Get ready for a smart weather forecast

  •  AI, AR and diverse data sources are changing how weather information is processed and consumed
  • Gathering large amounts of data, noticing patterns in the observations and then presenting all of this in real-time is not an easy task. This is where AI now comes in

Since its inception in 1875, the India Meteorological Department has seen a great deal of change in its use of technology for Met observations, weather services and forecasting—going from using weather telegrams for collecting observational data and sending warnings to having its own geostationary satellite for continuous weather monitoring.

In fact, weather forecasting all over the world has undergone a massive transformation. Predicting the weather can be a tricky affair. Data sets from numerous weather satellites and sensors are collected every day. Gathering such large amounts of data, noticing patterns in the observations and then presenting all of this in real-time is not an easy task.

This is where Artificial Intelligence (AI) now comes in. The 2018 World Economic Forum report, “Harnessing Artificial Intelligence For The Earth", mentions that a new field of “climate informatics is already blossoming". In 2016, IBM, one of the biggest names in AI solutions, purchased The Weather Company, a forecasting and information technology company.

In January, the company announced a new forecasting system that will provide accurate and more finely-tuned weather forecasts: the global high-resolution atmospheric forecasting system, or GRAF. This will be the first hourly-updating weather system that will be able to predict something as small as a thunderstorm anywhere on the planet. Compared with existing models, it will provide a nearly 200% improvement in forecasting image resolution worldwide (from 12km to 3km). “Delhi is the perfect example of a tropical zone.... It is really difficult to predict the weather," says Himanshu Goyal, India business leader, The Weather Company. “If we do a 15x15 km forecast in Delhi, we’ll say the conditions are cloudy and there might be rain in some parts of the city. But with the 3km model, we can come back and say, in Delhi, we believe there’s a possibility of rain in a specific area, while other parts of the city won’t get any rain. So, for citizens and some businesses, this becomes key information."

Screen grabs of an August 2018 monsoon day in India, shown at left by the standard current weather model that operates at 13km resolution. Photo: IBM
Screen grabs of an August 2018 monsoon day in India, shown at left by the standard current weather model that operates at 13km resolution. Photo: IBM

The service will be available later this year and its benefits will trickle down to consumers through The Weather Channel app. “Weather is the only problem that modern-day stats and technology have not been able to solve 100%. The key in using weather data and prediction is all about three ingredients," adds Goyal. “The first thing is the kind of observations and their frequency. The second is physics —how much physics (physical) capability you have to start churning these data sets for valid information. This relates to science and modelling. The third is the power of computing you possess. Can your models take this data, pass it through their system and provide an output as soon as possible?" he says.

The availability of a wide array of sources for climate data has also helped. Apart from historical weather data, forecasting services now get information from meteorological stations, a network of personal weather stations, satellites, radars, flights and even smartphones, which record barometric pressure measurements through different apps.

“We have sort of tried to reimagine the app (The Weather Channel), especially for markets like India. Given the capability of analytics today, we started to look at what is the most convenient interface people would like to see and what is the weather information people will want to consume," explains Goyal. Apart from daily weather updates, the app also gives you the option of viewing a mosquito and allergy index, among other things. “With Watson’s (IBM’s AI-powered computing system) capability, we tried to look at the reasons which can cause users to take small personal decisions—everything from preparing for fog to carrying an umbrella if it’s raining."

In order to make forecasts more personalized, some weather companies have even leveraged the power of augmented reality (AR). In December, Vuzix, a US-based provider of smart glasses and AR, entered into a partnership with AccuWeather to bring interactive AR-based weather experiences to their smart glass users across the world. This technology—which was demonstrated at the January edition of CES, an annual trade show for consumer technology, in Los Angeles —will let individuals view localized weather forecasts and updates on their Blade smart glasses. Just as smartphone users see regular weather forecasts on their mobile screens, users of Vuzix Blade AR glasses will be able to see comprehensive weather forecasts as a visual overlay that pops into their field of vision.

Keeping a weather eye

Carrot Weather 

Think of a weather app with a personality. Carrot Weather makes forecasting and checking the weather fun. The app’s interface gives you access to current, hourly and daily forecasts—with a tinge of sarcasm and fun. The app’s forecasts are powered by Dark Sky, one of the world’s most accurate weather services.

Available on: Android and iOS

Weatherbug 

Weatherbug lets you see the weather with the help of 18 interactive maps, including Doppler radar, lightning alerts, a satellite map, precipitation, temperature, wind chill, heat index, etc. The app also gives you important information on air pollution, pollen count and allergies. The app is known for its real-time forecasts and “Spark" lightning alerts, which can be useful during thunderstorms and severe weather conditions.

Available on: Android and iOS

AccuWeather

AccuWeather is one of the oldest weather apps around, and relies on the AccuWeather Forecast Engine. According to the official website, the engine generates highly localized forecasts for 2.3 million locations, with more than 100 parameters, from temperature and humidity to rain and snow and ice amounts, “for every hour over the next 90 days and for every minute for the next 2 hours". The app also lets you personalize local forecasts.

Available on: Android, iOS and WindowsPhone

FlowX

If you are into using maps and graphs for weather forecasting, then FlowX is one app to look at. The intuitive display on FlowX makes it one of the best weather apps around. The app gives you access to more than 30 data types, including precipitation, cloud, wind, temperature, pressure, humidity, waves and more. A highlight feature of the app is its ability to show multiple data layers together. For example, you can look at the cloud layers, precipitation and wind streamlines of one region together.

Available on: Android

Close