Urban Revival: Mapping urbanization impacts on surface temperature and vegetation using remote sensing and machine learning.

Urban development is altering land cover globally, impacting ecosystems, biodiversity, and climate. Our study investigates how satellite imagery and remote sensing can identify patterns of urban development and analyze related environmental changes. We used initial field observations from the GLOBE Observer App, where citizen scientists collect high-resolution, ground-level images using smartphone cameras. Each observation included geotagged photographs taken in standardized directions, and ultimately, we compiled a dataset of over 1000 photos. This dataset revealed that rapid urbanization and recent land cover changes were not accounted for in classification products such as ESA WorldCover (2020/2021) and Dynamic World (2015-2024). This led to our primary research question: How effectively can remote sensing detect urban expansion and its impacts on surrounding communities? Using urban cities like Chicago, Atlanta, Los Angeles, Detroit, and Austin, we studied impacts such as green space gains/losses, heat island formations, and nighttime brightness changes. To do this, we utilized Landsat—derived NDVI (2001-2020) and Sentinel-2 imagery (2015-2025), ECOSTRESS (2018-2025), and VIIRS data (2012-2025). Integrating multiple remote sensing datasets, we explored different machine learning models that classified geographic zones into three categories of development: regenerative, unequal, and stable. This classification differentiated areas that experienced positive environmental and urban recovery, harmful patterns including spectral patterns such as decreasing vegetation reflectance and increasing thermal emissions, and minimal spectral change over time. Our findings show remote sensing can reveal patterns of urban growth, such as in southwest Chicagoland, where impervious surfaces increased 50% relative to 2001 levels in 2021. Moving forward, the zone classifications could be a valuable resource for urban planners and local communities. By identifying areas at risk of environmental degradation, intervention could be more targeted and effective. This framework could also be extended to other cities to support global comparisons of urban development.