Sensitivity Analysis of Modelled Flood Inundation Extents over Hawkesbury–Nepean Catchment

Abstract

Rainfall runoff and topography are among the major factors controlling the accuracy of modelled riverine inundation extents. We have evaluated the sensitivity of both these variables on a novel 1-D conceptual flood inundation model employing Height Above Nearest Drainage (HAND) thresholds within sub-catchment units called Reach Contributing Area (RCA). We examined the March 2021 flood extent over the Hawkesbury–Nepean Valley (HNV) with 0.05′ gridded runoff derived from the Australian Water Resources Assessment (AWRA) modelling framework. HAND thresholds were enforced within each RCA using rating curve relationships generated by a modelled river geometry dataset obtained from Jet Propulsion Laboratory (JPL) and by modelling Manning’s roughness coefficient as a function of channel slope. We found that the step-like topographic nature of HNV significantly influences the back-water effect within the floodplain. At the same time, the improved accuracy of the GeoFabric Digital Elevation Model (DEM) outperforms SRTM DEM-derived flood output. The precision of HAND thresholds does not add significant value to the analysis. With enhanced access to river bathymetry and an ensemble point-based runoff modelling approach, we can generate an ensemble runoff-based probabilistic extent of inundation.