Cutting-Edge Flood Warning Tool Developed by DTU Researchers Shows Promise in Early Detection

Researchers from the Technical University of Denmark (DTU) have collaborated with Jammerbugt Municipality to create a pioneering early warning system for local flooding.

Innovative solution to address local flooding challenges

Susanne Nielsen, a resident of Aalborg, Denmark, has expressed concerns about potential flooding affecting her parents’ summer house in Slettestrand, North Jutland. The house’s proximity to the bay of Jammerbugt poses a risk of water intrusion, particularly with rising groundwater levels.

To mitigate this risk and provide timely warnings to residents and decision-makers, DTU researchers have developed an advanced early warning tool. Unlike traditional flood warning systems, this tool offers localized predictions, giving stakeholders up to 48 hours notice of impending flooding along rivers, streams, and coastal areas within Jammerbugt Municipality.

Central to the system is the “wet index,” a model based on artificial intelligence (AI) trained on diverse datasets, including satellite imagery, weather forecasts, ground and seawater levels, and landscape topography. This multidimensional approach enables a nuanced understanding of water dynamics and interactions with the surrounding environment.

Roland Löwe, an Associate Professor at DTU specializing in hydrology, underscores the complexity of water movement in open landscapes. By leveraging AI and carefully curated datasets, the researchers have developed a tool capable of accurately predicting local flooding events.

Trial run and prospects

In 2023, Jammerbugt Municipality trialed the early warning tool, yielding promising results during the wet spring months. However, challenges arose during the drier summer, highlighting the need for further refinement.

Heidi Egeberg Johansen, Project Manager at Jammerbugt Municipality, acknowledges the tool’s potential while emphasizing the importance of accuracy and reliability. Plans to retrain and adjust the model are underway, with funding sought to support ongoing development efforts.

In parallel with flood warning advancements, DTU researchers have pioneered scientific machine-learning techniques to enhance water management strategies. Combining machine learning with scientific computing has significantly reduced computation time without sacrificing accuracy.

Allan Peter Engsig-Karup, an Associate Professor at DTU, underscores the benefits of this approach in predicting water movement through drainage systems. By harnessing the power of scientific machine learning, calculations are performed up to 100 times faster than traditional methods, empowering decision-makers with real-time insights.

Transforming flood management practices

Integrating AI-driven solutions in flood management represents a paradigm shift in resilience planning. With faster and more precise predictions, municipalities can proactively allocate resources, implement preventive measures, and effectively adapt infrastructure to mitigate flood risks.

Roland Löwe emphasizes the practical implications of this technological advancement, enabling decision-makers to convene and explore various scenarios in real time. By streamlining the decision-making process, communities can optimize their response strategies and enhance overall resilience to flooding events.