Improving the Characterization of Urban Form and Function with Spatial Metrics using Remote Sensing Data

To understand changes in urban form and function, and how these changes relate to urban development processes that drive these changes or that are affected by it, increasing use is made of computer-based urban growth models. The performance of these models strongly depends on the availability of time series of data on urban land use, needed for calibration of these models. Producing such data from available data sources is time-consuming. Land-use interpretation is also a rather subjective process, which may lead to inconsistencies in land-use maps derived for different time steps or for different areas. This hampers the use of such data in developing urban growth models, and in comparative studies on urban dynamics. These obstacles call for a formalisation of the land-use interpretation process and for the development of (semi-) automatic approaches for inferring urban form and function from available data sets.

Several studies have shown the potential of spatial metrics for identifying distinct types of urban form and function. Such metrics can be calculated for relative homogeneous areas that constitute the urban fabric, e.g. urban blocks. Urban blocks are typically composed of built-up and non built-up areas, with a specific composition and spatial configuration. The wide spread use of GIS-based tools makes that for most cities large-scale vector data sets are produced, containing detailed information on the structure of the built-up area. However, while these data sets usually include detailed representations of the outline of individual buildings, in most cases no information is included on the physical characteristics of the areas that surround the buildings, and together with these buildings characterize the block as a whole. This makes it difficult to distinguish different types of urban form and function based on this type of data.

High-resolution remote sensing data may complement large-scale building data sets by providing information on the structure of the non-built area, e.g. parking lots, vegetation patches, and thus enable a better characterisation of urban form at block level. This research focuses on the added value of high resolution remote sensing data, in addition to urban building data, for describing and mapping urban form and function, using spatial metrics. The approach proposed is applied on the Brussels Capital Region, using the UrbIS large-scale reference data base of the region as well as high-resolution Ikonos data covering the area. A typology of urban form and function, based on traditional and newly developed spatial metrics is presented.