MASS Seminar - Christian Jakob (Monash)
Date Time Location
June 2nd, 2014 12:00pm-1:00pm 54-915
Title: Insights into tropical convection and its parametrisation gleaned from radar and satellite observations

Christian Jakob1, Laura Davies2, Vickal Kumar1, Karsten Peters1, Jackson Tan1

1ARC Centre of Excellence for Climate System Science, Monash Univeristy, Melbourne, Australia
2The University of Melbourne, Melbourne, Australia

Abstract: The representation of atmospheric convection in weather and climate models has been a challenge since their inception and it remains the Achilles heel of global modelling. The fundamental underpinning assumptions in representing convection in coarse-resolution models date back to the late 1960s and early 1970s and are still widely applied today. Arguably, our ability to observe convection and the large-scale environment in which it exists - both fundamental in tackling the “parametrisation problem” - have evolved significantly since then.

In this presentation we use modern observations, ranging from weather radars to global satellite observations to re-analyses to revisit the basic behaviour of tropical convection in relation to its large-scale environment. We will show that the overall amount of rainfall from convection is strongly determined by the area covered by convective clouds, and less by the rainfall intensity produced in them. We will also demonstrate that relating area-average convection to stability measures proves difficult. Finally, we quantify the stochastic component in the relationship between large and small scales in convecting atmospheres, showing that convection the degree of stochastic behaviour is a function of the large-scale state itself, with weaker forcing being associated with more strongly stochastic behaviour.

On larger scales, we find that convective features are strongly organised beyond the scale of a single GCM grid box. Using simple statistical models we conclude that future treatments of convection, which must include a stochastic component, must also include memory in time and ideally some knowledge of the convective behaviour in the surrounding of a single grid area.

Having identified key relationships between large and small scale sin convecting atmosphere we conclude by outlining a new approach to cumulus parametrisation making use of stochastic cloud models.

Speaker's website: http://users.monash.edu.au/~cjakob/