URBAN SOCIAL VULNERABILITY ASSESSMENT USING OBJECT-ORIENTED 
ANALYSIS OF REMOTE SENSING AND GIS DATA. 
A CASE STUDY FOR TEGUCIGALPA, HONDURAS. 
A. Ebert 3 , N. Kerle b 
d UFZ - Helmholtz Centre for Environmental Research, Permoserstr. 15, D-04318 Leipzig, Germany - 
annemarie.ebert@ufz.de 
b ITC - International Institute for Geo-Information Science and Earth Observation, PO Box 6, 7400 AA Enschede, The 
Netherlands - kerle@itc.nl 
Commission VII, WG VII/7 
KEY WORDS: Social vulnerability assessment, Remote sensing, Object-oriented analysis, Proxy variables, Tegucigalpa 
ABSTRACT: 
This paper deals with the assessment of social vulnerability (SV) as a critical component of comprehensive disaster risk assessment. 
Indicators for SV relate to aspects on different scales. Individual characteristics, such as gender, age and education level, have to be 
assessed on a very local (individual) scale, whereas indicators such as living conditions, economic development and location of the 
household can be assessed on the scale of a building, building block, an administrative neighbourhood or city district. In turn, 
measures to reduce SV and thereby the disaster risk are taken on different levels. Information on SV is notoriously difficult to obtain, 
and traditionally either detailed field studies or census data have been used.This research, which was done in Tegucigalpa, Honduras, 
is not focused on individual people, but on the level of buildings and administrative neighbourhoods in the city, with the intention to 
analyse SV for a central area of 3*3 km as a potential starting point for more detailed analysis if needed. The central novelty is the 
use of image-based contextual, object-oriented analysis and the focus on physical proxies as indicators for SV, whereby we focus on 
landslide and flood hazards.Very high resolution remote sensing data, as well as GIS data and city maps were applied to delineate 
proxy variables with the goal to analyse four indicators for social vulnerability: (i) socio-economic status, (ii) commercial and 
industrial development of a neighbourhood, (iii) abundance of infrastructure/lifelines, (iv) and distance to those. The validation of the 
results was done using a Social Vulnerability Index (SVI) created based on census data. A subsequent stepwise regression analysis 
showed that eight out of 47 proxy variables were significant and could explain almost 60 % of the variation of the SVI, whereby the 
slope position (i.e. location of a building) and the proportion of built-up area in a neighbourhood (i.e. neighbourhood composition) 
were found to be the most valuable proxies. To make the approach transferable to other study areas with different data availability we 
also indicate where data can potentially be substituted with lower quality information than applied in this study. This work shows that 
contextual segmentation-based analysis of geospatial data can substantially aid in SV assessment, and, when combined with field- 
based information, leads to an optimisation of the assessment in terms of assessment frequency and costs. 
1. INTRODUCTION 
The number of casualties and damage caused by natural 
disasters, while fluctuating strongly, has been increasingly in 
recent decades. Higher monetary values lead to large losses in 
developed countries, whereas in less developed countries the 
amount of people injured or killed and the loss of basic 
infrastructure are the main concerns after a disaster struck an 
area. Different living conditions and standards usually lead to 
different abilities of people to prepare for and cope with natural 
disasters, thus lead to a certain spatial pattern of vulnerability 
and resilience. This important fact needs to be considered in risk 
management. Knowledge about present and future hazards, 
elements at risk and different types of vulnerability are the 
place-dependent variables that need to be assessed in order to 
mitigate disaster risk. In practice, a lack of sufficient data is a 
main constraint in disaster risk assessment, especially in data 
scarce regions, such as less developed countries. 
2. STUDY AREA AND DATA BASE 
The hypothesis that remote sensing data can be used for SV 
assessment was tested on 87 neighborhoods (3*3 km) of 
Tegucigalpa, the capital of Honduras, a city with more than 1 
million inhabitants. Due to a lack of urban planning instruments 
many people are living in areas that are prone to hazards such as 
landslides and floods, which are often the only available and 
affordable spaces for building construction (Angel et al., 2004). 
The strong population growth leads to ecological and land use 
changes that increase the hazard and thus the risk. When 
Hurricane Mitch struck the city in 1998, numerous landslides 
and severe flooding caused thousands of fatalities and destroyed 
large parts of the city, including buildings and infrastructure. 
Very high resolution satellite data, a digital elevation model 
(DEM), a normalized digital surface model (nDSM), a set of 
different GIS data, census data and scattered ground truth 
information were available for the analysis (Table 1).