SATELLITE IMAGE BASED URBAN CHANGE DETECTION: THE CASE OF HAWZIEN TOWN IN NORTHERN ETHIOPIA

Abstract

Information on Land use/Land-cover change is imperative to local, regional and wide-range planning, decision making and ecosystem monitoring. Such planning and management activities are seriously vulnerable to blunders due to the deficiency of reliable and timely information on dynamics of land cover /land-use change. Thus, it is indispensable to study urban changes to evaluate what has changed and to predict future changes of growth and expansion. So far, minimal effort has been applied in Hawzien to precisely and accurately track the land use/land cover and urban changes using widely and commonly used satellite images. Therefore it is necessary to device a model that can automatically detect changes given satellite images of different points in time. Such automated systems will assist to identify, monitor and predict the growth and expansion of urban development This study presents the methods and results of pixel-based, object based and feature based change detection algorithms to spot changes in urban areas between two specific time periods using Land sat images. The implemented model classifies developed areas of different densities as well as undeveloped lands that are likely to be degraded by close proximity to development. This tool can also analyze land cover data for multiple time periods for a given city. For each time period, two maps are generated showing the density classes of the built-up area and the undeveloped land degraded by proximity to the built-up area. A prototype that automatically detects changes that happened between 1986 and 2000 in Hawzien city is implemented. This prototype was developed using Remote sense and has an ability to be integrated into ArcGIS. An easy to use graphical user interface is also developed using Microsoft visual studio 2010 to assist users with minimal or no skills of using ArcGIS. The result of the work shows that combination of feature based, object based and pixel-based Change detection methods exploiting spatial metrics and texture measurements is a potential new avenue to extract detailed urban land-use information from high or medium resolution satellite imagery. Managers, policy makers and planners could use this data as a decision-support tool. The model was evaluated by professional users on its learn ability and efficiency. The results show that the model has a capability of extracting urban change information with few modifications required. The overall feeling of users towards using the model is found to be 86.67%.