ClearScan - A Machine Learning System for Customized, Site-Specific Radar Image Filters

Aug 3, 2026

Upload by Miles

Radar meteorology relies on removing anomalous propagation (AP) echoes—non-precipitation returns—from radar images, yet existing heuristic and machine-learning approaches often target familiar architectures and may underperform under differing radar hardware or RF interference. ClearScan presents a flexible machine-learning system with an interactive training interface to adapt filtering to the specific conditions of each radar site. The thesis emphasizes transferability across environments through customizable model building and evaluation.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages52

Reading Time2 h 11 min

AccessFree to download

Abstract

Acknowledgements

Table of Contents

List of Figures

Chapter 1. Introduction

Background Overview

Motivation

Methodology

Thesis Organization

Chapter 2. Background and Related Work

Chapter 3. ClearScan System Operation

Sample Import

Labeling

Network Builder

Test Builder

Chapter 4. Implementation Details

Chapter 5. Case Study: Bangladesh Radar

Introduction

Dual Networks

Results

Alternative Network Comparison

Quantitative Results

Potential Drawback

Chapter 6. Conclusion

Contribution

Future Work

References

List of Figures

University of Alabama in Huntsville LOUIS | Theses | UAH Electronic Theses and Dissertations | | --- | --- | | 2024
ClearScan : a machine learning system for customized, site-specific radar image filters
Erick Jones
Follow this and additional works at: [https://louis.uah.edu/uah-theses](https://louis.uah.edu/uah-theses) | | Recommended Citation Jones, Erick, "ClearScan : a machine learning system for customized, site-specific radar image filters"(2024) . Theses. 683. [https://louis.uah.edu/uah-theses/683](https://louis.uah.edu/uah-theses/683) This Thesis is brought to you for free and open access by the UAH Electronic Theses and Dissertations at LOUIS. It has been accepted for inclusion in Theses by an authorized administrator of LOUIS. CLEARSCAN: A MACHINE LEARNING SYSTEM FOR CUSTOMIZED, SITE-SPECIFIC RADAR IMAGE FILTERS Erick Jones A THESIS Submitted in partial fulfillment of the requirements for the degree of Master of Science in The Department of Computer Science to The Graduate School of The University of Alabama in Huntsville August 2024 Approved by: Dr. Huaming Zhang, Research Advisor/Committee Chair Dr. Jacob Hauenstein, Committee Member Dr. Deepak Acharya, Committee Member Dr. Letha Etzkorn, Department Chair Dr. Rainer Steinwandt, College Dean Dr. Jon Hakkila, Graduate Dean Abstract CLEARSCAN: A MACHINE LEARNING SYSTEM FOR CUSTOMIZED, SITE-SPECIFIC RADAR IMAGE FILTERS Erick Jones A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science Computer Science The University of Alabama in Huntsville August 2024 In the field of radar meteorology, a perpetual problem is removal of so-called anomalous propagation (AP), i.e. , non-precipitation echoes, from the produced images. Much work has been done in this area already, including conventional heuristic algorithms as well as machine-learning systems such as neural networks. Often the focus is on certain familiar radar architectures such as WSR-88D, also known as NEXRAD. However, a large number of radars exist which are not identical to NEXRAD; and there are also environmental differences such as RF interference which can affect the success rate of existing AP removal strategies. The focus of this paper is to present a flexible machine-learning system for this task which provides a convenient training interface so it can be adapted to the specific conditions present at any given radar site to create a customized filter. We have named this system ClearScan. iii Acknowledgements I would like to acknowledge Baron Weather for providing facilities, equipment, funding, and other support including encouraging other employees to contribute and participate. I would also particularly like to thank Sherman, my department lead, who encouraged me to pursue this project and has often been at least as enthusiastic about it as I have. And my co-worker Emily, who has done the most hands-on work of using this software and providing valuable feedback. Finally, I would like to thank my advisor, Dr. Huaming Zhang, for cluing mein on how to do a thesis. Table of Contents Abstract .................................... ii Acknowledgements .............................. iv Table [of Contents](of Contents .............................. vi)[ ..............................](of Contents .............................. vi)[ vi](of Contents .............................. vi) [List of Figures](List of Figures ...............................)[ ...............................](List of Figures ...............................). vii Chapter 1 . Introduction ........................ 1 1.1 Background Overview ...................... 2 1.2 Motivation ............................. 3 1.3 Methodology ........................... 3 1.4 Thesis Organization ........................ 6 Chapter 2 . Background and Related Work ............. 7 Chapter 3 . ClearScan System Operation ..............

FAQs

What problem does ClearScan address in radar meteorology?

ClearScan targets removal of anomalous propagation (AP) echoes, which are non-precipitation returns, from produced radar images.

Why do existing AP removal strategies sometimes fail?

They may focus on specific radar architectures (e.g., WSR-88D/NEXRAD) and can be affected by differences in radar types and environmental factors such as RF interference.

How does ClearScan enable customization for different radar sites?

It provides a flexible machine-learning system with a convenient training interface, allowing a customized filter to be created based on the conditions at each radar site.

Aug 3, 2026

Upload by Miles