Evaluating Satellite Land Cover Accuracy in a Suburban Environment Using Citizen Science: New Hyde Park, NY

Abstract

Accurate land cover classification is essential for environmental monitoring, urban planning, and climate research; however, suburban landscapes remain difficult to characterize due to heterogeneous land cover. This case study evaluates the agreement between satellite-derived land cover tools and ground-based citizen science observations in New Hyde Park, New York. Using the Adopt-a-Pixel 3 km methodology, thirty-seven primary sampling units were established within a standardized area of interest. Field observations were collected using National Aeronautics and Space Administration (NASA)’s Global Learning and Observations to Benefit the Environment (GLOBE) Observer application and supplemented with high-resolution reference classifications generated through Collect Earth Online. These datasets were compared with satellite observation, specifically from European Space Agency (ESA) WorldCover, Dynamic World, ESRI Land Cover, Landsat time-series, and Meta/WRI Global Canopy Height datasets. Results indicate frequent over-generalization of developed land cover and underrepresentation of tree canopy, with the strongest agreement observed where surface water is present. Qualitative field documentation and community accounts revealed that storm damage, aging trees, and housing management practices contributed to long-term greenery loss, helping to explain some discrepancies between ground and satellite observations. Findings demonstrate that integrating citizen science, community knowledge, and reference data will continue to improve land cover assessment in suburban regions and support more inclusive and reliable environmental monitoring.