Identification of hot/cold spots inside the Surface Urban Heat Island of the main cities in North-Eastern Romania using Landsat imagery

Urban Heat Island (UHI) is caused by inadvertent climate modification due to human spatial concentration, being generally defined as the difference in temperature between urban areas and their rural surroundings. However, seen at fine spatial resolution, the heat island seems rather to look like an archipelago of hot spots delimited by colder areas. This study analyses the Surface Urban Heat Islands (UHI_Surf) of 16 cities located in Romania’s North-East Development Region for the identification of these hot and cold spots.

For each city, in order to identify UHI_Surf’s hot and cold-spots, Landsat series of satellites were used due to their potential to provide Land Surface Temperature (LST) product at a high spatial resolution, which is commonly required for micro or local scale studies. For this purpose, LST is derived from the Landsat 4, 5, 7, and 8 (1988-2021), collection 1 (using the Statistical Mono-Window algorithm), implemented in Google Earth Engine platform.

For hot /cold spots identification, Hot Spot Analysis (Getis-Ord Gi*) tool from ArcGIS Pro 3.0 was used. This tool calculates the Getis-Ord Gi* statistic for each feature in a dataset by looking at each feature within the context of neighboring features. To be classified as a statistically significant hot spot, a feature will have a high value and has to be surrounded by other features with high values. The local sum for a feature and its neighbors is compared proportionally to the sum of all features and a statistically significant z-score results. The larger the z-score is (positive), the more intense the clustering of high values defined as hot spots. The smaller the z-score is (negative), the more intense the clustering of cold spots. When the False Discovery Rate (FDR) correction is applied, statistical significance is adjusted to account for multiple testing and spatial dependency.

This procedure was applied for each of 16^th cities, summing up 10900 images which cover an area of 2526,8 km². LST is strongly controlled by surface properties (radiative, thermal, geometric, moisture and aerodynamic), these giving a greater surface temperature variability compared to air temperature, particularly during the day. Inside the identified hot spots, the LST is with 8-10C higher than the mean of UHI_Surf LST. Generally, light industrial, warehouses and transportation infrastructure (airports) are often relatively hot-spots, while cold-spots, are obviously more heavily vegetated areas, water bodies and areas of well-watered vegetation, but their futures are related to each city characteristics. The obtained results are designated to be used as the main assessment of urban heat island, delivering for stakeholder a clear image of the target regions inside the cities for the policies dedicated to the mitigation of the urban heat island effect.