Cooling Machines: Exploring the Heat Mitigation Effect of Urban Trees with Computer Vision

Abstract

As the impacts of climate change on cities become more pronounced, urban authorities are under pressure to prepare existing streetscapes for increased levels of heat stress. While many aspects of existing urban morphology have an impact on heat exposure (e.g. sky view factor, glazing levels, facade materials), they cannot be rapidly changed at large across existing urban infrastructures. Urban authorities across the world increasingly turn to planting trees as a way of cooling urban streetscapes. Urban vegetation is indeed known to have a cooling effect, primarily due to trees providing shade and preventing urban materials from heating up, as well as due to their ability to maintain their own internal temperature due to evapotranspiration. Even though the positive impacts of urban trees on thermal comfort are long known and well-studied, little work is dedicated to how these impacts vary across trees of different species and morphology. This is due to both the complexity of studying vegetation life cycles at sufficient scale, as well as due to the dispersed nature of the issue across disciplines of biology, urban climate, design, and data science. Nevertheless, this specific knowledge is vital to urban planners for deciding which trees have the most cooling effect in specific parts of the city. This thesis embraces the notion of trees as ‘cooling machines’ and dissects the diverse morphological and contextual factors that affect the role of individual trees on local urban heatscape. Leveraging a set of computer vision methodologies, including species recognition, context-aware segmentation, and photogrammetry, the thesis examines a large dataset of thermal imagery of urban trees collected in Los Angeles and Dubai to describe the impact of individual tree species, height and form, as well as spatial context on the cooling effect. Building on this approach, the thesis proposes a prototyping framework for architects to cure urban heatscapes via targeted curation of tree planting schemes, tying the visual and thermal aspects of urban greenery. This approach will allow cities to leverage the power of urban vegetation in the most efficient way, and tame urban heat in a scalable and globally affordable manner.