Date of Award

5-1-2026

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Coastal and Marine Systems Science

College

College of Science

First Advisor

Erin Hackett

Second Advisor

Louis Keiner

Third Advisor

Richard Viso

Additional Advisors

David Flagg; Thomas Hanley

Abstract

Atmospheric remote sensors have always offered promising technological advancements for measuring atmospheric properties over large spatial areas at high spatiotemporal resolution – a feat not currently possible with present atmospheric measurement technologies. However, remote sensors do not measure these atmospheric properties directly; they rely on robust calibrations or conversions, complex mathematical inversion methods, and/or machine learning to retrieve these properties. These inverse methods rely on the selection of an objective function, a machine learning technique, and an accurate parameterization for atmospheric property estimation. By leveraging inverse methods to improve the characterization of properties measured by remote sensors, this work enables high-resolution monitoring of the marine atmospheric surface layer (lowest 100 m of the atmosphere over the ocean) over extensive spatial domains.

This research advances remote sensing inversion methodologies that utilize horizontally pointed X-band radar operating within the marine atmospheric surface layer to detect air properties such as temperature, humidity, and refractivity. Advancements of this science and technology occur along three fronts: (i) by fusing a remote sensing inversion framework with numerical weather prediction ensemble data to benchmark its accuracy against traditional methodologies; (ii) by expanding and evaluating refractivity inversion methods that retrieve estimates of thermodynamic properties, such as temperature and humidity; and (iii) by adapting the inverse modeling framework to account for heterogeneous atmospheric conditions, evaluating how this heterogeneity impacts inversions, and determining when such conditions are most likely to occur based on numerical weather prediction datasets.

It is found that regularizing inverse methods utilizing numerical weather prediction ensembles can offer some distinct advantages in the accuracy and speed of the inverse method, but results also highlight a need for balance between regularization and ensemble diversity. In terms of extending these inverse methods to retrieve temperature and humidity, results indicate that humidity can be accurately estimated, whereas temperature remains more challenging to characterize. Finally, it is found that accounting for lateral heterogeneity in inverse methods can lead to more accurately estimated atmospheric properties as long as the environment is truly heterogeneous; otherwise, it may actually degrade accuracy due to overfitting. It was also found that not accounting for heterogeneity, when present, can lead to propagation prediction errors well above 5 dB in transitional trapping environments. Together, these advancements in data fusion, property retrieval, and heterogeneity modeling significantly broaden the utility of ground-based remote sensing for monitoring the marine atmospheric surface layer.

Download

Share

COinS