Mapping Informal Settlements Using Machine Learning Techniques, Object-based Image Analysis and Local Knowledge

The presence of informal settlements within Riyadh City, in the Kingdom of Saudi Arabia (KSA), has led to various urban planning issues. The current study examines these settlements using object-based machine learning (ML) methods (Random Forest (RF) and Support Vector Machine (SVM)), expert knowledge (EK) and very high resolution (VHR) WorldView-3 imagery. The study objective was to improve current mapping and planning techniques by incorporating EK in the identification process. Unique informal settlement indicators identified by the urban specialists were used in a combined ML and image analysis (OBIA) approach. Results show that combining ML, EK and remotely-sensed data in the analytical work can efficiently, effectively and accurately distinguish unplanned areas from other areas. This provides a very useful method for mapping informal settlements.