Work

Intercomparison of flood inundation models across land use types and hydrological flood stages

Year: 2026

Type: article

Abstract: Flood Inundation Mapping (FIM) model selection is a key operational decision because accurate, rapid mapping underpins early warning and resource allocation. FIM performance is context-dependent as it can vary with hydrograph phase, land-use/land-cover (LULC), and the evaluation benchmark. Intercomparison studies typically assess a single near-peak snapshot against one reference dataset. We seek to understand the importance of hydrograph phases, LULC, and evaluation dataset on the performance of FIM solvers of varying complexity (TRITON, LISFLOOD-FP, HEC-RAS 2D, ARC-Curve2Flood, and OWP HAND-FIM). We also investigate the accuracy–runtime trade-offs across the five models. We used the flooding event during Hurricane Matthew (2016) in the Neuse River Basin, North Carolina, USA, as a case study. Using high-resolution remote-sensing FIMs, hand-labeled points, and building footprints, we assess model skill across two rising and two falling hydrograph limbs and for LULC types. The results show that, even within a single event, no model consistently performs best across all phases. LISFLOOD-FP ranks highest in three of four flood phases, while TRITON leads during one of the rising limb phases. LISFLOOD-FP was most accurate in vegetated areas, followed by TRITON and HEC-RAS; in agricultural and urban areas, the order reverses, with HEC-RAS surpassing TRITON. Performance also depends on the benchmark used: in the late falling limb, LISFLOOD-FP is most accurate for flooded-building detection, whereas TRITON ranks highest against hand-labeled points. The results provide an important understanding of flood-model behavior and offers insights about designing large-scale intercomparison studies.

Source: Journal of Hydrology

DOI: 10.1016/j.jhydrol.2026.135410