Logo Logo
Help
Contact
Switch language to German
Organization of tropical convection
Organization of tropical convection
Atmospheric moist convection is an important source of uncertainty in current climate models. Compared to the resolved spatial scales, convective updrafts are small scale processes which, however, have an important influence on the atmospheric state. To determine their effect on the resolved scales, clouds are usually approximated as non-interacting despite their ability to trigger or favor new convection in their surrounding. In particular, convectively induced cold pools trigger new updrafts along their boundaries and thus induce correlations between successive generations of clouds, while convectively induced humidity perturbations accumulate over many generations and cause self-aggregation of convection on time scales much longer than the individual cloud life times. While the cold pool triggering and the instability of the atmosphere to humidity perturbations have been studied in detail, there is a lack of understanding of how these interactions affect the large scale properties of a convective field and thus impact the resolved scales. In this thesis we will introduce minimally simple models to represent the effects of cold pool triggering and convectively induced humidity perturbations and relate them to models known from statistical physics to determine how these processes lead to the formation of large scale structures. We start by introducing a modified percolation model which can reproduce the observed size distributions of shallow cumulus clouds. Analyzing size distributions using satellite data, we argue that it is the merging of smaller subclouds to larger clouds which leads to power law size distributions over a range of scales which increase as the total cloud fraction approaches a critical value. Modifying the standard percolation model to include clustering leads to quantitative agreement. The second model describes the spreading of convective activity using a 2D lattice model based on directed percolation. Motivated by the cold pool induced spreading of convective activity we argue that the ability of convection to trigger new convection can explain the observed continuous phase transition in precipitation strength. While we find that the model can reproduce the spreading of convection in an atmospheric model simulation, we also propose a more direct test for this hypothesis. The third model addresses the spatial evolution of self-aggregation, which organizes convection on long space and time scales. We argue that the convectively induced humidity perturbations lead to a diffusive spatial interaction and conclude that the upscale growth of moist and dry regions can be described as a phase-separation process called coarsening. Comparison with previous studies shows that coarsening can explain the frequently noted domain shape and size dependence and, at least initially, the upscale growth of moist and dry regions. Using models known from statistical physics we find that we are not only able to qualitatively describe the emergence of large scale properties but that, due to universal properties of these models, we are able to make quantitative predictions.
Not available
Windmiller, Julia Miriam
2017
English
Universitätsbibliothek der Ludwig-Maximilians-Universität München
Windmiller, Julia Miriam (2017): Organization of tropical convection. Dissertation, LMU München: Faculty of Physics
[img]
Preview
PDF
Windmiller_Julia.pdf

15MB

Abstract

Atmospheric moist convection is an important source of uncertainty in current climate models. Compared to the resolved spatial scales, convective updrafts are small scale processes which, however, have an important influence on the atmospheric state. To determine their effect on the resolved scales, clouds are usually approximated as non-interacting despite their ability to trigger or favor new convection in their surrounding. In particular, convectively induced cold pools trigger new updrafts along their boundaries and thus induce correlations between successive generations of clouds, while convectively induced humidity perturbations accumulate over many generations and cause self-aggregation of convection on time scales much longer than the individual cloud life times. While the cold pool triggering and the instability of the atmosphere to humidity perturbations have been studied in detail, there is a lack of understanding of how these interactions affect the large scale properties of a convective field and thus impact the resolved scales. In this thesis we will introduce minimally simple models to represent the effects of cold pool triggering and convectively induced humidity perturbations and relate them to models known from statistical physics to determine how these processes lead to the formation of large scale structures. We start by introducing a modified percolation model which can reproduce the observed size distributions of shallow cumulus clouds. Analyzing size distributions using satellite data, we argue that it is the merging of smaller subclouds to larger clouds which leads to power law size distributions over a range of scales which increase as the total cloud fraction approaches a critical value. Modifying the standard percolation model to include clustering leads to quantitative agreement. The second model describes the spreading of convective activity using a 2D lattice model based on directed percolation. Motivated by the cold pool induced spreading of convective activity we argue that the ability of convection to trigger new convection can explain the observed continuous phase transition in precipitation strength. While we find that the model can reproduce the spreading of convection in an atmospheric model simulation, we also propose a more direct test for this hypothesis. The third model addresses the spatial evolution of self-aggregation, which organizes convection on long space and time scales. We argue that the convectively induced humidity perturbations lead to a diffusive spatial interaction and conclude that the upscale growth of moist and dry regions can be described as a phase-separation process called coarsening. Comparison with previous studies shows that coarsening can explain the frequently noted domain shape and size dependence and, at least initially, the upscale growth of moist and dry regions. Using models known from statistical physics we find that we are not only able to qualitatively describe the emergence of large scale properties but that, due to universal properties of these models, we are able to make quantitative predictions.