IBM and NASA collaborate to develop AI foundational model for enhancing efficiency in extreme weather forecasting.

Technology giants IBM and NASA have announced a new collaboration aimed at developing an AI-based model to significantly improve the accuracy and efficiency of weather forecasting and climate applications.

This new model combines IBM's extensive artificial intelligence capabilities with NASA's vast amount of Earth science data and expertise. It aims to ingest diverse meteorological datasets and run complex simulations faster than existing forecasting techniques, while also inspiring new climate insights.

While artificial intelligence has already been applied in weather prediction models, such as NVIDIA's Fourcastnet and Google DeepMind's GraphCast, IBM points out that these are narrow-focused AI-based "simulators" rather than versatile foundational models.

These AI simulators can quickly produce weather forecasts, but due to limitations in training data, they lack a fundamental understanding of weather physics. The IBM-NASA foundational model has several key advantages:

• Analysis of a wider variety of data types and sources
• Faster data ingestion and inference
• Easier adaptation to new climate modeling requirements
• Improved accuracy for existing forecasting purposes

Potential applications include inferring high-resolution insights from low-fidelity data, predicting extreme weather events, forecasting disasters like wildfires, and more.

This is not the first time IBM and NASA have invested in climate-centric foundational models. Their earlier release of the geospatial artificial intelligence model was designed to help analysts extract insights about climate change impacts from satellite imagery data.

At the upcoming COP28 climate conference, IBM will showcase how this model has been fine-tuned to adapt to various applications. In Kenya, IBM's model assists in visualizing tree planting activities, addressing water scarcity, and combating deforestation. In the United Arab Emirates, in collaboration with MBZUAI, it is used to study urban heat island effects, aiding in understanding and mitigating their formation.

Curioni said, "AI-based models using geospatial data can accelerate the discovery of climate solutions. We hope these technologies can help address global health threats faster than ever before."

The new weather and climate model will also leverage IBM's AI platform and NASA's modeling capabilities to handle multi-petabyte datasets. Over time, it could provide valuable guidance for adaptation strategies in the face of intensifying global warming.

As life and ecosystems become increasingly vulnerable to extreme weather, IBM and NASA's model may prove to be a crucial tool for officials, businesses, and communities on the front lines of climate crisis.