Hydrodynamic and sediment transport tsunami models at the Salmon River estuary, Oregon

Delft3D-FLOW hydrodynamic model nested grids and bathymetry were constructed to propagate simulated tsunamis from the region offshore the Oregon coast to the Salmon River estuary. Four grids in total at 1650 m, 400 m, 50 m, and ~10 m grid-cell resolution, respectively, were created. Bathymetric data from two NOAA NCEI products (Amante and Eakins, 2009; Carignan and others , 2009) were converted from spherical to cartesian coordinates (WGS 84/UTM Zone 10 N [EPSG:32610]) and interpolated onto the grids using the grid-cell averaging function in Delft3D-QUICKIN software.

Static sea surface deformation predicted by the earthquake source models from Witter and others (2013) and Gao and others (2018) were interpolated onto each nested grid using the grid-cell averaging function in the Delft3D-QUICKIN software and used as initial water level conditions in each Delft3D-FLOW model. The boundary condition on the outermost, 1650 m resolution grid is a uniform Riemann-boundary that is weakly reflective to absorb outgoing waves. Boundary conditions along the edges of the 400 m, 50 m, and ~10 m grids were derived from the hydrodynamic results observed in each respective outer nested grid. These hydrodynamic observations were prescribed as Riemann-boundary conditions at six second intervals and generated usingDelft3D nesting functions. The “flood” advection scheme was turned on for all model runs.

A variable resolution, three-dimensional final nested grid at the Salmon River estuary was used to simulate potential sediment transport resulting from tsunami propagation. Grid cell size is, on average, 10 m, with m and n dimensions ranging from 3 meters to 61 meters. Vertical stratification is represented by 10 vertical “sigma” layers, with increasing resolution towards the bed. Bathymetry and topography in Delft3D depth files were adjusted to account for the coastal subsidence predicted by each earthquake source at the Salmon River. A table relating subsidence to each of the earthquake sources can be found in La Selle and others (2024) or the supplemental file in this section of the data release. For each earthquake source scenario, depth files were also adjusted to represent tsunami arrival at MHHW (2.54 m above NAVD88 at the South Beach, OR tide gauge) or the hindcast low tide at 21:00 Pacific Time on January 26th, 1700 CE, 0.53 m above NAVD88 (Mofjeld and others, 1997). Depth file names reflect the tidal adjustment (either “1700tide” or “MHHW”) and the subsidence adjustment in meters. Time zero in the model runs represents the time of earthquake rupture. We use an arbitrary date and time in DELFT3D-FLOW as time zero (2000-01-01 00:00:00). The 1650 m, 400 m, and 50 m grids are run with a timestep of 0.01 minutes (0.6 seconds), for 180 minutes. The 10 m sediment transport grid was run with a timestep of 0.005 minutes (0.3 seconds), for 60 minutes. The sediment transport models were run for 60 minutes because deposit thicknesses did not change appreciably within the estuary after 60 minutes (La Selle and others, 2024).

Maps of erodible sand sources and variable surface roughness in the detailed models were generated by classifying 1-m resolution ortho images of the Salmon River estuary taken in July, 2014 by the National Agriculture Imagery Program (NAIP). Imagery was auto-classified into water, unvegetated sand, marsh vegetation, and forest using the Spatial Analyst Image Classification tool in ArcGIS desktop 10.6. Roughness values were based on suggestions from Bricker and others (2015). The beach, river channels, and offshore areas were assigned a Manning's n roughness coefficient of 0.025 and an erodible bed thickness of 10 m. Vegetated areas of dunes were assigned a Manning's n coefficient of 0.030 and an erodible bed thickness of 10 m. The marsh surface and forest was assigned 0.040 and the bed was modeled as an unerodible surface. The van Rijn (2007) formulations were selected to represent suspended and bedload transport in Delft3D-FLOW.

Files need to run Delft3D-FLOW models were compiled into compressed archives for distribution.