DigiShape day – AI Impulse Program for Water Movement Models

Wanssum, 21-09-2021, Hoogwatergeul Ooijen - Wanssum in Limburg aan de Maas. Maaspark Ooijen-Wanssum is een natuurgebied aangelegd op oude Maasarmen. De tienduizend jaar oude rivierlopen van de Maas zijn herkenbaar, kreken, moerasbossen en met grondwater gevoede laagtes maken het tot een zeer gevarieerd gebied met een grote ecologische rijkdom. Eens in de zoveel tijd stroomt er Maaswater door het gebied.  Bij hoge waterstanden stromen de hoogwatergeulen als eerste mee. Wanneer de waterstanden in de Maas verder stijgen, zal de gehele tien kilometer lange Maasarm van Ooijen tot Wanssum zich vullen en op spectaculaire wijze meestromen met de rivier Maas. Bij extreem hoogwater zorgt dat voor een waterstandsverlaging van wel 35 cm op de Maas. Er zijn fietspaden en wandelpaden, bruggen, avontuurlijke oversteken met stapstenen. Op de foto paaltje dat bij rijsend water de diepte aangeeft voor verkeer dat over de weg rijdt.
Wanssum, 21-09-2021, Hoogwatergeul Ooijen - Wanssum in Limburg aan de Maas. Maaspark Ooijen-Wanssum is een natuurgebied aangelegd op oude Maasarmen. De tienduizend jaar oude rivierlopen van de Maas zijn herkenbaar, kreken, moerasbossen en met grondwater gevoede laagtes maken het tot een zeer gevarieerd gebied met een grote ecologische rijkdom. Eens in de zoveel tijd stroomt er Maaswater door het gebied. Bij hoge waterstanden stromen de hoogwatergeulen als eerste mee. Wanneer de waterstanden in de Maas verder stijgen, zal de gehele tien kilometer lange Maasarm van Ooijen tot Wanssum zich vullen en op spectaculaire wijze meestromen met de rivier Maas. Bij extreem hoogwater zorgt dat voor een waterstandsverlaging van wel 35 cm op de Maas. Er zijn fietspaden en wandelpaden, bruggen, avontuurlijke oversteken met stapstenen. Op de foto paaltje dat bij rijsend water de diepte aangeeft voor verkeer dat over de weg rijdt.

How can we use AI to predict faster and better what the water is going to do? And what does that mean for daily water management? Within the AI Impulse Programme for Water Movement Models (AIWaM), Rijkswaterstaat, Deltares, KNMI and DigiShape are investigating how AI and machine learning can help predict water levels, currents and waves more quickly.

On Thursday 19 November, we will share the first developments and experiences from this programme. What applications are being built? Where are we and what challenges do we face? We look not only at the technical results, but also at the approach and what other organizations can learn from it.

Faster calculations, better prepared

Rijkswaterstaat uses water models to predict how water levels, currents and waves will develop, for example during high water, drought or storms. Calculating different scenarios can take a lot of time. With AI and machine learning, some calculations can be performed significantly faster. This makes it possible, for example, to better determine inaccuracy, to investigate more scenarios and to gain insight into possible developments more quickly.

AIWaM is working on applications for the major rivers, including the Rhine and the Meuse, the IJsselmeer region and the North Sea. This is not only about developing new AI models, but also about how they can connect to existing models, systems and work processes.

From AI models to digital assistants

During this DigiShape day, the parties involved will show what they are now working on. For example, AI models for the North Sea and the Markermeer will be discussed, we will discuss the essence and development of high-quality datasets and we will look at how predictions can be better substantiated with new tools.

One of the highlights of the DigiShape day is the announcement of the winner of the Machine Learning Model for Golf Parameters challenge. Three market parties have each developed a prototype for this. The winner presents and demonstrates the application, also paying attention to the reliability and uncertainty of the predictions. We also discuss the experience that has been gained.

We also look ahead. Rijkswaterstaat is preparing a new tender for a digital assistant that supports professionals in making and assessing water expectations. In addition, we will be introduced to the Intelligent Modelling Lab for Weather, Climate and Water (iML-WKW) programme. This broader initiative focuses on a new digital infrastructure to develop and accelerate complex calculations and models, with standard and benchmarking. And is also intended as an ecosystem for public and private partners.

What can we learn from each other?

Although AIWaM is initially focusing on Rijkswaterstaat’s main water system, the developments are also relevant for other organisations in the water sector.

For example, water boards deal with water level forecasts and flood risk management at their own scale. Many organisations also work with similar models and systems, such as Delft-FEWS, a platform from Deltares for processing data and making water forecasts.

That is why this meeting is not only about the results of AIWaM, but also about the way in which we apply AI and machine learning in water management. How do we combine new techniques with existing models and knowledge? How do we ensure that applications become usable in practice? And where can organizations learn from each other’s experiences?

For whom is this day interesting?

The DigiShape day is intended for professionals who work on water expectations, water safety, hydrology, modelling, software development and the application of AI in water management.

We also expressly invite employees of water boards, consultancy and engineering firms, software suppliers, AI specialists and knowledge institutions. Whether you develop your own models, work with water forecasts or want to discover the possibilities AI offers for your organization: this day provides insight into the developments and offers an opportunity to exchange knowledge and experiences.

Program

During the day, the following topics will be discussed:

  • Progress of AIWaM: what has been developed since the start of the programme and what are the next steps?
  • Intelligent Modelling Lab for Weather, Climate and Water: introduction to a broader initiative, a collaboration between Deltares, KNMI and RWS. A national digital infrastructure facility for future models. With standards for quality and benchmarking, an ecosystem for public and private partners. This is also a practical example.
  • AI models for the North Sea: how can time series models and Graph Neural Networks help predict water movements?
  • AI for the Markermeer: how are datasets and models developed to calculate water movements faster?
  • Challenge golf forecasts: announcement of the winner and demonstration of the developed AI model.
  • Forecast Assistant: market exploration of a digital assistant for professionals who make water forecasts.
  • Forecast verification tool: how can we check how well water forecasts match reality?
  • River explorer: introduction to an application for exploring situations on the rivers.

The programme also includes time to discuss challenges and experiences in the field of water expectations.

Subscribe

Do you want to discover how AI can contribute to faster water expectations and exchange experiences with other professionals in the water sector? We look forward to welcoming you on November 19 in Amersfoort!