Object-Oriented Analysis of Very High-Resolution QuickBird Data for Mega City Research in Delhi/India

Abstract

The 21st century is the century of the cities and of urbanisation. Mega cities in rapidly urbanising regions are hot spots of demographic and socio-economic dynamics. Their rapid growth results in uncontrolled processes of fragmentation which counteracts governance and steering . Due to the high dynamics of mega cities, the traditional methods such as statistical and regional analyses or fieldwork are limited to capture the urban process. New monitoring and planning methodologies are therefore required to provide tools to simplify planning processes in complex urban structures. Remote sensing provides the opportunity to monitor spatial patterns of urban structures with high spatial and temporal resolution. Very high-resolution remote sensing data from the QuickBird satellite are used to identify urban structures and dynamics within Delhi/India. The paper presents a first semi-automated, object-based classification approach which allows the distinction of different settlement types within the urban area. The method was developed for a representative data set within a selected test site and afterwards transferred to the whole test site as well as to a second region to prove the transferability and general validity of the methodology. The research is focused on the identification of "informal settlements" since these represent those characteristic municipal areas which are subject to particularly high dynamics, population density as well as marginalization. The developed methodology allows the recognition of similar settlement types within the urban area. In turn this outcome is compared with a pixel-based classification result to get an idea about the limits of conventional pixel-based classification methods. One of the project aims is the critical assessment of methods and the development of analytical tools. According to this fact the classification results were embedded in a framework for the determination of different socio-economic attributes (e.g. population density) within a certain settlement. This assumption is evaluated - in combination with quantitative and qualitative socio-economic information from about 800 questionnaires emanate from two intense field campaigns - in regard to place- specific infrastructure deficits. In addition, the methodology developed was tested on its transferability and general application. For this purpose other test sites in the Delhi area characterized by similar settlement structures were selected. This is understood to be a first step to the development of indicators which will help to identify and understand the different forms, actors and processes of vulnerability in mega cities.