Data and code for: From roughness uncertainty to model-structure effects in 2D flood inundation models: a Monte Carlo comparison of SFINCS and HEC-RAS

This dataset contains the model setups, raw simulation outputs, tabular data and analysis code required to reproduce all results reported in: Navas, S., Galán, Á., & del Jesus, M. (2026). From roughness uncertainty to model-structure effects in 2D flood inundation models: a Monte Carlo comparison of SFINCS and HEC-RAS. Environmental Modelling & Software (submitted July 2026). One thousand Monte Carlo hydraulic simulations were performed with each model — 1,000 with SFINCS and 1,000 with HEC-RAS — using stochastic Manning roughness coefficient combinations for nine land-use classes in the Besaya River floodplain at Corrales de Buelna, Cantabria, Spain (T = 500-year flood event). Five simulations (0.5%) affected by numerical non-convergence were removed, leaving 995 valid paired simulations for analysis. The dataset includes: Tabular inputs: Manning roughness reference values, fitted distributions, 1,000 × 9 Manning combination matrix, T=500 inflow hydrograph. Tabular outputs: per-simulation SFINCS and HEC-RAS statistics, matched 995-pair ensemble, anomalous simulation diagnostics. Raw maximum-depth GeoTIFF rasters for all 1,000 simulations from each model (~290 MB total, 5 m resolution, EPSG:25830). Model setup files for HEC-RAS 6.6 (~1.4 GB) and SFINCS v2 (~7 MB). Seven Jupyter notebooks reproducing the full analysis pipeline. Publication-quality figures (PDF) for all main and supplementary figures. The tabular data in data/ are sufficient to reproduce all tables and figures without rerunning the hydraulic models. Raw simulation rasters are included for spatial reproducibility. Analysis code uses the pyhydra Python package: https://github.com/navass11/pyhydra.