Uncertainty related to high resolution topographic data use for flood event modeling over urban areas: toward a sensitivity analysis approach

¹ Institut de Radioprotection et de Sûreté Nucléaire & Lab. I-CiTy URE 005, EPU Nice Sophia, University of Nice, France,
e-mail : abily@polytech.unice.fr, corresponding author
² Lab. J.A. Dieudonné UMR 7351 CNRS & EPU Nice Sophia, University of Nice, France,
e-mail : delestre@math.unice.fr
³ EPU Nice Sophia, University of Nice, France,
e-mail : amoslau1@aquacloud.net
⁴ Institut de Radioprotection et de Sûreté Nucléaire, France,
e-mail : nathalie.bertrand@irsn.fr
⁵ Institut de Radioprotection et de Sûreté Nucléaire, France,
e-mail : claire-marie.duluc@irsn.fr
⁶ Institut de Radioprotection et de Sûreté Nucléaire, France,
e-mail : richet.yann@irsn.fr...
⁷ Lab. I-CiTy URE 005, EPU Nice Sophia, University of Nice, France,
e-mail : philippe.gourbesville@unice.fr
⁸ IRMA, UMR 7501 CNRS, Unistra, France,
e-mail : navaro@math.unistra.fr

Abstract

2D Free-surface hydraulic modeling tools are commonly used to assess flood hazard for production of maximal water depth (h_max) maps, as support for flood risk assessment. High Resolution (HR) topographic data are big data getting commonly available and used by hydraulic modeling community. Topographical information and its strategy of inclusion in models, are inputs of great importance for overland flow h_max calculation. To strengthen the assessment of confidence level in these deterministic hydraulic models outputs, uncertainty analysis (UA) and global sensitivity analysis (SA) can provide useful information that is required by practitioners and decision makers. UA and SA approaches allow to identify effective strategies to reduce the uncertainty of a model output. In this paper, developed approach consists in parameterizing three factors which introduce uncertainty related to HR topographic data use with hydraulic models: the measurement error (var. E), the level of details of above ground element representation in DEM (buildings, sidewalks, etc.) (var. S), and the spatial discretization resolution (grid cell size of a regular mesh) (var. R). Parameter var. E follows a probability density function whereas parameter var. S and var. R are discrete operator choices. The coupling of an environment for parametric computation (Prométhée) and a code relying on 2D shallow water equation (FullSWOF2D), Promethée-FullSWOF2D (P-FS) tool has been set up. P-FS tool allows launching directly numerous set of computation using R software. 1200 simulations of a river flood event scenario were performed on the regular computational mesh, spatially discretizing a 17.5 km² urban area (Nice, France). The aim is to produce UA over points of interests and SA through Sobol index maps production.