Wow! AI Designed to More Accurately Predict Extreme Weather

Table of Contents

Korean Rumit.CO.ID, JENA — An international team of scientists led by Markus Reichstein and Vitus Benson from the Max Planck Institute for Biogeochemistry has developed an early warning system concept based on artificial intelligence (AI). The research aims to assist humanitarian organizations and disaster response agencies in dealing with extreme weather such as droughts and heavy rains.

This research published in Nature Communications marks an important leap in utilizing AI for predicting disaster impacts with high precision. This system allows for more directed and efficient distribution of resources, while also providing more time for preventive actions. Reichstein said that early warning systems were previously only designed for the short term, from a few weeks to a few months.

"However, we also need to start thinking strategically for the long term — up to decades — so that we can design impactful preventive policies," he said, as quoted from the official website. Max Planck Institute , Saturday (28/6/2025).

This system not only provides information for adapting infrastructure or relocating settlements, but also helps the agricultural and forestry sectors adapt by planting drought-resistant varieties. The AI approach allows for disaster impact predictions up to a 20-meter scale, covering each field or garden. This is made possible through the utilization of data from the Copernicus satellites, which are then studied by AI to understand the patterns of extreme weather impacts on areas with similar geological and ecological characteristics.

"The conventional climate model cannot accurately predict the impact of extreme weather because there are too many local factors that influence it, such as soil type, vegetation, and topography," Reichstein said.

Timnya successfully predicted the impact of drought on various ecosystems, while another group of researchers developed an algorithm to predict the effects of extreme rainfall. The system was designed in six modules, starting from high-resolution spatial and temporal measurement of the impacts of extreme weather, precise weather prediction, ecological and economic impact projections, to communication and psychological methods so that warnings can be understood and acted upon effectively.

"Prediction alone is not enough, we must ensure that the warning is understood, taken seriously, and turned into action — both at the individual and policy level," Reichstein said.

AI also plays a crucial role in communication aspects, such as conveying warnings through voice, images, and visualizations of impacts. The long-term goal of this project is to create a warning system capable of recognizing the impacts of various extreme weather disasters globally, providing effective warnings, and even suggesting mitigation measures.

However, significant challenges remain, one of which is the need for explainability or the ability to explain decisions made by AI. "Today's AI still relies on statistical correlations rather than causal relationships; we need AI that can understand the physical conditions underlying extreme weather events," said Benson.

As an initial step, the team at the Max Planck Institute is developing a public-access early warning application, including communities in developing countries. "Early warnings in the Global South are still general. We want to make small-scale information accessible to everyone," Benson said. With this approach, AI will not only serve as an analysis tool but also as a support for strategic decision-making to face the increasingly unpredictable impacts of climate crises.