Damage Assessment with Very-High Resolution Optical Imagery Following the December 26, 2003 Bam, Iran Earthquake Luca Gusella 1 , Charles K. Huyck 2 , Beverley J. Adams 2 , Sungbin Cho 2 , Hung Chung 2 This paper presents a methodology for quantifying the number of buildings that collapsed following the Bam earthquake. The approach is ‘object’ rather than ‘pixel’-oriented, commencing with the inventory of buildings as objects within high-resolution Quickbird satellite imagery captured before the event. The number of collapsed structures is computed based on the unique statistical characteristics of these objects/buildings within the ‘after’ scene. A total of 18,872 structures were identified within Bam, of which the results suggest that 34% collapsed - a total of 6,473. Preliminary assessments indicate an overall accuracy for the damage classification of 70.5%. INTRODUCTION Over the past decade, the value of optical remote sensing technology for city-wide damage assessment has been increasingly recognized (see Matsuoka and Yamazaki, 2000; Eguchi et al., 2000, 2003; Estrada et al., 2001; Yamazaki, 2001; Adams, 2004, Huyck et al., 2004). As very high-resolution commercial imagery has become available from sensors such as Quickbird and IKONOS, it is now possible to identify damage to individual structures (Chiroiu et al., 2002; Adams et al., 2003, 2004; Saito et al., 2004). This paper summarizes an “object-oriented” methodology for counting the number of buildings that collapsed in Bam - a valuable statistic that was unavailable in the aftermath of the event a . Analytically, this approach differs from traditional “pixel-based” optical analyses used to develop damage maps following the 1995 Kobe, Japan (Matsuoka and Yamazaki, 1998), 1999 Marmara, Turkey (Eguchi et al., 2002; Huyck et al., 2004), and 2001 Boumerdes, Algeria (Adams et al., 2003; Yamazaki et al., 2003) earthquakes. The paper commences with an outline of the methodology and goes on a summary of conclusions and directions for future work METHODOLOGY An object-oriented methodology (De Kok et al., 1999, Benz et al., 2004, Bitelli et al., 2004) is used here to quantify the number of collapsed buildings in Bam i . In theoretical terms, this approach focuses on the analysis of “objects” within the imagery (in this case buildings), rather than individual pixels. For traditional pixel-based studies, each pixel is treated as an analytical unit characterized by reflectance (DN), thematic and processed (e.g. texture) values (Schowengerdt, 1983). In contrast, objects are vector (line/polygon) units encompassing a group of related pixels. They are characterized by the intrinsic properties of the constituent pixels, together with additional relational features that operate in a multi-level hierarchy with respect to their neighbors (Baatz and Schape, 2000). From Table 1, the present study employs high-resolution standard product Quickbird imagery, collected before and in the immediate 1 DISTART, University of Bologna, Viale Risorgimento, 2 – 40136 Bologna - ITALY 2 ImageCat, Inc, Union Bank of California Building, Ste 1050, 400 Oceangate, 90802 Long Beach (CA) -US Huyck