Discontinuous Galerkin modeling of the Columbia River’s coupled estuary-plume dynamics

Abstract

The Columbia River (CR) estuary is characterized by high river discharge and strong tides that generate high velocity flows and sharp density gradients. Its dynamics strongly affects the coastal ocean circulation. Tidal straining in turn modulates the stratification in the estuary. Simulating the hydrodynamics of the CR estuary and plume therefore requires a multi-scale model as both shelf and estuarine circulations are coupled. Such a model has to keep numerical dissipation as low as possible in order to correctly represent the plume propagation and the salinity intrusion in the estuary. Here, we show that the 3D baroclinic discontinuous Galerkin finite element model SLIM 3D is able to reproduce the main features of the CR estuary-to-ocean continuum. We introduce new vertical discretization and mode splitting that allow us to model a region characterized by complex bathymetry and sharp density and velocity gradients. Our model takes into account the major forcings, i.e. tides, surface wind stress and river discharge, on a a single multi-scale grid. The simulation period covers the end of spring-early summer of 2006, a period of high river flow and strong changes in the wind regime. SLIM 3D is validated with in-situ data on the shelf and at multiple locations in the estuary and compared with an operational implementation of SELFE. The model skill in the estuary and on the shelf indicate that SLIM 3D is able to reproduce the key processes driving the river plume dynamics, such as the occurrence of bidirectional plumes or reversals of the inner shelf coastal currents.

Challenges and Prospects in Ocean Circulation Models

“…Recent progress in developing unstructured models such as ADCIRC (Luettich and Westerink, 2000;Dietrich et al, 2012), FESOM (Wang et al, 2014;Sidorenko et al, 2015;Danilov et al, 2017;Wang Q. et al, 2018), FVCOM (Chen et al, 2003), ICON (Korn, 2017), MPAS (Ringler et al, 2013), and SLIM (Delandmeter et al, 2018;Vallaeys et al, 2018) is making multiresolution approaches more flexible and natural. One challenge is that new users find it difficult to write analysis software, and modeling centers are addressing this problem through the development and distribution of analysis packages in higher-level languages such as Python and Matlab.…”

An implicit wetting–drying algorithm for the discontinuous Galerkin method: application to the Tonle Sap, Mekong River Basin

“…The bed shear stress is evaluated with the help of the Chézy -Manning -Strickler formulation with n being the Manning coefficient. These parameters have been calibrated in SLIM for various domains of interest over the world [10,11,[22][23][24][25][26][27]. Using a spatially variable Manning coefficient could probably improve the quality of simulations.…”

“…The mesh resolution varies from 200 m to 2000 m in order to take into account the wide range of physical processes occurring in the computational domain. The application of this multi-scale mesh allows to simulate both small-scale and large-scale processes within a single model without nesting [26]. Our model does not take into account the interaction between groundwater and surface water because [34] it is estimated that groundwater discharge to the lake is approximately 4 -8 km 3 per year, accounting for only 5 -10 % of total annual inflow.…”

“…Unlike more standard numerical methods, such as finite volumes, it introduces a minimal amount of numerical dissipation and thus preserves small-scale flow features such as recirculation eddies. The model has been applied to various domains such as Lake Tanganyika, Africa [21], the Great Barrier Reef, Australia [22], the Scheldt estuary, Belgium and Netherlands [23], the Mahakam Delta, Indonesia [24,25], the Columbia River, USA [26], and various lakes of Titan, Saturn's moon [27]. Flow field, sediment transport, water renewal time scales and ecological issues were tackled.…”

Simulating Lake Tanganyika's hydrodynamics under a changing climate

“…The 3D component of the Second-generation Louvain-la-Neuve Ice-Ocean Model (SLIM 3D, www.slim-ocean.be) is a baroclinic model that solves the hydrostatic flow equations under the Boussinesq approximation by means of a discontinuous Galerkin finite element method on an unstructured grid (Blaise et al, 2010; Kärnä et al, 2013;White et al, 2008). It has been applied to coastal waters such as the Burdekin plume in the Great Barrier Reef (Delandmeter et al, 2015), the Columbia River region of freshwater influence (Vallaeys et al, 2018), the Congo estuary (Vallaeys et al, 2020 (submitted)) and Lake…”