Defense Presentation Title: Advancing Climate-Resilient Agriculture and Disaster Management: Applications of Remote Sensing and Machine Learning in Data-Scarce Bangladesh

Advisor: Dr. Mirela Tulbure, faculty fellow and professor in the Department of Forestry and Environmental Resources

Abstract: Timely, accurate, and high-resolution flood maps are essential for estimating the impact of flooding on agriculture and infrastructure. In Bangladesh, where rice is critical to food security and economic stability, flooding has threatened rice production and livelihoods and caused economic loss. Accurate and reliable information on flooding and rice production in Bangladesh could help understand the nexus between flooding and agriculture and could be useful for mitigating those challenges.

While limited studies have explored the nexus between floods and agriculture in Bangladesh, there is often a lack of spatial consistency, systematic validation, and proper uncertainty quantification. Additionally, the effectiveness and suitability of data-driven models in tackling these challenges in data-scarce regions have yet to be fully explored.

This dissertation leverages Earth observation data and machine learning models to (1) develop an automated framework for in-season rice area mapping, (2) quantify and predict rice yields and trends, and (3) quantify flooding and its impact on rice. This dissertation supports climate-resilient agriculture and disaster management, providing critical understanding for policymakers and stakeholders in Bangladesh.