The weather is an unforeseeable, ungovernable force and EarthRisk is using big data to predict it.
EarthRisk Technologies has developed a new model for predicting extreme weather events. The model identifies weather patterns based on over 82 billion calculations and 60 years of data. It then compares those patterns to current conditions and uses predictive analytics to predict the weather up to 40 days in advance.
The technology is derived from research at the University of California at San Diego’s Scripps Institution of Oceanography. Cofounder and CEO John Plavan said the old standard for weather prediction is built on subjective forecast models that are not accurate beyond a week.
“Hundreds of thousands of atmospheric variables are changing constantly around the globe and the old models aren’t robust enough to take these into account,” Plavan said in an exclusive interview with VentureBeat. “If there is a change to the initial conditions, the whole thing breaks down. We use statistical relationships to predict eventual outcomes and this technique is not subject to the same chaos. We are applying analytics to an industry that is begging for reinvention.”
EarthRisk has collected data from the U.S. and U.K. governments as well as observational data from thousands of scientists and researchers working in the field and the database is updated every day. EarthRisk’s engine searches for correlations and patterns of “statistical significance” and generates forecast probabilities based on this information. The approach uses the past to predict the future.
“Utilities corporations, energy traders, and energy producers are majorly impacted by big temperature changes and spend hundreds of millions of dollars trying to predict them,” Plavan said. “If they know there will be an extreme cold event a month from now, they can use that data to make an actionable decision, and these guys will do anything to gain a small edge.”
EarthRisk’s flagship product TempRisk is the first commercial application of this technology and is geared towards the research and energy trading communities. The company has been developing, refining, and testing the technology for a few years and now plans to expand the business dramatically and explore more commercialization opportunities. There could be more consumer-focused applications down the road, like the ability to check weather in a tropical location before booking a vacation.
Our upcoming GrowthBeat event — August 5-6 in San Francisco — is exploring the data, apps, and science of successful marketing. Get the scoop here, and grab your tickets before they're gone!