Modelling and mapping the urban thermal environment

Abstract

In order to develop a methodology and tools that could help on thermal urban planning, this paper proposes a modelling of urban thermal environment based on the application of Artificial Neural Networks (ANN) and GIS tools. The study area was a residential neighbourhood in a medium sized city, where urban air temperatures at the pedestrian level could be collected. At the same time, rural temperatures that were registered and made available by the city meteorological station site, have been obtained. Urban features of these points were determined by the following characteristics: street's orientation, building volumes, proportion of green areas, proportion of built and non-built areas and sky view factors (SVF). Based on the differences of urban and rural air temperatures registered at 7 p.m., a model applying Artificial Neural Networks was developed. The results of that model allowed the simulation of values for the entire area and the creation of thermal maps in a GIS environment.