The Role of Large-scale Organisation of Convection for Tropical Weather and Climate
Author: Boon Sze Jackson Tan
Publisher:
Published: 2014
Total Pages: 224
ISBN-13:
DOWNLOAD EBOOKTropical deep convection is a critical process in the climate system, influencing quantities such as cloud cover, precipitation and circulation. Despite its importance, convection is poorly represented in global climate models (GCMs). This shortcoming manifests through significant model biases such as in tropical clouds, precipitation and variability on various time scales, and limits our ability to make accurate projections such as changes in rainfall patterns in a warming climate. Due to the coarse resolution of GCMs, convection has to be represented through parametrisation schemes, in which the subgrid-scale behaviour of convection is determined through the resolved large-scale variables. However, our understanding of the relationship between large-scale variables and convection of different degrees of organisation is limited.In this thesis, we investigate the properties and organisation of tropical convection using cloud regimes, with the aim of improving the representation of convection in models. These cloud regimes are derived from the International Satellite Cloud Climatology Project and they identify various states of convection at a resolution comparable to a GCM. These states range from convectively-suppressed environments to a convectively-active atmosphere of congestus clouds and one with cirrus clouds to, most importantly, a regime of organised deep convection.Using these cloud regimes as proxies for different states of convection, we examine its relationship to large-scale variables. Compositing the cloud regimes with traditional measures of convection, we ascertain that they indeed represent different states of convection. Relating them to large-scale variables, we discover that the environments of different convective states are statistically distinct but possess considerable overlap, a result consistent with the stochastic nature of the relationship between convection and large-scale variables. This motivates us to use our knowledge of this relationship to investigate convective organisation in statistical models of varying complexity. After extending the resolution of the regimes from one day to three hours through an innovative technique, we model them statistically using its large-scale environment and infer that a stochastic parametrisation scheme that ignores spatial and temporal memory may struggle to reproduce deep convection organised beyond the grid box and time step. This outcome is worrying because an analysis of the time-series of these regimes suggests that organised deep convection has increased in the past twenty-seven years, driving a corresponding change in the spatial trends of precipitation. Therefore, advancing our understanding of deep convection and addressing deficiencies in its representation in GCMs are of paramount importance.