ORCID

https://orcid.org/0000-0001-7881-3683

Date of Award

Summer 2024

Language

English

Embargo Period

7-31-2024

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

College/School/Department

Department of Atmospheric and Environmental Sciences

Program

Atmospheric Science

First Advisor

Robert Fovell

Committee Members

Ryan Torn, Justin Minder, David Fitzjarrald

Keywords

Fire weather, Downslope windstorm, Numerical Weather Prediction, WRF

Subject Categories

Atmospheric Sciences

Abstract

Downslope windstorms in the western United States often contribute to the ignition and ex- acerbation of existing wildfires due to their hot, dry, and gusty nature. Accurate forecasts of these windstorms are imperative for mitigating wildfire risk. However, forecasting downslope windstorms remains challenging due to their sensitivity to environmental conditions.

The Sundowner Windstorms of the Santa Barbara, California region have been asso- ciated with many large wildfires in the area. This study analyzes special observations from the Sundowner Wind Experiment field campaign to better understand the diurnal cycle of these windstorms. Analysis of the strongest windstorm during the campaign indicates that the diurnal cycle of the marine boundary layer significantly influences the environment prior to windstorm onset. High-resolution simulations of this event further reveal that the marine layer air acted to delay the onset of downslope winds. The role of the marine layer was fur- ther tested using semi-idealized simulations to simulate an early sunset, leading to the early onset of downslope winds. Finally, the predictability of Sundowner windstorms was found to be limited using current forecasting tools. However, a simple statistical model demonstrated utility in providing accurate predictions of strong downslope wind occurrences.

Additionally, the downslope windstorm associated with the 2021 Marshall Fire in Boul- der, Colorado, also exhibited limited predictability from operational forecasts. This limited predictability was traced to subtle shifts in upper-level winds, which controlled whether a strong downslope windstorm occurred. A novel approach of downscaling coarse synoptic scale forecasts, focusing on the period surrounding the windstorm, was found to be as ac- curate or better than traditional downscaling of the full forecast. Furthermore, we show that a Machine Learning Weather Prediction model, such as Google’s GraphCast, can be competitive with traditional operational forecasting models.

License

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

Download

Share

COinS