Investigation of the Integration of AVlRlS and IFSAR for Urban Analysis

Most attempts at urban analysis using remotely sensed imagery lack the capabilities necessary to define the detailed geometry and differentiate the textures of the complex urban landscape. This paper presents a proof-of-concept study of the potential for integrative analysis of Interferometric Synthetic Aperture Radar (IFSAR) and Airborne Visible/Infrared Imaging Spectrometer (A VIRIS) hyperspectral imagery for a study area in Los Angeles, California. Recent advances in the use of interferometric radar allow the definition of high resolution, three-dimensional ( 3 ~ ) geometry of surface features, topography, and impervious surfaces in urban areas. The radar analysis i s enhanced using hyperspectral imagery to mask surfaces adjacent to structures in order to assist in the determination of baseline topography and segmentation of building footprints for improved geometric measurement of the complex urban area. Introduction The global tendency has been to locate cities in areas of sensitive environments, such as coastal zones, floodplains, and prime agricultural lands. Thus, the changing structure, size, and activity composition of cities has profound and disproportionate impacts on regional and global environmental change. With most of the Earth's population living in cities, there exists a great need for a means of rapidly and accurately inventorying and monitoring urban areas. This is especially true in developing nations, where the urban information infrastructure necessary for urban planning and management is not in place (IGBP, 1995). The urban landscape is extraordinarily complex. It is the manifestation of both physical and human processes expressed in intricate structural geometry of roads, buildings, and land-cover mixtures in relatively unpredictable spatial patterns. Current theories on urban processes, which generate these geometric and land-use patterns evident in the urban landscape, are inadequate to explain, let alone predict, urban patterns of land use and land cover (Thrall, 1987). Lacking a universal basis in theory to explain and predict the changing urban areas around the world, better methodologies and empirical tools must be developed to specify and analyze urban structure and processes. Improved methodologies and better analysis of these urban areas should provide the insights necessary for urban planning and management policies in the future. G.F. Hepner is with the Department of Geography OSH270, University of Utah, Salt Lake City, UT 84112 ([email protected]). B. Houshmand, I. Kulikov, and N. Bryant are with the Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109-8099 (bh8athena.jpl.nasa.gov). Previous Research Numerous studies have been undertaken to use remote sensing for analysis of urban areas (Haack et al., 1987; Khorram et al., 1987; Chavez and Bowell, 1988; Wang, 1993). In general, previous investigations have been plagued by these interrelated problems: limited analytical capabilities due to reliance on a single sensor; spectral and spatial resolution of sensors are too coarse for detailed analysis of urban geometry and texture; inability to discern detail of urban structure, while being extendible for regional scale analysis; for example, large-scale aerial photogrammetry can define detail, but can be cost prohibitive for large regions; and sensors/data products were not and cannot be optimized for characteristics of the urban environment, because urban analysis was not the primary objective in the sensor refinement. Recent research has pointed towards the need for sensor data fusion using various available sensors and ancillary data sources to increase the accuracy of land-cover classification and the inference of human activity and land use. Among the available sensors used for integrative urban analysis are SPOT High Resolution Visible (HRV) multispectral data, SPOT panchromatic imagery, Landsat TM, aerial photographs, and synthetic Aperture Radar (SAR) (Henderson and Xia, 1997; Gouinaud et al., 1996; Ridd, 1995; Weydahl et al., 1995; Gong and Howarth, 1992; Webber and Hirsch, 1992; Haack et al., 1987; Jensen, 1981; Jensen et al., 1994). While the combination of the above imagery data sources used with ancillary data has produced improved capabilities to inventory and map the urban system, research has identified a number of shortcomings in urban classification accuracy and the availability of synchronous ancillary data (Webber and Hirsch, 1992). Census data and existing maps important information for urban area analysis are unavailable in many cities, particularly in the developing world. Also, the above sensors have not been able to measure critical features accurately, such as surface roughness, topography, structural dimensions, and environmental variables (Xia and Henderson, 1997). Topography, which includes the three-dimensional ( 3 ~ ) geometrical patterns of human structures, in addition to the natural topography, is valuable in the assessment of land coverlland use. Interferometric Synthetic Aperture Radar (IFSAR) data were used recently to classify general land covers in several areas of California and Oregon (Rodriguez et al., 1998). According to studies sponsored by the European ComPhotogrammetric Engineering & Remote Sensing, Vol. 64, NO. 8, August 1998, pp. 813-820. 0099-1112/98/6408-813$3.00/0 O 1998 American Society for Photogrammetry and Remote Sensing PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING A u g u s t 1 9 9 8 813 mission DG XI11 Application of Remote Sensing to Urban Areas, a horizontal spatial resolution of 1 to 2 metres is required for extraction of urban geometrical patterns (Casciati et al., 1996). This level of spatial resolution has only become available in recent SAR and IFSAR systems (Madsen et a]., 1993; Soumekh, 1995). Additionally, an important attribute of IFSAR is its orthorectification properties, which eliminates foreshortening of terrain found in conventional SAR imagery. These properties aid in the coregistration of the radar imagery to other sensor data and maps. This paper will attempt to address some of the limitations of previous urban research using an integration of two sensor systems. Every urban area consists of a natural topography overlain by a human-created three-dimensional structure of buildings, roadways, and other features. The Interferometric Synthetic Aperture Radar (IFSAR) system is used to investigate an improved method for definition and measurement of the structural geometry of both the natural and the human-created topographies in the urban area. However, accurate geometric measurement of these features requires knowledge of the textural components of the urban landscape. In addition, the urban area is more than simply the structural geometry of discrete features. The linkages, composition, and relative location of the mix of continuous natural and human phenomena expressed as land covers and land uses are essential to the inventorying and analysis of the urban area. The many spectral channels and the vertical viewing perspective of the Airborne VisibleIInfrared Imaging Spectrometer (AVIRIS) are used in this study to complement the use of IFSAR. Objectives for Integrative Use of IFSAR and AVlRlS The focus of this study is the acquisition of the 3D geometry of the urban area, not the development of a comprehensive land-cover classification for the study area. The objectives are to provide verification that the integration of AVIRIS and IFSAR data can assist in the delineation of urban features to allow more accurate measurement of both human and natural threedimensional geometry; use the AVIRIS data to provide urban textural information corresponding to the geometric information obtained by the IFSAR; and assess the potential research requirements for future integrative use of the two sensors for more detailed and comprehensive urban analysis, inventory and monitoring. Study Area/Sensor Data Used The study area is in the Westwood section of the Los Angeles region, including the University of California, Los Angeles (UCLA) campus. The approximate area is shown in the AVIRIS natural color image and map insert displayed in Plate