Development of a new mountain drag parametrization scheme

Mountain drag schemes in NWP and Climate models are known to suffer from a number of deficiencies. For example model stability problems often arise as a result of significant mountain wave accelerations being applied over a single model level, an approach which is somewhat unphysical as recent numerical modelling studies have indicated that mountain wave breaking accelerates the flow over a depth that is related to the vertical wavelength of the mountain wave. Other deficiencies include the assumption that waves react instantaneously to changes in the atmospheric profile, and the assumption that waves propagate purely vertically.

As a first step in addressing some of these deficiencies a new mountain drag scheme has been developed and implemented in the Met Office Unified Model, which is used for forecasting on time-scales from a few days to hundreds of years. The new scheme is largely based on the Lott and Miller (1997) scheme but includes some new features, such as:

1. The application of gravity-wave drag over a vertical gravity wave-length rather than over a single model level.

2. A Froude number dependent low-level drag coefficient

3. An introduction of the concept of gravity wave propagation times into the scheme.

4. A new choice of averaging depth for the low-level flow.

5. An attempt to represent the effects of waves which do not propagate vertically and therefore exit the grid-column laterally.

All of these changes are physically motivated based on recent research, but are also chosen to be beneficial to the numerical stability of the scheme.

In this talk the motivation for developing a new scheme and the science behind the scheme will be presented. Comparisons will be made between the predictions of the scheme, gravity wave observations from a limb-sounding satellite instrument and a month long series of high-resolution (4km) simulations of flow across the southern Andes (a well-known mountain wave hot-spot region).