Ontology of Informal Settlements in Riyadh, Saudi Arabia with Geospatial Intelligence

Abstract

Any management policies developed for managing the urban growth of a city, and to ensure the sustainability of that growth into the future, must understand the spatial distribution of informal settlements found within the city boundaries. These settlement types can be found in a large number of metropolitan areas around the world. Accurate identification requires an understanding of the various characteristics which tend to be associated with these settlement areas, including the types of materials used in building construction and the unique street patterns found within the settlement neighbourhoods. Due to the dynamic nature of these settlements, however, the existence of a universally-accepted framework which can be used to define and map these areas is lacking. This study aims to integrate local knowledge, remote sensing data, and machine learning to investigate and develop an informal settlements ontology for use within the Arabian Peninsular region. Information used included very high to medium resolution satellite images, field surveys, expert opinion regarding local conditions, and a wide range of geographic data. Object-based image analysis (OBIA), machine learning methods such as random forest (RF), and support vector machine (SVM), expert knowledge, and various geographic datasets were employed to identify the distribution of informal settlements over time and space within Riyadh, the capital city of Saudia Arabia. Many variations in settlement character can be found, so the development of a local ontology of informal settlements (LOIS) has been proposed. The major findings of the current work are: (i) the inclusion of local expert knowledge (EK) about the various spatial, spectral and textural image attributes identified, can enhance the identification of informal settlements over time and space; (ii) a combination of OBIA-RF and OBIA-SVM, augmented by local knowledge, can improve the image-based classification of informal settlements, and; (iii) the approach taken, involving the selection of thirty unique geospatial indicators, was found to be very useful in the study of informal settlements over time, particularly when processing the very high and medium resolution satellite images in tandem with the Landtrend tool. The efficacy of the proposed approach, in regards detailing the spatiotemporal growth of identified areas, was shown by the accuracy of the mapping result outputs.

The findings of this work may prove to be of value to urban planners and decision-makers when designing and constructing new infrastructure (roads, water, and energy plants), and when assessing the potential for any adverse environmental, social, or financial impacts associated with informal settlements - specifically within the Arabian Peninsula, but also in other regions of the world with similar characteristics.