Effects of urban vegetation on the surface urban heat island: a UAV approach with a focus on private gardens

Context In the current context of climate change, the effects of the Surface Urban Heat Island (SUHI) are of increasing concern worldwide. Urban Green and Blue Infrastructure (UGBI) is widely recognized as an effective strategy to mitigate SUHI. However, not all UGBI elements are considered equally in adaptation plans, and evidence on the SUHI-mitigating capacity of different UGBI elements is still missing. Objectives We aimed to utilize Unoccupied Aerial Vehicles (UAVs) to assess the effectiveness of various UGBI elements in mitigating SUHI during a heatwave in Tartu, Estonia. We further addressed private gardens, and assessed how their composition and structure influence SUHI. Methods We used high-resolution multispectral UAV imagery and a Random Forest algorithm to classify UGBI elements alongside other urban land-cover types to enable comparative analysis. UAV-derived Land Surface Temperature (LST) was used to quantify their SUHI mitigation capacity. We combined LST with Generalized Additive Models (GAMs) to examine the effects of private garden composition and structure on SUHI. Results UGBI maps achieved high classification accuracy (OA = 0.95). LST differed significantly between land-cover classes. Mown grass showed high temperatures (40.4 ± 5.04 °C), comparable to impervious surfaces, whereas unmown grass and woody vegetation exhibited substantially lower temperatures. GAMs further showed that neighbourhood context is associated with cooling patterns, with deviance explained increasing from 47.2 to 73.0% when neighbourhood-specific relationships were included. Conclusions UAV imagery enabled the identification of fine-scale variation in UGBI and SUHI, highlighting the importance of vegetation type, management practices, and neighbourhood context in relation to urban cooling effects.