To estimate how thermal stress varies with infrastructure changes.

WRI

Sentinel2-derived FabDEM (30m)​, Overture building footprints (Google, Microsoft, OSM)​, UT-GLOBUS building height (ICESat-2 and GEDI), WRI/Meta tree canopy height (1m)​, WRI OpenUrban land use (1m)​, ERA5 meteorology (~9km).

1 meter

Areas of interest identified by cities

hour

We are providing data corresponding to one historical day of reference based on highest registered air temperature during the previous year.

A limitation of this open-source modeling approach is the error created by input datasets. UTCI data is heavily influenced by the placement, extent, and height of trees and buildings. A reliance on open-source, global datasets with varied error produces a range of misplaced, missing, or erroneous shadows that, accordingly, affect UTCI. While presented at a 1m resolution, inputs to the UTCI layer (shown above in ‘Data Sources’) include datasets at coarser resolutions (e.g., 30m DEM) and datasets with spatial errors known to be well above 1m (e.g., building heights and their shadows). Thus, at a very high resolution, an expert local viewer will see features that they know are wrong or missing in the maps. We do not recommend that the data be used to guide specific interventions. The UTCI values across a neighborhood in sun, tree shade, and building shade, as well as on varied urban surfaces, are realistic and can be compared and used to understand the dynamics of hyperlocal urban heat. These values, and the metrics calculated across a neighborhood or city, can be used to support and guide policy.