Development of a one-dimensional version of the Hirlam-model in Sweden

Background:

This model has been run operationally for about nine years now.

Mainly for TAF-production, but also for other types of site-specific forecasting.

The model is regarded as a useful tool in the forecast production, but of course there are occasions where it is not performing that well.

The most problems lies of course in predicting the boundary layer structures (fog/low clouds).

During the years the model has been updated at a number of occasions with new parameterisations developed within the Hirlam project.

Development of a one-dimensional version of the Hirlam-model in Sweden

New update for the model:

Surface parameterisation.An updated version of the ISBA surface scheme will be introduced to the 1D-model.

In the original scheme each gridpoint is divided in five different tiles (water, ice, open land, open land with low vegetation and forest).

For each of these tiles there is a separate calculation of all the surface fluxes of heat, moisture and momentum. The different fluxes are then area-weighted together to give the model the lower boundary conditions.

The ISBA-scheme has not performed that well when the ground is covered with snow, so the scheme has recently been revised in Sweden with a new treatment of the snow.

This is done by introducing a sixth tile; snow. The snow-covered parts of the tiles open land and open land

with low vegetation are transferred to this snow-tile and separate calculation is done for this new tile.

Development of a one-dimensional version of the Hirlam-model in Sweden

Surface parameterisation cont.

Tests have shown that with this scheme it is possible for the model to better describe the extremely low temperatures that sometimes occurs over the snow-covered parts of northern Scandinavia.

Turbulence

For vertical diffusion processes in Hirlam there is at present a scheme that treats the turbulent kinetic energy (TKE) as a separate variable in the model.

In this scheme the vertical mixing of the variables, temperature, moisture and cloud water is treated separately.

In an updated moist variant of the scheme the mixing is done on the variables liquid potential temperature and total water content.

Development of a one-dimensional version of the Hirlam-model in Sweden

Turbulence cont.

These variables are conserved under non-precipitating moist adiabatic mixing.

The primary benefit is that in-cloud buoyancy effect on TKE is incorporated so that cloudy turbulence is also properly predicted.

The cloudy profiles are well mixed, i.e. they follow moist adiabats.

This new moist turbulence scheme requires a very high vertical resolution, initially about 150 levels will be used with a 30-m resolution in the boundary layer.

These two significant upgrades to the one-dimensional model will be implemented shortly and forthcoming tests will show how much better the model will be on forecasting fog and low clouds.

New Hirlam 1D-modelStatus report

Dataassimilation (ISBA) using observed t2m and q2m.– Same as 3D-model.– Gives new values of surface temp and moisture.– Seems to be working well so far.

Using observation to modify initial profile.– How to relate observed cloud amount with cloud

water and specific humidity in the model?– Cloud tops?

New Hirlam 1D-modelStatus report

New surface scheme with special snow treatment.– Seems to be performing well.– Snow evaporation?– A bit cold on sunny spring days.

New turbulence scheme.– Using 140 vertical levels.– Still some tuning to do.– Problems with partial cloudiness.

New Hirlam 1D-modelVertical levels

New Hirlam 1D-model