Stochastic and deterministic multi-cloud models for tropical climate dynamics

Boualem Khouider, University of Victoria

11:45 - 12:30

The tropical belt between roughly 30° South and 30° North receives most of the solar energy absorbed by the Earth surface. This energy is then redistributed poleward through various oceanic and atmospheric

trap most of Earth’s longwaves through the phenomenon known as the greenhouse effect. Cloud condensation is accompanied by latent heat release, which drives local and large scale atmospheric circulation patterns. This results in complex two-way interactions between clouds and water vapour and the climate system, which hinders the ability of climate models to represent clouds and the highly turbulent, unresolved processes of convection and precipitation in the tropics.

The rotation of Earth around itself induces an important force on its atmosphere and oceans known as the Coriolis force, resulting in some important circulation patterns such as the trade winds and the jet stream. Because the Coriolis force vanishes at the equator, this later acts as a waveguide for many important atmospheric and oceanic waves. Convection in the tropics is organized into a hierarchy of cloud clusters and superclusters ranging from the convective cell of a few kilometres to synoptic and planetary scale wave disturbances, with a global impact, such as the Madden Julian oscillation (MJO) and the monsoon intra-seasonal oscillations (MISO). The MJO is a planetary scale disturbance in winds, cloud cover, and rainfall which circulation patterns. The heat absorbed by the ocean surface is transferred to the troposphere through the phenomenon of convection where the lighter--moist and warm air rises, condenses water vapour and forms clouds and rain. In return, clouds and water vapour affect greatly the radiation budget as they radiate a significant portion of solar energy back to space and develop over the Indian Ocean warm pool and slowly propagates eastward at roughly 5 m/s. Phenomena such as El-Nino, hurricanes, extreme droughts and floods, both in the tropics and mid-latitudes (particularly the Pacific Northwest region), are linked to the MJO.

Most global climate models (GCMs) simulate poorly these convectively coupled wave phenomena and as such they lack fidelity in representing the tropical weather and climate. Due to computing power limitations GCMs use grid resolutions on the order of 100 km. Clouds and convection processes are thus represented through subgrid models known as parameterizations. Traditional parameterizations are to blame for the poor performance of GCMs in the tropics. This is mainly because they are based on the quasi-equilibrium theory; convection is assumed to instantaneously consume atmospheric instability and restore equilibrium, i.e, within one time step of the GCM, which is unphysical and more so for finer GCM resolutions around 10-50 km.

In this lecture, I will present a hierarchy of stochastic and determinsitic multicloud models for the parameterization of organized tropical convection and convectively coupled tropical waves. The multicloud models rely on the observed three cloud type paradigm which characterizes multiscale tropical convective systems. In particular, the stochastic multicloud models rely on a lattice multi-particle interacting system to represent the missing variability in climate models due to the unresolved process of convection and deviate further from the quasi-equilibrium loophole.

Each lattice site is either occupied by a certain cloud type or is clear sky. Random transitions from one cloud type to another are based on conditional probabilities, depending on whether the environment is favourable for convection or not, i.e, atmospheric instability and mid-tropospheric moisture. After coarse-graining, the lattice model reduces to a stochastic birth death process for the cloud area fractions, which can be easily coupled to a GCM without any significant computational overhead. Both reduced model linear and nonlinear analysis and GCM simulation results will be presented to showcase the ability of the multicloud models to represent various tropical weather and climate phenomena such as the MJO and MISO.