How ArabiaWeather battles the forces of nature
Residents of the Middle East don't need me to tell them that this year so far has been a doozy in terms of weather.
First, we had Zina:
Lebanon after storm Zina. 8 people died in freezing temperatures, and thousands of Syria... http://t.co/fshGeQK4uO pic.twitter.com/vfRjsHt7Of— British Red Cross (@BritishRedCross) January 27, 2015
Then, we had Yohan, which shut down the Suez Canal for at least a day:
Waves from Storm Yohan crash on the beach of Saida, Lebanon (Image via The Daily Star).
Then, half the Sahara desert, apparently, landed in Cairo, Amman, Jerusalem, and Beirut.
— Tamer El-Ghobashy (@TamerELG) February 11, 2015
As a latest injustice, Mother Nature sent Storm Windy to dash our hopes of an early spring.
While we, bundled up and crowding around our space heaters, forget what it was like to bask in the sun, the weather professionals at ArabiaWeather have had a busy start to the year, to say the least. But busy doesn’t have to mean chaotic, as is apparent by the focused, analytic approach of Chief Product Officer Yousef Wadi, obvious even over a Skype call.
ArabiaWeather’s products range from weather forecasting for users throughout the region on desktop and mobile to specialized packages for corporate customers like Royal Jordanian airlines and Al Arabiya television station; they rely on a carefully balanced system with five main inputs, says Wadi, that can predict weather events up to a month in advance.
These include weather algorithms, or models, as they’re called in the weather business, from the US, and Europe and the UK, as well as open source algorithms they’ve developed themselves that pertain specifically to the Middle East “based on the learnings of our meteorologists around the region.” These include algorithms that predict dust storms like the one that engulfed the Levant last month that are a particularity of the Middle East. The various models the team uses have been assigned weights based on their individual records of accuracy.
The second input is data from weather stations (pictured left), of which ArabiaWeather has nearly 200 in the GCC and Levant, where solar-powered sensors track wind speed, wind direction, volume of rain, temperature, and other metrics. The team also gathers information from weather stations in Saudi (which comprises their major market share) whose information is publicly available, Wadi says.
Third, the meteorological team culls raw images from the company’s “massive” satellite coverage from EUMETSAT, a global satellite agency, on which the team then does its own image processing. Next, they collect lightning strike data region-wide, which “gives an indicator of the intensity of the storm.”
If all this fails, the team relies on “human intelligence,” meaning actual people stationed throughout the region whom the team can call up and ask about the weather.
Because of the extremely precise and measured predictions of both ArabiaWeather’s weather models and the know-how of its meteorologists, the frequency of extreme weather events so far this year haven’t fazed the team. In other words, there are always potential extreme weather patterns on the horizon; it’s just that this time, the potential extreme weather events actually materialized in a way that was unusual for the region.
So how does the team translate the raw data to something its customers can use?
Once the probability of an extreme weather pattern reaches a certain level, the trick then becomes how to communicate the weather patterns to users and clients. “The second a weather pattern moves into a five-day signal, meteorologists and the weather team look for errors,” and add a human corrective element to algorithm predictions based on their own experience. “At this point [in the weather cycle], human intelligence is better” than algorithms, Wadi says.
Case in point: Storm Huda, the first extreme weather pattern in the region this year, was originally predicted to be a “light snow,” Wadi says. “Only Mohammed [al-Shaker, ArabiaWeather’s founder and CEO] predicted heavy snow based on what he's seen in past storms,” says Wadi. Huda (or Zina, as the storm was named in Lebanon) went on to wreak havoc with snow, rain, hail, and wind in Lebanon, Syria, Jordan, Egypt, Palestine, and Saudi, and ArabiaWeather users were properly warned.
Using the predictions compiled from algorithms and human intelligence, the weather team takes over, writing articles (the seven person editorial and weather content teams produce 24 articles per day), producing videos, and organizing mobile push and email notifications to users of the Arabic language app who have opted in. The teams are also invited to discuss the weather forecast on regional television stations SkyNews Arabia and Al Arabiya. ArabiaWeather’s product for its corporate clients is a bit different, depending on each client’s reasons for needing to know about the weather (we'll cover more of this in a subsequent article).
While the team doesn’t release weather data for public consumption more than two weeks in advance, Wadi gave us a tiny scoop as to an extreme weather event that might materialize somewhere in the Middle East. All I’ll say is: watch out for March 7th.