International audienceUnderstanding the physical vulnerability of buildings and infrastructure to natural hazards is an essential step in risk assessment for large cities. We have interpreted high spatial resolution images, conducted field surveys, and utilized numerical simulations, in order to assess vulnerability across Arequipa, south Peru, close to the active El Misti volcano. The emphasis of this study was on flash floods and volcanic or non-volcanic hyperconcentrated flows, which recur on average every 3.5 years across the city. We utilized a geographic information system to embed vulnerability and hazard maps as a step to calculate risk for buildings and bridges along the Río Chili valley and two tributaries. A survey of ~1,000 buildings from 46 city blocks, different in age, construction materials, and land usage, provided architectural and structural characteristics. A similar survey of twenty bridges across the three valleys was based on structural, hydraulic, and strategic parameters. Interpretation of high spatial resolution (HSR) satellite images, which allows for quick identification of approximately 69 % of the structural building types, effectively supplemented field data collection. Mapping vulnerability has led us to pinpoint strategic areas in case of future destructive floods or flows. Calculated vulnerability is high if we examine structural criteria alone. We further consider physical setting with the most vulnerable city blocks located on the lowermost terraces, perpendicular or oblique to the flow path. Statistical analysis conducted on 3,015 city blocks, considering nine criteria identified from HSR images, indicated that building-type heterogeneity and the shape of the city blocks, along with building and street network density, are the most discriminant parameters for assessing vulnerability

38The focus of this study is an analysis of building vulnerability through investigating impacts 45This study assesses the aspects of building design and site specific environmental hazard proxy in areas where more detailed hydrodynamic modeling data is not available. 50Building design and site-specific environmental conditions determine the physical vulnerability. 51The mathematical approach considers both physical vulnerability and hazard related 52 parameters and helps to reduce uncertainty in the determination of descriptive parameters, 53parameter interdependency and respective contributions to damage. This study aims to (1) 54 enable the estimation of damage probability for a certain hazard intensity, and (2) obtain data 67show that 90% of these tests have a success rate of more than 67%. Probabilities (at building 882011; Thouret et al., 2013Thouret et al., , 2014, and apparent locally high vulnerability of buildings and 89 critical infrastructure in Arequipa, are major motivations for this study. 90Risk in the context of disaster risk management is commonly defined as a potential loss for a 91 given probability function (Crichton, 1999; Kaplan and Garrick, 1981). In the standard 92 conceptual framework, risk is the product of hazard, vulnerability and exposure (Cardona, 93 2004; Carreno et al., 2006). While the hazard is generally described by its severity, e.g. 94inundation height for a given return return period, exposure relates to the number and value of 95 elements potentially affected (Hiete and Merz, 2009). Many different definitions, concepts and 96 methods to systemize vulnerability exist in the current literature (Birkmann, 2006; Cutter, 2003; 97 Wisner et al., 2004; Thywissen, 2006; IPCC, 2007; Bründl et al., 2009). In this study we follow 98 the definition for physical vulnerability proposed by Glade (2003) and Villagran de Leon (2006) 99 as the predisposition of an element or system to be affected or susceptible to damage as the 100 result of the natural hazard's impact. Vulnerability assessment for hydro-geomorphic hazards conducted to study and record structural damage following a hazard event, these data are then 112 generally correlated to the process intensity, frequently derived from deposition height or 113 inundation height, in order to develop empirical fragility curves (Fuchs et al., 2007a,b; Holub 114 and Fuchs, 2008 133Flash floods are common in semi-arid areas, such as Arequipa, and can, despite their 134 infrequent nature, have a devastating effect in both geomorphological and human terms 135 (Gaume et al., 2009; Jonkman and Vrijling, 2008;Martínez Ibarra, 2012). The occurrence of 136 flash floods is highly variable, both spatially and temporally, most occurring as the result of 137 localized intense storms (Graf, 1988;Reid and Frostick, 1992; Hooke and Mant, 2000 , 2007a,b; Holub and Fuchs, 2008). 160Several recent studies (Martelli, 2011;Santoni, 2011; Ettinger et al., 2014a,b; Thouret et al., 161 2013 Thouret et al., 161 , 2014 192On 8...