To main content

Surface water and urban flooding

The aftermath of severe weather events, whether on a national or international scale, leads to substantial infrastructure damage, often resulting in considerable insurance claims. The anticipated rise in extreme weather incidents emphasizes the critical need for robust flood protection measures and advanced predictive simulators capable of forecasting flooding and stormwater events, including those triggered by intense rainfall.

Contact persons

Historical background

Our engagement with surface water and flooding events dates back to the early 2000s and originates from mathematical research on high-resolution schemes for hyperbolic conservation and balance laws conducted by our current research manager before joining SINTEF. In 2003, we initiated investigations into utilizing graphics cards for solving partial differential equations (PDEs). Shallow-water simulations became a logical starting point, given their explicit stencil computations and involvement of three components (water height and momentum in two axial directions) aligning with the colors red, green, and blue, which we manipulated through shaders to solve PDEs. Dambreak problems served as effective test cases, offering the potential to integrate simulations with advanced graphics for visually appealing visualizations; see, e.g., Hagen et al. (2007).

A series of developments unfolded, leading us to actively engage in GPU-accelerated simulation of dambreak-induced flooding in collaboration with the National Center for Computational Hydroscience and Engineering,  Mississippi; see Brodtkorb et al. (2012), Sætra (2014), and Sætra et al. (2015). The code that emerged from this collaboration was later tailored for investigating urban flooding scenarios and subsequently licensed to a reputable engineering firm in the UK. 

The second avenue of exploration begins with algorithms designed to identify the potential for structural trapping in CO2 storage, drawing inspiration from natural analogs in watershed analysis. Guided by our commitment to transferable knowledge (and encouraged by results from our master student  Voldsund (2017)), we later found ourselves modifying these algorithms to aid the startup company Spacemaker in developing tools for predicting urban flooding and mitigating damage from heavy rainfall. This adaptation was successfully integrated into their product, as detailed in articles on and (Spacemaker was subsequently acquired and transformed into AutoDesk Forma).

What do we do?

We are corrently working in three main directions:

  1. Static modelling: our methods for watershed analysis and their like have been reimplemented in the Julia programming language and designed to rapidly process large terrain sections, giving a software called SWIM.
  2. Dynamic modelling: we are currently in the process of modernizing our old dambreak/urban flooding simulator and re-implementing it in Julia.
  3. Integration: a comprehensive analysis ideally requires a combination of static and dynamic tools that interact, and we will shortly start integrating our two software tools. 

We also have the possibility of using our GPU Ocean code to study storm surges resulting from extreme weather events.




Software for static modeling and prediction of surface water and urban flooding based on analysis of topography/terrain.