Pelmorex this week introduce a predictive weather forecasting solution developed for energy traders.
The Oakville-based weather information and data management firm says its cutting-edge new solution, dubbed Gaia Energy, is “poised to revolutionize weather predictions and analysis within the natural gas, electricity, agriculture, and weather derivatives space.”
“This is a revolution in how weather forecasts are made,” believes Chris Scott, who functions as Pelmorex’s chief meteorologist.
Powered by Gaia, Pelmorex’s proprietary Deep Learning Numerical Weather Prediction model provides advanced predictive modeling, real-time weather insights, historical weather data, and customizable data visualizations.
These tools enable energy traders to make informed and strategic trading decisions faster and with greater accuracy, giving them a competitive market advantage, posits Scott.
“Instead of using traditional methods that are based on the laws of physics, our model learns from observed weather data, producing more accurate forecasts and eliminating the consistent errors found in traditional models,” says Scott. “The greatest advantage is that this AI model can run in a fraction of the time, providing critical weather updates sooner than anyone else.”
Pelmorex claims Gaia Energy can deliver forecasts up to 18x faster than the world’s current top conventional models while also delivering higher accuracy.
That kind of performance did not come easily to the firm, however.
“The process of bringing this project to completion has been a humbling experience,” commented Jonathan Weisbaum, who serves as director of meteorological engineering for Pelmorex. “It took a substantial amount of brain power to develop and train this model from scratch, in a distributed environment with an enormous amount of data, while providing operational support.”
Weisbaum added that Pelmorex has some of “the best engineers and scientists in the world, and it shows in the model’s forecasting skill and inference speed.”
Even possessing top tech talent, collaboration was critical.
“Our Gaia forecast model is a perfect example of what talented and collaborating teams from meteorology and data science can build together by leveraging new and powerful AI methodologies,” said Mark Gibbas, Managing Director of B2B Enterprise Solutions for Pelmorex.
“While conventional physics based models are impressive with their intricate modeling of atmospheric dynamics, these models also tend to grow forecast error very quickly,” added Gibbas. “What we are seeing with our Gaia DLNWP based model is that this error is more suppressed, leading to more accurate forecasts.”
Gibbas says Pelmorex believes in “using AI for good.”
“Gaia Energy achieves this by providing the energy trading community with more accurate forecasts that are also available an hour faster than previous generation forecasts,” he stated.
Established in 1989, Pelmorex is the owner and operator of popular weather brands including The Weather Network, MétéoMédia, and Weather Source.