AI News Monitor

AI models can make weather forecasting faster, more precise, and more efficient.

Tech giant Nvidia and the European Center for Medium-range Weather Forecasts (ECMWF) have both announced improved weather forecasting systems using generative AI models.

The ECMWF announced on Tuesday that it is starting operations with a new AI forecasting system. The ECMWF is an intergovernmental organization that provides global weather data for Europe. According to the body, its new AI model outperforms the best non-AI models by up to 20 percent for many measures. It also uses 1,000-times less energy.

In a similar development, US tech giant Nvidia claimed that its new generative AI model, CorrDiff, can transform how communities manage extreme weather phenomena. According to Nvidia, the model can also improve renewable energy management and agricultural planning.

CorrDiff has better prediction scales than other weather models. While typical AI models are limited to 25 km resolution, CorrDiff is able to zoom into areas as small as 2 km to provide more localized and detailed weather information. This high resolution forecast makes it possible to predict more precise weather patterns and their likely impacts.

This means that Nvidia’s CorrDiff can now provide much more accurate information about events like hurricanes and floods at the neighbourhood level. CorrDiff is also up to 10,000 times more energy efficient than conventional weather models.

Agencies in Taiwan, a high-risk region for extreme weather events, are using the technology to improve their disaster preparedness by receiving early warning signs. Various other meteorological and disaster management agencies worldwide are now also adopting CorrDiff, according to Nvidia.

Traditional weather forecasting involves collecting vast amounts of atmospheric data to run physics-based equations on supercomputers. AI models, on the other hand, are trained on historical weather data to quickly predict future conditions without relying on complex equations. This makes them much faster and more efficient. While weather calculations take hours to run, AI models can generate similar results in only a few minutes.

Weather forecasting is not only useful for managing extreme weather events that threaten communities. It is also used in other sectors, such as in the energy sector to predict wind speeds or solar radiation levels, or in shipping to make better route decisions and avoid poor conditions. In agriculture, it can help farmers make appropriate schedules to reduce losses and improve productivity.

While there has been some uncertainty about using AI for an activity as critical as weather forecasting, it seems those doubts have largely been put aside. However, the ECMWF says its AI forecasting system will run alongside the more traditional method that uses physics-based equations.