A study of the interactions between the boundary layer and moist convection combining ARM observations, cloud-resolving and single-column model simulations of TWP-ICE

The Tropical Warm Pool-International Cloud Experiment (TWP-ICE), dedicated to tropical convective systems during the Australian monsoon in the Darwin area, provides a wide range of ground-based and in-situ measurements of surface fluxes, boundary layer and cloud properties. The period of observations, January to February 2006, includes different cloud regimes: intensive oceanic mesoscale convective systems associated with heavy rainfall during the active period, congestus clouds remaining lower than 8 km and much lower precipitation during the suppressed period, fair-weather cumulus during a clear period of three consecutive days and finally continental diurnal deep convection during the break period.

The aim of this study is to better understand the interactions between the boundary layer and moist convection using simulations in which turbulence and clouds are explicitly resolved, and then improve the representation of physical processes involved in parameterizations used in general circulation models. In particular, while previous studies using TWP-ICE observations have largely focused on moist updraft properties, we will consider the impacts of downdrafts on the life-cycle of clouds and on the underlying boundary layer.

First, high-resolution simulations performed with the DHARMA cloud-resolving model (CRM), using bulk and size-resolved microphysics schemes, including several sensitivity tests and compared with available observations and other CRM simulations of the same time period, are used to estimate downdraft properties and study their characteristics during the active and suppressed monsoon periods, their impact on the boundary layer and their potential effect on convection depth. Second, the representation of those downdrafts in parameterizations of deep convection is addressed by running simulations with the single-column version of two different general circulation models: GISS ModelE and LMDz. In ModelE the moist convection parameterization is based on an entraining plume model including a vertical velocity equation and saturated downdrafts. In LMDz the deep convection scheme is based on an episodic mixing and buoyancy sorting approach and includes a representation of unsaturated downdrafts and cold pools. Results are analyzed in order to propose further improvements in the representation of downdrafts in GCMs.