This thesis is in the alternative format and comprises two separate journal articles and an intermediate chapter that together form a coherent research project. Synoptic-scale differences between drylines that produce deep, moist convection and those that do not are determined in two papers. In the first paper, a dataset of drylines within a region of the Southern Great Plains is constructed from surface analyses. Doppler radar, visible and infrared satellite imagery are used to identify convective drylines, where deep, moist convection was deemed to have been associated with the dryline circulation. Composite synoptic analyses of 179 convective and 104 non-convective dryline days reveal previously unidentified differences between convective and non-convective drylines. Convective drylines feature more amplified upper-level flow, associated with a deeper trough in the western US and a stronger downstream ridge than non-convective drylines in the three days preceding a dryline event. As a result of greater poleward low-level moisture transport, significant differences are observed between the composites on the day of a dryline event. The convective composite features greater specific humidity at low to mid levels and higher CAPE than the non-convective composite. A more objective method of analysis, machine learning, is then investigated as a tool for predicting dryline convection and identifying its sensitivity to numerical weather prediction model output. Gradient boosting is applied to model data obtained for 205 dryline days identified during the climatology. The model has a high probability of detecting convective drylines, but labels too many drylines as convective, resulting in false positives. Analysis of model feature importance reveals variable performance varies spatially. The model attaches high importance to the strength of the upper-tropospheric jet over the Rockies when predicting dryline convection. However, instability and mid-level moisture are important variables in locations immediately east of the dryline. These results are consistent with analysis of the composites, and provide evidence that synoptic-scale processes can help determine whether or not a dryline will produce deep, moist convection.
Date of Award | 1 Aug 2021 |
---|
Original language | English |
---|
Awarding Institution | - The University of Manchester
|
---|
Supervisor | Geraint Vaughan (Supervisor) & David Schultz (Supervisor) |
---|
- Thunderstorms
- Climatology
- Convection
- Dryline
Environments Associated with Dryline Convection in the Southern Great Plains
Mitchell, T. (Author). 1 Aug 2021
Student thesis: Phd