Brian Jeffery Beggerly on

Urban areas are hot, and are getting hotter. The higher temperature in urban areas – the urban heat island (UHI) effect – is an emergent property of urbanization and a major research area in urban climatology.

Most UHI studies are conducted on a city-by-city basis with variable methodologies, data sources, and quality control criteria. Satellite-based estimates of the UHI, the surface UHI (SUHI), reduces some of these inconsistencies by using data collected by the same satellite-carried sensors. The SUHI is usually calculated based on the land surface temperature (LST) difference between an urban area (or unit) and its surroundings, called the “rural reference.” Irrigation patterns, agriculture, and the phenology of the background vegetation of the region all influence this rural reference.  Since vegetation cover changes modulate the LST, the rural reference has an influence on the SUHI intensity, especially at seasonal and annual scales.  Most previous multi-city studies on the SUHI have used the same buffer width to specify this rural reference for all cities, which is problematic due to the dependence of the SUHI footprint on city size.

Another major issue with studying the SUHI is with defining the urban unit. Administrative boundaries of urban areas are not always based on the urban (built-up) characteristics of the surface. If these units are used to estimate the SUHI, as they frequently are, we get a biased SUHI intensity.  For instance, if a city’s administrative boundaries include non-urban areas, the estimated SUHI would be lower due to the mix of urban and non-urban land use. The discrepancies in the choice of both urban and rural units make it hard to compare the SUHI intensities of different cities.

We attempted to resolve these methodological deficiencies in the UHI literature by designing a new algorithm, known as the simplified urban-extent (SUE) algorithm, to estimate the SUHI of urban clusters at a global scale.

For the first deficiency, we forgo the use of buffers and calculate the UHI as the LST difference between the urban and rural pixels, as defined by their spectral characteristics, within each urban unit. To address the second issue, we use urban clusters, which may include multiple contiguous cities, as the urban units instead of individual cities. We calculated the SUHI for over 9000 such urban clusters using roughly 17 years of satellite-derived LST data, and 12 years of satellite-derived land use data. Finally, we investigated the influence of vegetation on the seasonal and temporal characteristics of the SUHI intensity at a global scale and for each climate zone. The database created in this study is the most comprehensive characterization of the SUHI to date.

To make our data more accessible, we created an interactive web application to display the UHI intensity of all urban areas. The app allows the user to query the annual, seasonal, and temporal trends of the SUHI intensity for almost any urban area on Earth. One can download the seasonal and temporal plots of the UHI intensity, as well as the raw data, for the urban areas of interest. This app can potentially provide baseline data to allow urban residents, city planners, and policymakers take actionable steps to mitigate heat stress for their specific cities.

The image below shows a screenshot of the app with its main features. Clicking on any of the urban areas on the map generates the data and plots them on the data panel. More information about the app and the link to it can be found here.

More details of the SUE algorithm and the results of this study are in Chakraborty and Lee, 2019. Data from the SUE algorithm are also featured in the upcoming launch of the Urban Environment and Social Inclusion Index (UESI) this December. If you have any suggestions or comments about the study, the dataset, or the app, please contact me at

Featured photo by Brian Jeffery Beggerly on

Chakraborty, T. & Lee, X. (2019). A simplified urban-extent algorithm to characterize surface urban heat islands on a global scale and examine vegetation control on their spatiotemporal variability, International Journal of Applied Earth Observation and Geoinformation. 74, 269-280. doi: