Session summary
This GeoField 2026 case studies session presents applied examples from the Geospatial Impact Evaluation in Practice textbook. The session focuses on how geospatial methods can support impact evaluation when field data are limited, interventions are spatially targeted, or outcomes are difficult to measure through conventional surveys.
Moderator: Rachel Sayers
Geospatial Synthetic DID of an Agricultural Intensification Program
Sanchi Lokhande presents a geospatial synthetic difference-in-differences evaluation of an agricultural intensification and irrigation program in Niger. The study focuses on irrigation perimeters constructed downstream of existing dams to improve food security and climate resilience. Using Landsat-derived vegetation indicators, soil moisture data, weather controls, and weighted comparison pixels, the analysis estimates changes in vegetation after irrigation infrastructure was completed. The findings suggest substantial vegetation gains, while also showing that impacts vary across sites.
Assessing Forest Loss in Ghana Caused by Illegal Gold Mining
Foster Mensah presents a case study on the causal impact of illegal small-scale gold mining on deforestation in Ghana. The analysis combines mapped mining footprints, forest loss data, terrain, roads, rivers, and other spatial covariates in a matched difference-in-differences design. The results show that illegal mining caused additional forest loss above the counterfactual, with impacts concentrated near rivers, tributaries, forest reserve boundaries, and road-accessible areas.
Measuring Local Economic Impacts of Typhoons
Boi Yen Ha presents a textbook chapter on estimating the local economic impacts of typhoons in Vietnam. The study combines storm-track data with harmonized nighttime lights to measure changes in economic activity at the province-year level. Using difference-in-differences methods for staggered exposure, the analysis finds that typhoon exposure reduced nighttime light intensity, with effects persisting for several years in some specifications.
Together, the session shows how geospatial impact evaluation can be adapted to different policy domains and data constraints. Irrigation infrastructure, illegal mining, and typhoons each involve spatially defined exposure, uneven local impacts, and limited conventional data. The session demonstrates how remote sensing and causal inference can generate actionable evidence, while underscoring the importance of careful treatment definition, credible comparison groups, robustness checks, and transparent workflows.