be a very active hurricane season,” he said. “So the hope is that will make up for the lack of aircraft data this summer.”
But improvements could be on the horizon.
As many countries began to reopen, air traffic density began to climb, said Petchenik with Flightradar24. It was slow through May, he said, but sped up through June as airlines increased total available flights.
“Think of it like different lengths of a board at different inclines. For each month, we see a similar ramp-up as airlines add a significant number of flights at the beginning of each month,” he said. “That was the same for May and June, they reconfigured their schedules, added a set number of flights and we saw that increase play out throughout each month.”
So far in July, FlightRadar has seen a plateau in both total flights and commercial flights globally.
Petchenik believes that’s attributable to a “pullback” in North America, particularly in the U.S., “as people in the U.S. aren’t really allowed to go anywhere.”
Whether the gap in data has long-term implications, as Chen’s research suggests, is hard to be certain about, said Farnell.
“Right now, with the uptick in aircraft, things are looking better, and that may not have a detrimental effect on some of these longer-term models,” he said.
“But it’s definitely possible. It all depends. Will there be a second wave? How long will this go on for?”
Comments