Modern weather forecasts are typically done by powerful computers performing detailed physics calculations. This taxing method required solving complex equations in an attempt to predict temperature, wind, rain, and other weather events.
Utilizing modern artificial intelligence techniques could prove a useful alternative. A newly developed global weather model makes predictions based on the past 40 years of weather data. This analysis of past weather patterns could yield more efficient and potentially accurate results than current models.
The simple, data-based AI model can simulate a year’s weather around the globe much more quickly and almost as well as traditional weather models. This is according to a paper in the Journal of Advances in Modeling Earth Systems. The model does so by taking similar repeated steps from one forecast to the next.
“After training on past weather data, the AI algorithm is capable of coming up with relationships between different variables that physics equations just can’t do,” lead author Jonathan Weyn says. This work comes as part of his doctorate research in atmospheric sciences.
To merge successful AI techniques with weather forecasting, the team mapped six faces of a cube onto planet Earth, then flattened out the cube’s six faces, like in an architectural paper model.
The data-driven model would need more detail before it could begin to compete with existing operational forecasts, the authors say, but the idea shows promise as an alternative approach to generating weather forecasts, especially with a growing amount of previous forecasts and weather observations.
This new bite was adapted from an article originally written for WE Forum.
Interested in learning about machine learning and data science? Whether you just want to pick up a few new skills or launch a new career, Lantern Institute provides various programs to help you get started.