Dissertation
Advances in AI, cyber systems, and data-driven techniques for large-scale flood mapping
University of Iowa
Doctor of Philosophy (PhD), University of Iowa
Spring 2024
DOI: 10.25820/etd.007348
Abstract
Floods rank among the most devastating natural disasters, challenging humanity for over a millennium. Recent advancements in physical sciences have significantly enhanced our understanding of flood dynamics, leading to better preparedness and mitigation strategies. However, the rapidly evolving human society, coupled with changing climate and Earth surface conditions, has presented new challenges in flood modeling and mapping. Today, there is a growing demand for flood mapping models to be faster, ideally in real-time or near-real-time, to facilitate timely decision-making and adjustments. Additionally, there is an interest in discerning commonalities and differences in flood patterns across various global regions. While these objectives have been long-standing, it's only recently that they have become attainable, thanks to breakthroughs in cyber technologies, data-driven methodologies, and the significant improvement in data availability, particularly in terms of large-scale remote sensing datasets.This research concentrated on three promising areas to enhance flood mapping methods, aiming for speed, precision, and scalability. In the first phase, I explored the effectiveness of a DEM-based flood model, Height Above Nearest Drainage, and refined the algorithm for greater adaptability and enhanced performance in areas with intricate hydrologic and geospatial characteristics. In the second phase, I examined the influence of physical drivers like precipitation and other factors, including soil moisture and land cover, on Earth conditions, employing cutting-edge deep learning techniques. This led to the successful demonstration of predicting Earth Observation (EO) images at the pixel level by understanding the influence of these variables on Earth surface conditions. So far, most advancements are currently producer-centric, necessitating additional efforts to effectively communicate findings to end-users and decision-makers. Thus, the third and final phase of my research was dedicated to developing accessible, web-based, and cloud applications for extensive online data analysis and visualization in flood mapping. The goal was to offer customizable and user-friendly data and analytic products, catering to a diverse audience with varying needs and levels of expertise.
Details
- Title: Subtitle
- Advances in AI, cyber systems, and data-driven techniques for large-scale flood mapping
- Creators
- Zhouyayan Li
- Contributors
- Ibrahim Demir (Advisor)Nathan Young (Committee Member)Susan Meerdink (Committee Member)Witold Krajewski (Committee Member)
- Resource Type
- Dissertation
- Degree Awarded
- Doctor of Philosophy (PhD), University of Iowa
- Degree in
- Civil and Environmental Engineering
- Date degree season
- Spring 2024
- Publisher
- University of Iowa
- DOI
- 10.25820/etd.007348
- Number of pages
- xvii, 256 pages
- Copyright
- Copyright 2024 Zhouyayan Li
- Language
- English
- Date submitted
- 04/15/2024
- Description illustrations
- illustrations (some), color maps
- Description bibliographic
- Includes bibliographical references (page 196-252).
- Public Abstract (ETD)
The devastating effects of floods are manifold, encompassing human fatalities and injuries, disruptions to societal activities, and extensive damage to both private and public properties. The struggle against floods likely parallels the history of human civilization. Significant advancements in mathematics and physics over the past centuries have profoundly deepened our understanding of flood dynamics. Moreover, recent progress in data-driven and cyber-techniques, along with enhanced data accessibility, has opened new avenues for conducting comprehensive flood inundation mapping and studying flood processes and impacts from a broader perspective.
This research primarily focused on three key areas to further the field of large-scale flood inundation mapping. My work involved the examination of data-driven flood models, the prediction of large-scale Earth surface conditions using a blend of data-driven methodologies and remote sensing imagery, and the development of web- and cloud-based systems for flood and geospatial analytics and visualization. These systems are designed to provide customizable and personalized data and decision support tools, catering to a diverse array of users.
- Academic Unit
- Civil and Environmental Engineering
- Record Identifier
- 9984647647802771
Metrics
5 File views/ downloads
24 Record Views