Sealed urban surfaces absorb radiation throughout the day, heating up, and releasing it slowly overnight. Vegetation on the other hand absorbs radiation converting it into other forms of energy that will not cause a rise in temperature: latent heat through evapotranspiration or chemical energy through photosynthesis. Sealed surfaces cause urban areas to be significantly warmer than rural surroundings, creating the urban heat island. Urban vegetation cover depends on the function of the city sector. Typically a gradient in vegetation cover can be observed from almost zero in downtown areas through larger proportions in residential areas to almost complete cover in parks. Land cover in cities is highly fragmented: patches of vegetation mix with buildings, roads, parking lots, and other features.

A pilot study during a heat wave in July 2005 showed that pavement surface temperature differences between shaded and sun-exposed sites remain significant throughout the night. However, these were only point measurements. In this study aerial coverage across the city was achieved using satellite images. Coarse resolution data, such as from Landsat Thematic Mapper (TM), do not pick up fragmentation details, but provide a source of thermal data. Surface temperature was estimated from the thermal infrared channel for one day during the heat wave. The Normalized Difference Vegetation Index (NDVI) was used to analyze vegetation distribution at two resolutions (Ikonos: 4 m, TM: 30 m). The NDVI of the high resolution image was aggregated to the same pixel size as the temperature data and compared to the NDVI from the TM image. Temperature estimates were regressed against the NDVI.

The correlation between NDVI values from the aggregated high resolution and the TM images were strong, indicating that the coarser resolution image yields a good representation of the higher resolution product. Not surprisingly a highly significant inverse relationship exists between vegetation proportion and surface temperature. Even the coarse resolution allowed for the detection of hotspots that are caused by as little as a quarter of a city block of sealed surfaces (e.g. schools) in otherwise vegetated residential neighbourhoods. A single block of high vegetation cover in the middle of an otherwise highly sealed area, such as the Court House block in downtown Huntington or Marshall University campus, caused a clear depression in surface temperatures. The Ohio River with low NDVI but also low temperatures was an exception that can easily be explained by the fact that the river surface is not sealed and dry but provides an unlimited amount of water for evaporation.

Results of this study support a transition to a more park-like design of urban centres wherever possible. Particularly trees would be valuable in highly sealed areas. On the ground, where human activity occurs, trees' footprints are small. Instead they expand above the zone that is typically used by humans. There the canopy intercepts radiation, preventing it from heating the ground.