Seasonal forecasts
Over the past five years, the Institute for Environmental Studies (IVM) has developed a generic AI method for making seasonal forecasts. To predict temperatures, the startup uses machine-learning techniques to crunch data on soil moisture, sea-surface temperature and surface temperature at two metres above sea level. The forecasts have been found to hold up well for the spring and summer months, but could be improved for autumn and winter. The temperature of the sea surface in the northern Pacific Ocean appears to play a major role in the predictability of the surface temperature in Western Europe.
Example projects
The method has been used to predict westerly winds high in the atmosphere around the Arctic (known as polar vortex variability), Indian monsoon precipitation and extreme temperatures in the US. It is also being used to forecast temperatures for an energy-trading company and to produce operational software for the World Food Programme with a view to combating drought in Mozambique.
Can you tell us more about the project in Mozambique?
The World Food Programme (WFP) uses, produces and translates climate information to vulnerable societies to make them more climate resilient. It also helps governments and partners around the world to adapt to changes in food security. To this end, the organisation relies heavily on timely and reliable forecasts for droughts and floods that can be translated into early-warning systems. One of the WFP’s latest action programmes is Forecast-based Financing (FbF). In the event of an anticipated drought, an early-warning system triggers measures to be implemented through predetermined funding systems.
The WFP has strong, ongoing collaborations with different countries and is currently developing FbF in cooperation with the national meteorological institute of Mozambique. Previous collaboration between the IVM and WFP showed that our AI-based seasonal drought forecasts could yield promising results if implemented in the FbF system. The WFP operates under the auspices of the United Nations.
Startup
Sub-seasonal to Seasonal Weather Predictions Boosted by AI is a startup that provides insights into the climate system and offers practical applications for business clients. It is of particular interest in highly weather-dependent fields, such as the energy sector (grid operators, energy traders, energy producers) and agriculture (e.g. large agricultural companies and NGOs). With its motto ‘Bringing AI innovations for S2S forecasting to society’, the startup is also a good illustration of how researchers at VU Amsterdam are doing their part to make the world a better place.
The startup was founded by Jannes van Ingen, MSc Climate Econometrics, business and software developer; Sem Vijverberg, MSc Climate Physics, atmospheric and data scientist; and Dim Coumou, VU/KNMI professor on climate extremes and societal risk.
Partners
- Business partners: Cross Energy Trading, MetDesk
- NGO partners: Climate Hazards Center, World Food Programme, 510: An initiative of the Netherlands Red Cross
- Academic partners: VU Amsterdam, DLR Institute for Data Science, Royal Netherlands Meteorological Institute (KNMI)
This project is linked to three SDG's:
- SDG 13: Climate action
- SDG 2: Zero hunger
- SDG 7: Affordable and clean energy.