Continental-scale urban–reference patterns in greenness and land surface temperature across U.S. cities

Context Urban and nearby non-urban landscapes differ in vegetation and surface climate. How these differences vary with hydroclimate over multiple decades remains poorly resolved. Objectives We compared urban and reference greenness and land surface temperature across U.S. climates. We also evaluated differences in trajectories, spatial variability, and precipitation dependence. Methods We analyzed harmonized Landsat observations for 52 U.S. city–reference pairs over 1995–2024. Comparisons used fixed footprints, environmental covariates, and supplementary sensitivity analyses. Results Across 52 pairs, the median city was less green (ΔNDVI = − 0.17) and warmer (ΔLST = + 5.09 °C) than its reference. These differences reversed along the climatic gradient, with arid cities often greener and cooler and mesic cities generally less green and warmer than their references. In paired annual composites, cities warmed approximately 22% faster than references on average, while urban NDVI changed little or declined slightly and references greened. Greenness was more spatially variable in every urban footprint by CV and SD, and this difference persisted in 36 qualifying equal-area comparisons. Precipitation responses were strongest in dry settings and generally weaker in cities. Conclusions Urban–reference differences varied predictably with regional hydroclimate, while cities and references followed diverging long-term trajectories. Greater urban greenness heterogeneity persisted after accounting for sampling area and shape. These observations support an inter-urban water-deficit framework that considers regional hydroclimate and urban land cover when comparing landscapes and their trajectories.