Modeling Weed Distributions Using Soil, Topography, and Disturbance History

This research project investigates how weed species distributions are structured by interactions among soil properties, landscape position, and disturbance history in an agricultural field of the southeastern Coastal Plain. High-resolution field observations of weed occurrence are integrated with surface (0–15 cm) soil fertility data, depth-based pedological information, and environmental covariates derived from LiDAR and satellite imagery.

The project evaluates species-specific responses to soil chemical properties and landscape processes, emphasizing the role of hydrology and subsurface soil characteristics in shaping spatial patterns of weed occurrence. Machine-learning models are used to predict both weed distributions and subsurface soil properties across the field, allowing vertical pedon information to inform surface-level predictions. By linking soil profile data, remote sensing, and land-use history, this work aims to advance field-scale understanding of weed ecology and supports more targeted, soil-based management strategies.