Title: Radar Earth Observation Imagery for Urban Area Characterisation
Authors: MOLCH Katrin
Publisher: OPOCE
Publication Year: 2009
JRC N°: JRC50451
ISBN: 978-92-79-11731-2
ISSN: 1018-5593
Other Identifiers: EUR 23780 EN
URI: http://publications.jrc.ec.europa.eu/repository/handle/JRC50451
DOI: 10.2788/8453
Type: EUR - Scientific and Technical Research Reports
Abstract: This report introduces the use of medium to high resolution spaceborne radar (SAR - Synthetic Aperture Radar) Earth observation imagery for urban area mapping applications. Urban mapping can benefit from this type of satellite data since built-up structures induce strong backscatter and thus can be distinguished well on radar imagery. The objective is to raise awareness about the possibilities and some of the limitations associated with using SAR satellite imagery for characterizing the built-up area. The delineation and thus knowledge on the distribution of urban agglomerations at a global or continental scale is of importance to a wide range of applications such as determining hot spots in the framework of disaster preparedness, or modeling the impact of a disease outbreak. Population, however, is not distributed evenly throughout an urban area. For applications supporting e.g. humanitarian aid initiatives, the outline of the urban or built-up are, thus, might not be sufficient. These applications require reliable information on where the people are. At smaller scales, the density of built-up structures can serve as a first, coarse estimate of the population distribution within a city. Population density varies between different neighborhoods - and with time of day. Built-up density maps thus provide added value to binary built-up area delineations. Moreover, this density distribution changes over the years. It can be monitored by multi-temporal built-up density maps. A critical parameter to measure at a regional scale is the built-up stock. Measuring built-up stock takes into account the type and distribution of buildings. This can aid in estimating the population of a given area, specifically in regions where administrative data of this type are not readily available. Population information is essential for assessing the number people potentially affected should a crisis, natural or man-made, occur, and will help determine the type and amount of aid required. Most globally available land cover datasets, such as Global Land Cover (GLC) 2000, or Africover, merely provide urban area outlines. As a global population dataset, the Landscan data, updated every few years and available through the Oak Ridge National Laboratory, USA, depict population density in 1 km raster cells. This report addresses possibilities to use SAR data to improve these existing globally available datasets either with respect to spatial resolution or thematic information. As a result of their reliability, weather independence, and relatively low cost, satellite SAR imagery at approximately 23 m spatial resolution (ERS-1 /-2, ENVISAT) constitute an attractive alternative to optical imagery for mapping purposes at scales of up to 1:100,000. Higher resolution satellite SAR data (RADARSAT-1/-2, TerraSAR-X) are useful for inner-city differentiation. SAR data are different from optical data with respect to the surface parameters they measure and in the way the information is coded in the image. Moreover, the side-looking geometry introduces geometric effects. The sensor-specific image characteristics have to be taken into account during SAR data processing and information extraction. After a brief introduction, the representation of urban areas on SAR images is illustrated. Specific issues and limitations are discussed. In the final chapter methodologies towards automating built-up area delineation and characterization from SAR data are introduced.
JRC Directorate:Space, Security and Migration

Files in This Item:
File Description SizeFormat 
st_report_sar_urban_final.pdf4.93 MBAdobe PDFView/Open

Items in repository are protected by copyright, with all rights reserved, unless otherwise indicated.