The mirage of the silver bullet: Exploring the limitations of high-resolution data in flood model validation

Validation of inundation models presents persistent challenges, particularly in urban floodplains where structural complexity often exacerbates discrepancies between model outputs and observation datasets. Findings from the present study suggest that while availability of high-resolution field data can enhance validation efforts, it is not the final piece of the puzzle. The study highlights the (often inseparable) complex combination of observed data limitations, model uncertainties, and structural discrepancies between model and observed datasets, which strongly influence validation outcomes. Using the July 2021 Vesdre Valley flood in Belgium as an illustrative framework, the research evaluates the performance of a high-resolution 2D hydrodynamic model (WOLF). The unprecedented detail of the post-flood survey provides a unique opportunity for rigorous validation. Four so-called ‘reconciliation methods’ are explored to address structural discrepancies (in post-processing) between observation data and computational outcomes, highlighting the effect of the choice of method. At the highest model resolution (2 m), the critical success index (= 0.86) indicates strong spatial agreement across sectors, with a mean absolute error of 0.56 m in modelled maximum flood depths. By proposing and critically evaluating various methods for reconciling inherent differences between observational and computed datasets, this study highlights the complexity of model validation beyond data availability. Additionally, it offers recommendations for refining post-flood survey methodologies to minimise uncertainties associated with the validation process.