Project 27:
This project will employ multi-band spectral imaging from NASA satellites, machine learning algorithms, and advanced economic modelling to predict crop failures. By integrating Earth science data with predictive analytics, we aim to provide early warnings of agricultural disruptions, enhancing community resilience to natural hazards. This approach promises to transform remote sensing data into actionable economic insights, strengthening disaster risk reduction and management efforts.
The student will identify satellite datasets sensitive to indicators of crop stress and land use change, then compile a time-series dataset for regions cultivating diverse crops, annotated with historical crop success and failure outcomes and other descriptors of the growing season. They will subsequently train or fine-tune a machine learning model to recognize patterns predictive of crop failure and land use change. Finally, the student will analyze the model’s predictions in conjunction with our lab’s economic models to quantify the potential economic impact of varying scales and severities of crop failure, providing a direct link between satellite observations and economic resilience strategies.
Upon completion of the project, the student will present their findings at the NAU Undergraduate Symposium and other conferences. The work will also be prepared for submission to a peer-reviewed journal, contributing novel insights into the predictive capabilities of satellite imagery for agricultural resilience. Additionally, the results will support our lab’s application for a ROSES grant, demonstrating practical applications of Earth science data in economic and convergence science contexts. The student’s research may also enrich ongoing NSF-, DOD-, FEMA- and EPA-funded work within the lab; for example, supporting collaborations with farmers practicing regenerative agriculture, integrating predictive models into user interfaces built for emergency managers, and supporting greenhouse gas inventory work.