Doing Something About the Weather with Big Data
“Everybody talks about the weather, but nobody does anything about it.” You’re probably familiar with that old saw from 19th century Puritan Charles Dudley Warner. As it turns out, some enterprising folks here in the 21st century are finally doing something about the weather—by mixing it with big data.
Weather forecasting is, in itself, a big data endeavor. Twice a day, the latest meteorological observations from around the world are loaded into supercomputers and run through sophisticated models that give us a glimpse of what the weather probably will be like in three to seven days’ time. The forecasts simply are not that accurate beyond seven to 10 days, which shows us how much we still have to learn in this fascinating field.
While improving weather forecasting continues to be a big focus of the HPC community, the weather is also playing a role in big data endeavors. As Warner noted 200 years ago, the weather is a constant topic of conversation among humans. Whether it’s hot or cold, rainy or dry, the weather has a huge impact on where we go, what we do, and (most importantly to capitalists) how much we spend.
Here are three ways people are actually doing something with the weather in big data:
Maximizing Airport Parking Revenue
How much people will pay for short-term parking at an airport is dependent on a lot of factors. Being late for your flight, being unable to walk far, or having unruly kids in tow will generally lead people away from the cheaper off-site lots and towards the more expensive parking at the airport proper.
It turns out that the weather is another big factor impacting what people will pay, and it just so happens to impact everybody. This realization led the Dublin Airport in Ireland to implement a dynamic pricing scheme that automatically calculates the optimal price to charge people to park there.
“They know, given the weather conditions, how much more they can charge for the short-term parking close to the gate,” says Ulrik Peterson, COO of TARGIT, the business intelligence vendor supplying the software powering the system. “They’ve actually given the control over pricing to the system.”
The system has had a big impact on the airport’s bottom line, Peterson tells Datanami. “It runs real-time and it’s based on how many spaces they have left and how is the weather and the forecast for the next two hours,” he says. “They marry those together, and then they can know how much more they can charge.”
The airport is also using weather data to make staffing recommendations, also done in TARGIT’s Decision Suite. “They’re using the weather data now with the internal data to better predict their staffing on the various posts or departments at the airports,” Peterson says. For example, if it’s raining, the TARGIT software will alert the airport to add more staff to the gates, where they can speed up the boarding process by encouraging people to put away their umbrellas and extra clothing.
It’s not something that can’t be learned from experience, but it can be useful to have the computers gently reminding us of our duties as weather conditions change.
Preparing for Weather Emergencies
Bad weather can be inconvenience, but it can also be deadly. In 2014, natural disasters around the globe accounted for 7,700 deaths and $110 billion in losses, the majority of which were weather-related. Storms, floods, heat waves, cold waves, droughts–they all take their toll on the occupants of the Earth.
While it’s not possible to stop the disasters, we can help to save lives and stem monetary losses by improving how we respond to them. Earlier this summer, IBM announced a partnership with The Weather Channel aimed squarely at empowering emergency responders to better cope with impeding weather events.
As part of the deal, IBM will incorporate weather-related data and algorithms from The Weather Channel’s B2B division, called WSI, into its emergency management solution. That software, running at new Intelligent Operations Center (IOC) for Emergency Management, will collect historical and sensor data, and apply “deep analytics” and data visualization capabilities to help agencies better coordinate.
“Big data is revolutionizing emergency management and transforming how communities protect citizens and property in times of emergency – which can range from hurricanes and snowstorms to highway accidents and riots,” said Robert Griffin, General Manager of IBM’s Safer Planet initiative.. “The IOC for Emergency Management from IBM enables data sharing across organizational boundaries and applies sophisticated analytics to provide insight for future incident planning and response.”
Gaining an Edge in Energy Trading
Like other commodities, the price of electricity and natural gas fluctuates on the public market in response to supply and demand. Weather plays a major factor in the demand for electricity and gas. Consider that a heat wave will drive air conditioner use through the roof, while a cold snap will lead to more need to heat home and offices, and therefore drive up demand for natural gas and heating oil.
Despite the sensitivity of these markets to the weather, traders have been stuck with the same seven- to 10-day forecasts generated daily by American and European teams. Traders seeking an advantage must go elsewhere, and increasingly they’re looking to big data.
Energy traders aren’t so interested in knowing whether we should bring a sweater to the picnic in at the park this weekend. What they’re after are the likelihood of extreme events occurring. One outfit using big data to build a 40-day forecast of extreme weather events is EarthRisk Technologies, which, ironically, is based in the most boring weather city in the country, San Diego, California.
EarthRisk claims that its TempRisk Apollo solution “is the first commercial application with scientifically validated statistical methods that analyze the risks for significant heat and cold events up to 40 days in advance,” the company says on its website. The company uses “advanced algorithms” to process more than 60 years of global weather patterns.
Customers in the energy trading business say the approach is working. Adam O’Shay, the president and head of trading at Leeward Point Capital, says the software has “transformed our trading and analytics” by allowing for “targeted probabilistic temperature risk-assessment for all critical areas related to the North American natural gas market.” Josh Darr, a meteorologist at Chesapeake Energy, says the solution “allows users to quickly identify key drivers of temperature prediction from both a numerical weather prediction and analytic standpoint.”
There are numerous other ways that weather data can play in big data analytics. How are you using it? Send us a note at [email protected]