The effect of vertical mixing on along channel transport in a layered flow

Abstract

The Columbia River has a large, biologically productive estuary whose ecosystem depends on the balance of salt and fresh water. Outflow from the river also forms a vast buoyant plume which affects circulation for hundreds of miles along the coast. Both the estuarine salt balance and the initial state of the plume are determined by flow through the narrow entrance channel. I have made a three-part study of the effects of interfacial turbulence and bottom friction on along-channel transport through the Columbia River entrance channel.My observations in the Columbia River entrance channel show that both interfacial mixing and bottom friction significantly affect circulation. The pycnocline is thinned by lateral advection on flood and thickened by vertical mixing on ebb and flow. On late flood, the pycnocline is close to the surface and quite thin; on late ebb, its center is below mid-depth and it fills 3/4 of the water column. Bottom friction retards the near-bottom currents, so early flood currents are strongest at mid-depth, and peak flood currents are strongest at the surface. At peak ebb and peak flood, salinity transport is strongest at mid-depth.A two-layer time-dependent model (Helfrich, 1995) simulates along-channel currents and layer thicknesses. By assuming a near-critical bulk Richardson number, I estimated the pycnocline thickness from the two-layer model results. Bottom friction raises the pycnocline and causes tidal variations in vertical shear, which drive the changes in pycnocline thickness. This model replicates the observed pycnocline quite well, but cannot simulate mid-depth currents.I created a new three-layer time-dependent model, in which the middle layer represents the pycnocline. Mixing of salt and fresh water creates water of intermediate density, which is modeled as entrainment from the top and bottom layers into the middle layer. This model simulates along-channel circulation at all stages of the tide, including the mid-depth maximum at early flood. It also simulates the vertical distribution and tidal average of salinity transport. For the best fit to observations, the three-layer model requires significantly more bottom friction than the two-layer model; this is consistent with the formulation of the bottom roughness coefficient.