Uncertainty aspects of 2D flood modelling in a benchmark case study

In this study, we investigate the contribution of several uncertainty drivers towards the total uncertainty of a 2D flood model, in a benchmark case study under steady flow conditions. The simulator used for the analysis is the in-house FLOW-R2D software, whilst the benchmark case study consists of a compound trapezoidal channel, which represents the main channel and the floodplains. Unlike the conventional taxonomy of the uncertainty sources (input data, parametric and structural), we define five drivers: a) the forcing driver which consists of the inflow to the computational domain; b) the geometric driver which depends on the topography of the case study; c) the physical driver which incorporates all the parameters required to describe a physical process (such as friction); d) the computational driver which includes the parameters needed for computational reasons (e.g. space step); e) the structural driver which is metric for the weakness of the numerical model to capture an idealized analytical solution or observed data, due to the abstraction from reality. For the quantification of each driver contribution, we present the Uncertainty Index, which is based on the stochastic Monte Carlo technique.