ArcUHI: a GIS add-in for automated modelling of the Urban Heat Island effect through machine learning

Abstract

Increased urbanisation is boosting the intensity and frequency of the Urban Heat Island (UHI) effect in highly developed cities. The advances in satellite measurement are facilitating the analysis of this phenomenon using Land Surface Temperature (LST) as an indicator of the Surface UHI (SUHI). Due to the spatial implications of using remote sensing data, this research developed ArcUHI, a Geographic Information System (GIS) add-in for modelling SUHI. The tool was designed in ArcGIS, which was bound with R to run machine learning algorithms in the background. ArcUHI was tested using the metropolitan area of Madrid (Spain) in 2006, 2012 and 2018 as a case study. The add-in was found to predict observed LST with high accuracy, especially when using Random Forest Regression (RFR). Building height and albedo were identified as the main drivers of SUHI, whose magnitude and extension increased with time. In view of these results, strategic roof and wall greening was suggested as a measure to mitigate the street canyon effect entailed by buildings and offset the heat retention capacity of built-up surfaces.