P3.1 Modeling the Interfacial Layer in Large-Scale Models

Wednesday, 9 August 2000
Martin J. Otte, Penn State University, University Park, PA; and J. C. Wyngaard, N. L. Seaman, and D. R. Stauffer

Representing entrainment and the structure of the interfacial layer are among the most challenging aspects of boundary layer modeling. A recent intercomparison of commonly used boundary-layer models found that each had difficulty reproducing entrainment at the top of the PBL, resulting in significant errors in the predicted mean boundary-layer structure. Similarly, researchers have reported that the use of mesoscale-model generated refractivity fields in propagation models has met with only limited success, which they attributed in part to the poor representation of interfacial-layer structure.

Our results suggest that the interfacial layer shares many characteristics of stable PBL flows. For example, an eddy diffusivity relating the turbulent fluxes to the mean gradients exists in the interfacial layer, and it scales with the local rms vertical velocity and buoyancy length scale. Similiar scaling results hold for the scalar variances, structure-function parameters, dissipation and destruction rates, and the entrainment rate.

This scaling yields a closed set of equations for the interfacial-layer structure, given the vertical-velocity variance. We present a model for the vertical-velocity variance based on the moment equations. The modeled variance behaves well, predicting that the vertical-velocity variance scales with the convective velocity scale in free convection and with the velocity jump across the inversion in the case of a neutral, inversion-capped PBL with strong shear.

In weather-forecast models the problem of modeling the interfacial layer is compounded because the region may not be resolved well. The vertical thickness of the interfacial layer may be 100 meters or less, which may be less then the vertical grid mesh of the model. The effect of resolution on the interfacial-layer structure predicted by our model will be examined. For the cases where the interfacial layer is poorly resolved by the forecast model grid, we examine the use of polynomials to represent the vertical structure within the capping inversion to improve modeling the mean interfacial-layer structure.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner