Mapping and Monitoring Center Pivot Irrigation Data-Poor Settings

While the low 5% rate of irrigation in Sub-Saharan Africa leaves agriculture vulnerable to climate change induced disruptions in rainfall patterns, the recent uptake of medium-scale irrigation, particularly center pivot irrigation systems, may help buffer against such shocks. However, a lack of data on the installment and use or disuse of such systems precludes our understanding of this expansion, its potential and sustainability under climate change, as well as its impacts on food security and equitable economic development. Using PlanetLabs 5m basemap imagery, we employ the novel Segment Anything Model to segment all objects, and subsequently filter these objects to retain those depicting the circular shape characteristic of center pivot irrigation systems. This zero-shot approach effectively identifies and segments the vast majority of such ventures. To uncover the history of use and disuse of these systems, we retrieve Landsat time series information for these located center pivots and employ a gradient boosting model to classify activity status over time. This approach provides detailed insights into irrigation system usage patterns, vital for understanding the role of irrigation as a climate change adaptation tool.

Faculty Mentor: Kelly Caylor

Project Mentors: Yang Hu, Anna Boser