An analytical methodology to assess epistemic uncertainty of 2D flood models under steady flow conditions

We discuss the epistemic uncertainty observed in the output of 2D flood models in a steady-state conditions using an idealized benchmark setup. First, we propose a new taxonomy in uncertainty sources defining five drivers: a) forcing, b) geometric, c) physical, d) computational, e) structural. Then, we perform a sensitivity analysis to investigate the influence of the drivers’ variables to model outcome and an uncertainty quantification using several metrics, to include the Coefficient of Variation, the skewness and the newly proposed Uncertainty Index to quantify the contribution of every driver in the total uncertainty and its characteristics. We found that the driver with the major impact is forcing, followed by the geometric and the physical drivers, while the computational and the structural drivers have negligible impact, at the main channel. Given that in our era the accuracy of topographic information is high, future research shall focus on forcing and physical driver. • Uncertainties of a 2D hydrodynamic model under steady flow regime are evaluated. • The uncertainty drivers are: forcing, geometric, physical, computational and structural. • A framework coupling sensitivity analysis and uncertainty quantification is developed. • Several metrics were used to quantify the uncertainty in a benchmark study. • Forcing driver is the most important followed by the geometric and physical.