Temporal transferability of spatially derived Manning’s roughness across flood regimes in the Mississippi River

Abstract

Evaluating the temporal transferability of spatially derived roughness across flood regimes is important to balance the use of consistent roughness values for model reliability with adjustments needed to reflect changing conditions and improve model performance. We applied a high-resolution 2D unstructured grid hydrodynamic model to simulate flood events on the Middle Mississippi River focusing on two significant flood years (2011 and 2019) and a year with moderate flooding (2022). Model results were evaluated using gauge data from USGS stations and high-resolution data on flood inundation extents from Planet Labs Dove satellites for 2019 and 2022 to ensure robust evaluation. Simulations for all years used the same set of spatially variable Manning’s roughness (n) values, which remained temporally invariant. The model exhibited excellent performance across all years, with Nash–Sutcliffe Efficiency values of 0.98–0.99 for water surface elevations and 0.92–0.99 for discharge, as well as strong agreement with observed flood inundation extents. Discrepancies in model performance in 2011, particularly in flood peak magnitude and timing, were attributed to substantial unaccounted inflows (e.g., from tributaries) entering the model domain that were not represented in the boundary conditions. The consistently strong model performance using spatially derived roughness over a decade suggests that reductions in performance are more likely caused by hydrologic forcing (missing flow inputs) rather than by temporal changes in roughness. This study highlights the need for a careful assessment of whether sufficient evidence justifies using time-dependent roughness values and advocates against treating Manning’s n as a free, catch-all parameter.