Imagine trying to predict the weather using a map that has a few blurry spots; those spots are where our current climate models struggle. This work uses AI to fill in those gaps, specifically focusing on how clouds, oceans, and forests react to heat. It's like giving a weather forecaster a high-definition lens to see extreme events coming more clearly.
Current climate models have high uncertainty regarding how much the Earth will warm and how heat is absorbed by oceans. This makes it difficult for businesses to predict long-term environmental risks and societal impacts.
Beta versions of ML-enhanced observational data streams, a neural-network river-discharge prototype, and AI-integrated cloud and ocean schemes for ESMs.
If you are a reinsurance firm dealing with unpredictable extreme weather losses — this project developed ML-enhanced observational data and better process representations that improve the accuracy of climate extremes projections. This allows for more precise risk pricing and capital reserve planning.
If you are a data provider dealing with unpredictable crop yields due to carbon cycle shifts — this project developed hybrid parameterisations for photosynthesis and biomass dynamics. This helps in creating more reliable long-term land productivity forecasts.
If you are a fleet operator dealing with changing ocean currents and heat transport — this project developed 4D fields of temperature and carbon transport using AI. This provides better data for optimizing long-term route planning based on ocean heat uptake.
Based on available project data, no pricing or cost structure is mentioned as the project is a research-driven Horizon-RIA initiative.
The project has produced beta products and prototypes, such as a neural-network river-discharge prototype, but full industrial scaling is not yet detailed.
Based on available project data, specific licensing terms are not provided; however, it involves 20 partners across 10 countries.
The project integrates AI into existing models like ICON-A-MLe, HadGEM3-GC5.0/UKESM, and NEMO to reduce radiation biases and improve eddy closures.
The project is scheduled to run from 2024-04-01 to 2028-03-31.
The consortium is heavily academic, consisting of 20 partners from 10 countries, with 12 universities and 5 research organizations. There is a 0% industry ratio, meaning the current output is focused on scientific breakthroughs rather than immediate commercial products, though the scale of the network suggests high-quality validation.
Contact Max-Planck-Gesellschaft zur Förderung der Wissenschaften EV
Contact us to find a bridge between these academic beta products and your industrial application.
