Implementation of an E–ε Parameterization of Vertical Subgrid-Scale Mixing in a Regional Model

View More View Less
  • 1 Naval Research Laboratory, Monterey, California
© Get Permissions
Full access

Abstract

An E–ε parameterization of subgrid-scale vertical turbulent mixing has been installed in NORAPS (Navy operational Regional Atmospheric Prediction System). The 1.5-order parameterization uses full prognostic equations for turbulence kinetic energy E and dissipation ε with no mixing length l assumption. A stable numerical method has been developed to integrate the two prognostic equations; this method has time and memory requirements that are similar to first-order K-theory turbulence parameterization and avoids numerical instabilities reported with El (Mellor–Yamada level 2.5) schemes. The E–ε parameterization produces a more active mixed layer, compared to a first-order K-theory scheme. Improvements are noted in forecasts of mixed-layer depth and near-surface wind speed, with reduction or elimination of spurious noise in the predicted fields of temperature and wind that were related to deficiencies of the first-order K-theory parameterization. In a numerical simulation of the ERICA (Experiment on Rapidly Intensifying Storms over the Atlantic) IOP 5A storm, the E–ε parameterization provides an improved forecast of cyclone central pressure. The better cyclone forecast results primarily from more accurate prediction of wind speed near the surface and in the upper troposphere where first-order K theory may produce unrealistic vertical mixing of momentum and temperature.

Abstract

An E–ε parameterization of subgrid-scale vertical turbulent mixing has been installed in NORAPS (Navy operational Regional Atmospheric Prediction System). The 1.5-order parameterization uses full prognostic equations for turbulence kinetic energy E and dissipation ε with no mixing length l assumption. A stable numerical method has been developed to integrate the two prognostic equations; this method has time and memory requirements that are similar to first-order K-theory turbulence parameterization and avoids numerical instabilities reported with El (Mellor–Yamada level 2.5) schemes. The E–ε parameterization produces a more active mixed layer, compared to a first-order K-theory scheme. Improvements are noted in forecasts of mixed-layer depth and near-surface wind speed, with reduction or elimination of spurious noise in the predicted fields of temperature and wind that were related to deficiencies of the first-order K-theory parameterization. In a numerical simulation of the ERICA (Experiment on Rapidly Intensifying Storms over the Atlantic) IOP 5A storm, the E–ε parameterization provides an improved forecast of cyclone central pressure. The better cyclone forecast results primarily from more accurate prediction of wind speed near the surface and in the upper troposphere where first-order K theory may produce unrealistic vertical mixing of momentum and temperature.

Save