Does the introduction of a simple cloud-aerosol interaction improve the representation of drizzle in the operational Met Office Unified Model?

Over-production of drizzle in cloudy situations is a common problem in NWP models, and significant input from the human forecaster is required to convert the model output into an operational product.

The Met Office Unified Model frequently suffers from excessive drizzle production over the UK, especially within cloudy boundary layers and the model is particularly sensitive to the cloud droplet number concentration. The present formulation has one value of droplet number over land grid points and another over sea. However, this leads to an unphysical sea-land split in light rain and incorrect drizzle rates when a clean air mass passes over the land or a polluted air mass crosses the sea.

In order to improve the representation of drizzle, we are trialling the linkage between the aerosol used for diagnosis of visibility in the model and the cloud autoconversion rate. This has the advantage of being computationally inexpensive, yet gives a reasonable representation of whether the boundary layer is clean or polluted.

As not all aerosol particles become cloud condensation nuclei, some assumptions have to be used to link total aerosol number to cloud droplet number. We will present a simple case, where we assume a clean-polluted split given a total aerosol threshold and more complicated linkages, where the aerosol is allowed to vary with altitude and where the cloud droplet number has a strong dependence on the total aerosol number. We will show that careful choice of cloud droplet number has to be made in order not to interfere with the fog diagnostics calculated within the model.

The results can be evaluated in order to find the best representation. This is done by comparing the model data with rain rate-liquid water path observations in stratocumulus from several field campaigns.