Publication Date

2015-12-04

Availability

Open access

Embargo Period

2015-12-04

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PHD)

Department

Meteorology and Physical Oceanography (Marine)

Date of Defense

2015-10-30

First Committee Member

Brian E. Mapes

Second Committee Member

Chidong Zhang

Third Committee Member

Benjamin Kirtman

Fourth Committee Member

Anthony D. Del Genio

Abstract

Interactions between atmospheric deep vertical convection and larger-scale flow have been examined in three diverse modeling frameworks: a traditional climate model with cumulus parameterization scheme, an extremely computation intensive global convective-resolving model, and a simplified global primitive equation model with a linearized anomaly convective scheme. The first model is operational but has some chronic problems that call for more research, the second is promising but produces vast amounts of output data that are hard to interpret or even handle, and the third is satisfying for interpretation of the interaction processes, but illustrate some continuing challenges of atmospheric modeling, ultimately because this important process occurs in convective cells much smaller than the size of our planet. More specifically, in chapter 2, hindcast bias growth in the Climate Forecast System (CFS) is analyzed. Errors in the Cold Tongue – Intertropical convergence Complex (CTIC) are apparently initiated by the convection, since they appear very rapidly. The errors are interpreted as indicating weak sensitivity of the convection scheme to moisture, a common problem in climate models. This initial convective error apparently seems coupled feedback processes which gradually spread bias errors to other regions and components of the CFS. In chapter 3, explicitly simulated tropical convective rain events were examined, from a pioneering global nonhydrostatic 5-km mesh model (NASA GEOS-5). The data examined are like perfect observations – they are samples at full resolution – but unlike observations the values are know exactly. Composite profiles of larger-scale temperature and humidity evolution across the rain events show good agreement with published composites of observations, but not every case has all the composite characteristics. Diverse interaction mechanisms between convection and its environment are seen in the various cases, as in nature, indicating the model’s realism in that broad sense. In chapter 4, a linear matrix formalism for convection-large scale state interaction is explored. Heating and moistening rate anomalies are cast as weighted sums of temperature and moisture anomalies, based on pioneering work of a collaborator, Dr. Z. Kuang of Harvard. His matrix M is tested as a diagnostic model to help interpret the composite data from chapter 3. The composite humidity anomalies are found to be more consequential than temperature in shaping the evolution of convection, based on M’s weighting factors. The static stability implied by composite-averaged cool air at the surface, which is a consequence rather than cause of heavy rain, reduces the M-predicted rainrate, indicating one of the challenges of framing causality in terms of large-scale variables alone. In chapter 5, M is applied as an anomaly convection scheme in a global primitive equation model. Five experiments are described with modified versions of M. The modifications are motivated by a wish to understand the roles of M’s eigenmodes, and on the hypothesis from chapter 2 that free tropospheric moisture sensitivity is an important aspect of convection schemes. The experiments show substantial differences in large-scale M-coupled phenomena, although these first-ever model simulations are too unrealistic to judge in terms of observations. Several general conclusions emerge. Clearly the treatment of convection is important to large-scale climate and weather, especially in the tropics. Explicitly resolving cloud systems appears to be a promising approach as computation power grows, but will not be affordable for all climate problems. Still, output datasets from such models can be a useful resources for trying to improve parameterizations through better understanding of interaction processes as played out in the range of weather scenarios that occur in the tropics. Based on the linearized matrix approach, some clean interpretations can be deduced. For example, moisture sensitivity of convection does indeed appear to be a key issue for convection interactions, and having a new model where the sensitivity can be cleanly modified and tested could lead to knowledge that may feed back into improvements in conventional cumulus parameterizations.

Keywords

convection-environment interaction; cumulus parameterization; cloud-resolving global model; mesoscale convective system; linearized matrix; convective sensitivity

Share

COinS