High-resolution topography upscaling for overland flows – Top-up

Urban floods produced by exceptional precipitations may cause large damages both in terms of life losses and property destruction. Numerical modeling can be used to predict, anticipate and control the floods by helping to size and position protective systems including dams, dikes or rainwater drainage network. One of the major challenges in numerical modeling of urban floods is that the small structural features (like buildings, concrete walls, cars, etc.) may significantly affect the flow.

In order to include those small structures in the hydraulic simulations two major ingredients are requested:
1) high-resolution topographical data sets;
2) efficient numerical algorithms.

The first ingredient is provided by the modern terrain survey techniques including photogrammetry and LIDAR. In fact the high-resolution topography is largely available nowadays. For example, the topographical data sets of Métropole Nice Côte d'Azur that is going to be explored in our project covers approximately 1450 square kilometers with the resolution not less then 25 points by square meter.

The motivation of Top-up project comes from our believe that the second ingredient is still missing. The difficulty comes from a large contrast between a typical size of the simulation domain (10-100km) and the size of the relevant structural features, which in many cases have to be represented at metric or infra-metric scales. In other words the physical process that has to be simulated involves multiple scales, some of which are likely to be unresolved by the computational mesh. Top-up propose to address this problem using Domain Decomposition (DD) and Multi-scale (Ms) numerical methods (in the spirit of MsFEM). Those two complementary approaches are well understood and have a long record of successful applications to the multi-scale problem in porous media flow and elasticity. We believe that those two methods also provide an appropriate framework to address multi-scale flood modeling.  

The objective of the project is to design some efficient DD and Ms methods adapted to the multi-scale surface flow problems, to implement them as a parallel scientific computing software, and finally to validate them over a set of large-scale urban flood scenarios involving realistic high-resolution topography. The last objective will be achieved though the tight collaboration with Métropole Nice Côte d'Azur.