Evaluating the effect of vegetation in urban microclimate using remote sensing technologies

Abstract

At present time, more than half of the world’s population lives in cities, and many regions of the world are experiencing rapid urbanization. The increasing urbanization combined with the effects of climate change have resulted in the intensification of urban heat islands. Urban heat islands decrease outdoor thermal comfort and create feedback loops on energy consumption. Furthermore, the more frequent temperature extremes directly impact human health, particularly in the hot regions of the world. Microclimatic models are often used to improve the understanding of the influence of the complex interactions between urban surfaces and their impact on thermal comfort. Thus contributing to guide the development of thermally comfortable cities. Increasing urban green cover is often proposed as a mitigation strategy to decrease the urban heat island effect and enhance outdoor thermal comfort. However, urban microclimatic models have not fully addressed the quantification of vegetation effects on urban climate. The two key terms of this research are the radiative budget and the mean radiant temperature (Tmrt). The radiative budget summarizes the interaction between radiation and the urban surfaces. The radiative budget of a city is directly influenced by the complexity of the urban geometry, surface materials, direct solar radiation, incidence angle, and atmospheric diffuse radiation. It is a key parameter in the urban energy balance and urban climate models. In contrast, the Tmrt summarizes the effective radiative flux reaching a human body and is an essential human bio-meteorological parameter of thermal comfort. Several modelling tools exist with varying levels of detail. However, limitations exist regarding data retrieval and model assimilation, particularly when addressing complex 3-D geometries and fine-scale variation of vegetation properties and the optical properties of urban surfaces. Remote sensing technologies possess great potential in improving the assessment and monitoring of the thermal behavior of cities. On the one hand, they enable the direct mapping of urban structures and their optical characteristics. On the other hand, remote sensing enables the retrieval and mapping of vegetation cover and their properties. These products constitute useful information that more advanced modelling tools could assimilate. This work seeks to close the gaps mentioned above by adapting an advanced 3-D radiative transfer model to compute the radiative budget of any urban scenario and produce maps of mean radiant temperature distribution. This adaption is capable of providing accurate simulations of urban temperatures over existing conditions or design scenarios at different scales. The first objective of this work was to develop a methodology to generate 3-D urban scenes from satellite data on which simulation tasks can be performed over any desired location in a city. Then, radiative budget simulations were carried out over such datasets by repurposing the Discrete Anisotropic Radiative Transfer (DART) model. This allowed to evaluate the impact of urban typology, surface material, and the effects of changing leaf area density (LAD) on the radiative budget of different urban scenarios in Singapore. The results show that highly urbanized landscapes with no tree cover can absorb radiation up to four times more than a densely vegetated “natural” landscape. Furthermore, the simulation results show good agreement when compared against net radiometer data obtained from a local flux tower in Singapore. The second objective was to develop a new method for detailed modelling and mapping of Tmrt by adapting the DART model. The adaption of the DART model enabled the precise quantification of the effects of vegetation and surface materials on the Tmrt. The method was successfully evaluated against Tmrt observations obtained from two sets of sensors mounted on a custom-made mobile platform consisting of three net radiometers, a globe thermometer, and a Vaisala weather station. Last but not least, the third part of this research work consisted of developing a fully convolutional neural network (FCN) approach to automatically delineate individual tree crowns (ITC) over a tropical urban park using satellite and UAV hyperspectral imagery. The ITC is an essential product at the beginning of the pipeline of fine-scale vegetation assessments. This work explored the role of spatial resolution and spectral information for individual tree crown delineation tasks. This research work highlights the importance of radiative transfer modelling as a tool for better understanding the urban heat island effect, optimizing urban design, and enhancing thermal comfort in urban areas. This knowledge is indispensable for creating sustainable, liveable, and comfortable urban environments, especially in the face of rising temperatures and the challenges posed by climate change.