ORCID

https://orcid.org/0009-0009-1061-427X

Date of Award

Summer 2024

Language

English

Embargo Period

8-6-2024

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

College/School/Department

Department of Atmospheric and Environmental Sciences

Program

Atmospheric Science

First Advisor

Ryan Torn

Committee Members

Andrea Lang, Daniel Keyser, Lance Bosart

Keywords

Arctic cyclone, numerical weather prediction, cyclone predictability, ensemble sensitivity analysis

Subject Categories

Atmospheric Sciences | Meteorology

Abstract

Arctic cyclones (ACs) can transport warm, moist air into the Arctic region, which combined with strong winds may lead to rapid declines in sea ice during the summer. As a consequence, accurate sea ice predictions during the summer may rely on being able to predict cyclone-related wind speed and direction, which critically depends on the cyclone track and intensity. Despite this, there are relatively few studies that have documented the predictability of ACs during the summer, beyond a few case studies. In addition, there has been no extensive comparison of whether these cyclones are more or less predictable relative to comparable midlatitude cyclones, which have been studied in greater detail.

The first part of this thesis creates a climatology of comparable, long-duration, intense Arctic and Atlantic basin cyclones to compare the predictability of Arctic and Atlantic position and intensity forecasts over a large number of cases using the Global Ensemble Forecast System Reforecast V2. Using standard deviation (SD) and root mean square error as a proxy for predictability, Atlantic cyclone position is characterized by higher predictability relative to comparable ACs, but intensity predictability is higher for ACs. In addition, storms in both basins characterized by low ensemble SD and predictability are found in regions of higher baroclinicity than storms characterized by high predictability. There appears to be little, if any, relationship between latent heat release and precipitable water and predictability.

The source of position and intensity forecast variability of two Arctic and two Atlantic basin cyclones, similar to those analyzed in the climatology, is diagnosed using ensemble-based sensitivity analysis applied to Model for Prediction Across Scales ensemble forecasts. Results suggest that the position and intensity variability of these storms are largely associated with both upstream and downstream uncertainty of meso-scale features embedded within the larger-scale potential vorticity (PV) features, particularly downstream ridge building, in both regions associated with cyclogenesis. Further, the intensity variability of the ACs is associated with uncertainty to thermal boundaries near the cyclone or along the sea ice edge, whereas the Atlantic basin cyclone intensity variability is sensitive to low-level temperature and moisture along the polar front proceeding the cyclones. Further, AC intensity variability appears to be most sensitive to upper-tropospheric variability earlier in the forecast, transitioning to lower tropospheric variability later in the forecast, whereas results are less clear for Atlantic basin cyclones.

An additional AC that was observed during the THINICE 2022 field campaign is analyzed to compare the optimum metrics applied to sensitivity analysis to assess the source of forecast error of European Centre for Medium-Range Weather ensemble forecasts. Sensitivity of cyclone position variability (along the major axis direction) and 500-hPa geopotential height variability within a domain encompassing a tropopause polar vortex is compared, revealing similar sensitive regions primarily to the position of an upstream PV trough and amplitude of a downstream PV ridge. However, comparing sensitivity of cyclone intensity (minimum sea level pressure) variability and low-level wind variability encompassing a low-level jet associated with the AC, are less similar. Sensitivity of low-level wind variability reveals more systematic uncertainty associated with the upper-level PV field and a low-level thermal boundary.

License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Share

COinS