MTD: a new powerful method to select urban-rural pairs for Urban Heat Island quantification applied to Turin, Italy

Since Howard (1833) first suggested that air temperatures recorded in urban areas are higher than in the surrounding countryside, there have been hundreds of studies of the Urban Heat Island (UHI) phenomenon, which is due to the different thermal properties between urbanized and natural lands, anthropogenic heat emissions, human-induced pollution and limited wind blowing among buildings. The impervious land cover type and the presence of sheltering constructions trap heat during the day and release it during the night,  resulting in higher night-time temperatures. The UHI intensity is commonly computed as the difference between an urban and a rural measurement site and, therefore, the definition of station pairs is a crucial task for its evaluation. To this end, we propose a powerful method capable of highlighting the thermal pattern typical of each weather station: the Mean Temperature Difference (MTD) method. Principal Component Analysis (PCA) is adopted to cluster similar thermal behaviours, allowing an objective classification of the stations. The strength and novelty of this data-based approach, which employs hourly temperature measurements, lays in the fact that any preliminary assumption about the landscape characterizing each station (i.e. urban or rural) is not needed, making it less arbitrary and more objective than other methods. The application of the proposed method to the metropolitan area of Turin (Italy) shows that the joint use of MTD with PCA yields reliable and easily interpretable results, also in an area with complex morphology (orographic and hydrographic heterogeneity, different land uses, etc.). Once the best urban and rural pairs have been identified, the characteristics of the Urban Heat Island are shown, highlighting its seasonal and daily properties.