Modeling Convective Mixing: Improvements on the Industry Standard and It's Importance to Biological Production Modeling

By Scott Durski
Edward Levine

Presented at ASLO-99, on February 5, 1999, in Santa Fe, NM

During the LEO-15 Coastal Ocean Predictive Skill Experiments of 1998, improvements were made to the convective mixing models currently available. The figures below show a short summary of the improvements, and how they compare to real data taken from a Naval Underwater Warfare Center Turbulence REMUS cross shore transect.

Color Contours of Vertical Viscosity, with Density Overlay.

The above figure displays color contours of vertical viscosity with temperature overlayed for day 2 of an idealized upwelling simulation. A comparison is made between two vertical mixing schemes. The top frame displays the industry standard Mellor-Yamada level 2.5 closure approximation, while the bottom frame displays a modified version of the Large, McWilliams and Doney K-profile parameterization, which includes the effects of bottom generated turbulence. Our modified scheme is shown to produce viscosity profiles quite similar to the Mellor Yamada scheme in the offshore region, but differs significantly from it in the frontal region.

Convective Mixing During Moderate Upwelling

The above plots explain the high viscosities displayed on the previous slide. The frames display across-shore velocity (top left), density(top right), vertical viscosity (bottom left) and vertical velocity(bottom right). The high vertical mixing in the frontal region is caused by interior convection. The pycnocline is advected upward in the near shore region from a depth of onshore flow, to a depth in which offshore flow occurs. Thus unstable stratification is generated in the interior and convective mixing occurs. The magnitude of the vertical viscosity predicted by the model both here and offshore of the front agrees well with turbulence measurements made off the coast of New Jersey during the summer of 1998. These turbulence measurements are shown below.

Mixing Parameter Data from the NUWC Turbulence REMUS at 5 Meter Depth

The Naval Underwater Warfare Center (NUWC) Turbulence REMUS data acquired during an upwelling event on July23, 1998 is shown above. The four parameters shown were calculated from REMUS-based turbulence data acquisition in the upwelling gyre center (pink area/near shore) and in the offshore leg (blue region). The vertical eddy viscosity changes at the 5m level across the upwelling front agree reasonably well with idealized predictions of the modified KPP model for the case of ultilzing the Large et al (1994) model for the surface and bottom boundary layers, and including convection.

Comparison of Mixing Schemes with a Simple Active Biological Tracer

The above figure shows the potential importance of vertical mixing parameterizations to the ecosystem. Each model shows data from days 2, 4, 6, and 8 of an upwelling event. The color contours above show the time evolution of a simple biological tracer during a modeled upwelling event. The tracer represents a primary producer. The growth of the biological tracer is specified as dependent on three factors; nutrients (present only in the bottom layer waters initially), short wave radiation and intensity of vertical mixing. The mixing dependence is meant to simulate the effect of reduced grazing in highly turbulent conditions.

The two vertical mixing schemes are found to produce dramatically different concentrations of this tracer by day four. It is interesting to note that the physical fields (such as temperature) show far less sensitivity to the vertical mixing parameterization. Consideration of biological fields draws out the importance of properly parameterizing vertical mixing in such a situation.