Date of Award

August 2021

Degree Type

Thesis

Degree Name

Master of Science

Department

Atmospheric Science

First Advisor

Clark Evans

Committee Members

Jonathan Kahl, Sergey Kravtsov

Abstract

In recent years, the United States’ operational global numerical weather prediction model, the Global Forecast System (GFS), has been upgraded to include a new dynamical core and an updated turbulence parameterization. This updated turbulence parameterization uses a hybrid eddy-diffusivity, countergradient, and mass-flux formulation to approximate near-surface turbulent vertical mixing. The precise formulation used is based on the local stability, with the eddy-diffusivity, countergradient, and mass flux formulations used under stable, weakly unstable, and strongly unstable conditions, respectively. In this study, an objective classification of environmental regimes is used to verify GFS short-range vertical soundings, focusing on the planetary boundary layer where the turbulence parameterization plays an important role in determining vertical mixing, thus sounding characteristics. Observed temperature and dewpoint temperature from 15,488 soundings taken at 0000 UTC and 16,118 soundings taken at 1200 UTC between May – November 2019 are first interpolated into a height above-ground-level (AGL) coordinate and normalized to the pseudoadiabat defined by the surface-based parcel’s wet-bulb temperature. This allows for sounding shapes to be classified together regardless of their temperature and dewpoint temperature differences due to altitude, geography, or seasonality. A multivariate empirical orthogonal function (EOF) analysis is then performed on the normalized sounding data, after which a k-means clustering analysis is conducted on the leading two principal components retained from the EOF analysis. The output of this analysis classifies soundings into three different environmental regimes at each time, leading into a regime-specific model sounding verification. The classification method identifies distinct environments at each time, including deeply mixed layers (strongly unstable), shallow mixed layers (weakly/moderately unstable), and radiation inversions (stable), and each profile has varying biases due in part to turbulent mixing issues within the boundary layer.

Download

Share

COinS