Global warming is causing more extreme weather leading to societal impacts all around the world. By predicting long-term weather with the help of AI, much suffering can be prevented. AI is currently causing a revolution in this field.
Artificial Intelligence offers opportunities to analyse complex data and make weather forecasts for the long term. AI applications are currently developing rapidly and software offers more and more possibilities. Both for long-term forecasts of the weather itself, and for the expected damage from the weather, such as failed harvests. AI applications are also valuable for both energy production (by wind and sun) and energy demand (by extreme weather, among other things). The energy market needs solutions to properly coordinate supply and demand.
Improving seasonal forecasts
Within the Institute for Environmental Studies (IVM), the Climate Extremes research group is working on applying AI methods to improve seasonal forecasts and understand the underlying processes. A team of climate scientists and AI engineers from this research group has founded the startup Beyond Weather. The research group and startup train machine learning techniques on large datasets of ocean temperature, soil moisture and specific patterns in the atmosphere. The analyses show that the predictions for many regions and seasons are better than predictions based on classic methods using weather models. New processes can also be discovered this way. For example, the temperature of the ocean water in the tropical Pacific Ocean plays an important role in the predictability of heat waves in Western Europe.
Projects
The research group has developed several AI methods for predicting the wind blowing high in the atmosphere around the North Pole, also called the polar vortex, Indian monsoon precipitation and extreme temperatures in the US. Currently, this method is also applied for temperature forecasting for the energy sector and harvest forecasting for the agricultural sector.
Deep Learning
The team is currently looking at heat waves in Europe and predictability in the Mediterranean using new Deep Learning techniques. Together with partners University of Amsterdam and Microsoft AI, they are developing Deep Learning methods for long-term predictability of droughts and other extremes, such as forest fires, in the Mediterranean – a region that is hit hard by climate change.
Startup Beyond Weather
Beyond Weather was founded in 2023 and brings AI-supported forecasts to sectors such as agriculture and energy, but also to humanitarian organisations. AI-based forecasts are only useful if they are "brought" to society, instead of just described in scientific publications. A good example is a collaboration with the World Food Programme (WFP), in which seasonal forecasts are produced and translated to vulnerable societies to make them more climate-proof. In the energy world, you can think of grid operators, energy traders and energy producers. In the agricultural sector, you can think of large agricultural companies and NGOs. The startup also proves how research at VU Amsterdam can make the world a little better. Improved long-term forecasts can save lives and reduce costs.
Beyond Weather is a startup by Jannes van Ingen, Sem Vijverberg, Steven van den Tol, and Dim Coumou.
Partners
- Business partners: Cross options energy trading, MetDesk, KPMG, Microsoft AI, Gallagher RE, Rabobank, World Bank
- NGO partners: Climate Hazards Center, WFP, Red Cross Red Crescent, Climate Adaptation Services (CAS),
- Academic partners: In Netherlands: KNMI, UvA.
Internationally: Many partners throughout Europe and beyond
