Assessing Climatic Hazards in Coastal Socio-Ecological Systems using Complex System Approaches

Coastal socio-ecological systems face unprecedented challenges due to climate change, with impacts encompassing long-term, chronic changes and short-term extreme events. These events will impact society in many ways and prompt human responses that are extremely challenging to predict. This dissertation employs complex systems methods of agent-based modeling and machine learning to simulate the interactions between climatic stressors such as increased flooding and extreme weather and socio-economic aspects of coastal human systems. Escalating sea-level rise and intensified flooding has the potential to prompt relocation from flood-prone coastal areas. This can reduce flood exposure but also disconnect people from their homes and communities, sever longstanding social ties, and lower the tax base leading to difficulties in providing government services. Chapter 2 demonstrates a stochastic agent-based model to simulate human relocation influenced by flooding events, particularly focusing on the responses of rural and urban communities in coastal Virginia and Maryland. The findings indicate that a stochastic, bottom-up social system simulator is able to replicate top-down population projections and provide a baseline for assessing the impact of increasingly intense flooding. Chapter 3 leverages this model to assess how incorporating heterogeneity in relocation decisions across socio-economic groups impacts flood-induced relocation patterns. The results demonstrate how this heterogeneity leads to a decrease in low-income households, yet a rise in the proportion of elderly individuals in flood-prone regions by the end of the simulation period. Flood-prone areas also exhibit distinct income clusters at the end of simulation time horizon compared to simulations with a homogenous relocation likelihood. Lastly, Chapter 4 explores relationships between extreme weather and agricultural losses in the Delmarva Peninsula. Existing research on climatic impacts to agriculture largely focuses on changes to major crop yields, providing limited insights into impacts on diverse regional agricultural systems where human management and adaptation play a large role. By comparing various multistep modeling configurations and machine learning techniques, this work demonstrates that machine learning methods can accurately simulate and predict agricultural losses across the complex agricultural landscape that exists on the Delmarva peninsula. The multistep configurations developed in this work are able to address data imbalance and improve models' capacity to classify and estimate damage occurrence, which depends on multiple geographical, seasonal, and climatic factors. Collectively, this work demonstrates the potential for advanced modeling techniques to accurately replicate and simulate the impacts of climate on complex socio-ecological systems, providing insights that can ultimately support coastal adaptation.

Name thesis:degree_name

Doctor of Philosophy

Level thesis:degree_level

doctoral

Discipline thesis:degree_discipline

Biological Systems Engineering

Department dc:contributor.department

Biological Systems Engineering

Grantor dc:publisher

Virginia Tech

Year dc:date.issued

2024

Author dc:creator

• Nourali, Zahra

Chair dc:contributor.committeechair

• Shortridge, Julie Elizabeth

Committee members dc:contributor.committeemember

• Bukvic, Anamaria
• Little, John C.
• Easton, Zachary

• Socio-Ecological Systems
• Climate Change Impacts
• Sea-Level Rise
• Coastal Flooding
• Human Relocation
• Extreme Weather
• Agricultural Losses
• Complex Systems Approaches
• Agent-based Modeling
• Machine Learning

Statement dc:rights

• Creative Commons Attribution-ShareAlike 4.0 International

Licence dc:rights.uri

• http://creativecommons.org/licenses/by-sa/4.0/

Language dc:language.iso

en

Dc Identifier Other

vt_gsexam:40398

OAI identifier oai:identifier

oai:vtechworks.lib.vt.edu:10919/119212