To keep the Netherlands safe, liveable and accessible in a time of climate change, the AI Impulse Programme water movement models strives for more agile, faster and better substantiated water management. The programme is supported by Rijkswaterstaat in collaboration with KNMI, Deltares and DigiShape, among others. Within the programme, Rijkswaterstaat and DigiShape are now starting a Challenge around fast and reliable wave forecasts in the North Sea.
By developing machine learning models and high-quality datasets for the main water system (rivers, lakes, coast), Rijkswaterstaat wants to respond proactively, adaptively and data-driven to the increasing unpredictability of water.
In order to be able to deliver fast, flexible and reliable water models that can be used operationally and managed in a controlled manner – with the right quality, explainability and integration into existing systems as a starting point – the programme wants to join forces with knowledge institutions, market parties and start-ups.
One of the initiatives is to determine the expectations of wave parameters for the North Sea, mainly significant wave height and swell. These are used in the North Sea for, among other things, making tidal gates and issuing warnings for high waves during storms.
Currently, the operational system RWsOS of Rijkswaterstaat uses a SWAN model to model the waves in the North Sea. Due to the long calculation time of the model, only one deterministic weather forecast is currently calculated. Calculating a complete ensemble (such as the 50 ensemble members of ECMWF-EPS) is not possible due to the long calculation time of the model. Because uncertainty information from these wave parameters is desirable, the AI Impulse Program is investigating whether a fast computational model (emulator) can be developed to generate uncertainty information. That is why RWS has decided, together with Digishape, to organize a challenge to take this further.
Challenge for fast and reliable golf forecasts
With the Challenge, we invite market parties, start-ups and knowledge institutions to think about innovative solution options for future, fast, flexible and reliable water models for wave parameters.
Participants are asked to write a Plan of Action, which describes:
- what type of design is used;
- which datasets are used;
- how the model is trained;
- and how the model can ultimately be used operationally.
Important dates
NB: All communication, questions and registration regarding the Challenge are done via TenderNed.
- Online information session: 1 June 2026
- Deadline for the round of questions: 5 June 2026
- Submission of Action Plan via TenderNed: 15 June 2026
- Pitch session during DigiShape day: 19 June 2026 (Register via DigiShape)
If there is sufficient perspective, the challenge can be followed up by a multiple private request for a proof-of-concept and further upscaling towards a pre-production model for multiple North Sea locations.
More information and the full challenge description can be found via Tenderned.
