Weather Forecasting Through Machine Learning

What is the one thing we all need to ask before we leave the house each morning?

“What’s the weather going to be like today?”

Luckily, we live in the transformational age of advanced and ever-evolving technology, and we now have the gift of artificial intelligence and machine learning (ML).

By looking at years of weather data and pairing it with known weather patterns and outside factors, computers can now predict the weather like never before.

For instance, during the Paris Olympics this year, AI tools were used to provide a daily forecast of weather conditions.

Recent advances in ML-based weather prediction (MLWP) have produced ML-based models with less forecast error than simpler past models.

As noted by the science news magazine EOS, “It took fifty years to build and incrementally improve the sophisticated computer models needed to produce such accurate forecasts. It has taken ML approaches just a couple of years to match, and in some cases surpass, the skill of these traditional models.”

Now, Google DeepMind scientists have developed a weather model that largely beats the world’s most accurate modeling system.

Researchers believe the model may be able to accurately predict unprecedented events that climate change accentuates.

According to a report last month commissioned by the International Chamber of Commerce, “climate-related extreme weather events have cost the global economy more than $2 trillion over the past decade.”

According to Maginative, “Google DeepMind’s new high-resolution AI model, GenCast, delivers faster, more accurate weather predictions than traditional methods. AI will fundamentally change how we plan for everyday weather and extreme events.”

Part of the reason why this is such a significant development is because GenCast offers weather forecasts that are not only more reliable but also faster. Unlike traditional methods that rely on supercomputers and take hours, GenCast runs on a single TPU chip in just eight minutes, predicting a wide range of possible outcomes rather than a single scenario, as current weather prediction models do.

In contrast to traditional models, which make forecasts by solving physics equations, GenCast learns directly from historical weather data.

Jacob Mallinder, in CEO Today, outlines some of the top AI innovations that are making a difference in weather forecasting today.

He explains that aside from improved data gathering and predictive analytics, AI today allows for insight into climate change, agricultural optimization, improving efficiency in the energy sector, and enhancing aviation safety.

MAELSTROM is another large-scale R&D project, aiming to fundamentally improve weather and climate prediction. It aims to combine the powers of high-performance computing (HPC) and ML to cope with the extreme complexity inherent in weather and climate forecasts.

In August, NVIDIA Research announced a new generative AI model, dubbed StormCast, for emulating high-fidelity atmospheric dynamics.

The model was built to provide reliable weather prediction at the mesoscale — a scale larger than storms but smaller than cyclones — which is critical for disaster planning and mitigation.

Quantum ML is another option, but some experts believe it is not the optimal way to go as quantum computing is still in the early stages of development.

If it is the case that GenCast and other emerging technologies can predict serious weather events, it will introduce a monumental change. With such abilities, we can warn people ahead of time and prevent traffic accidents. Farmers can plan their farming and prevent crop loss. Energy companies will be able to run more effectively if they can understand sun and wind patterns in advance.

And when we walk out the door, we won’t need to wonder what the weather will be like because we’ll already know.