The testing of a convective parametrization scheme in an idealised tropical atmosphere

A cloud resolving model (CRM) of an idealised convective quasi-equilibrium representative of the tropical atmsosphere is used to generate datasets of temperature and humidity against which a convective parametrization scheme can be evaluated. By using an idealised framework, in which radiation and its interaction with convective clouds is neglected, it is hoped to gain a clearer insight into the behaviour of the UK Met Office Unified Model (UM) convective parametrization scheme in tropical latitudes than is otherwise possible from using case studies from observational field campaigns.

The CRM is run with a prescribed time-constant cooling function and a fixed sea-surface temperature (SST) for several days until a quasi-equilibrium climate is attained. The quasi-equilibrium profiles of temperature and humidity are then used to initiate a simplified version of a single column model of the UM (SCM) in which only convective, boundary layer and large-scale cloud parametrizations are included. The SCM is forced with the same prescribed cooling function and fixed SST as the CRM and run for twenty-four hours. Temperature and humidity profiles at the end of the SCM simulation are compared to those from the CRM. The SCM temperature profiles are up to 2K cooler than the CRM temperature profiles in the upper troposphere. This cooling is consistent with the upper tropospheric cold bias seen in tropical latitudes in global UM simulations.

By comparing the properties of the convection simulated in the CRM with that parametrized in the SCM it is seen that the convective cloud tops in the SCM are some 100mb lower than those in the CRM. Further analysis suggests that it is an overdilution of the convective parcels which inhibits the parametrized convection penetrating to higher altitudes. Changes to the convection scheme's entrainment rate in the SCM act to reduce the perceived cold bias by over half. Tests of these changes in the global UM similarly demonstrate a reduction in the tropical cold bias. A further series of quasi-equilibrium CRM and SCM experiments are performed with different fixed SSTs but the same prescribed cooling functions. It is seen that the upper tropospheric cold bias in the SCM temperature profiles decreases as SST increases. The implication that this may have for upper tropospheric temperature increases in climate change experiments is considered.

It is suggested that this above technique can be a powerful tool in rapidly exploring the parameter space of convection parametrization schemes and optimizing their behaviour for the tropical atmosphere.