Towards Model Resolution-Invariant Aerosol Activation for the UK Met Office Unified Model

The amount of solar radiation reflected by clouds depends on droplet size and number concentrations. Parameterizations in weather and climate models form (activate) droplets by the adiabatic cooling generated when a moist air parcel rises. Hence, accurate representation the vertical wind velocities (updrafts) in a model is critical for accurate representation of droplet activation process.

Updrafts in a model depend on amount of small-scale turbulence resolved by the boundary layer dynamics parameterization in the model, which in-turn depend on the horizontal grid spacing of that model. Updrafts can vary within a model grid. A higher grid-resolution resolves a higher amount of updrafts (more spread about the mean) and vice versa. To account for this variability, models can either use a turbulent kinetic energy (TKE) based parameterization or scale-up the grid-average updrafts to improve representation of droplet activation.

The UK Met Office Unified Model (UM) has, in theory, the capability to simulate atmospheric evolution, including clouds and aerosols, at any horizontal grid resolution from around 100 m to around 100 km. The double moment cloud microphysics scheme in the UM uses a TKE based parameterized updraft: Wact = Wavg+√TKE . Here Wact, the updraft used in the activation scheme, is a calculated by adding the square root of TKE to the grid- average updraft, Wavg. However, this crude parameterization does not perform very well: different model grid resolutions produce different cloud droplet concentrations. To test the accuracy of, and potential improvements to, this parameterization, we setup different regional model domains near Graciosa Island and Ascension Island, locations of recent field campaigns (ACE-ENA and CLARIFY). Domains with 100 m, 300 m, 1 km, 3 km, 10 km, and 30 km grid spacing were used.

Our simulations demonstrate the deficiencies in this parameterization. As expected, the Wavg term was highest and the √TKE term was lowest for the 100 m model. However, the Wact term differs between different model resolutions, which explains the large discrepancies in droplet number concentration. Following earlier work which only tested the cloud-resolving scale below 1 km grid spacing, we hypothesize that Wact should depend on grid-resolution and the planetary boundary layer height. We test different refinements to the parameterization above based on this hypothesis, and present an improved method which we believe could be applied in other atmospheric models.