Spatio-Temporal Modelling of Informal Settlement Using GIS and CA Markov Model; A Case of Dire Dawa City, Ethiopia

Abstract

Informal settlements are urban areas where residents live in conditions without legal recognition of land tenure. This research is aimed to model the spatio-temporal dynamics of informal settlements using GIS and CA-Markov model in Dire Dawa City, Ethiopia. This study utilized satellite imagery from the Landsat 5 MSS (1994), Landsat 7 ETM+ (2004), and Landsat 8 OLI/TIRS for 2014 and 2024. The Random Forest (RF) algorithm was used for classification due to its robustness and accuracy in handling complex remote sensing data. Preprocessing steps, including clipping, cloud and shadow masking, radiometric and geometric correction, were applied to ensure the accuracy of the Landsat imagery. Change detection was conducted using post-classification comparison, where independent classifications for each period were compared to identify areas of change. Landscape metrices such as Edge density, Largest Patch Index, Fractal dimension index and compactness were employed to detect informal settlements. CA-Markov model was used to predict the expansion of informal settlements. Findings of the research revealed that, over the past 30 years, Dire Dawa City has seen notable land cover changes. From 1994 to 2024, Dire Dawa experienced significant land use changes. Agricultural land grew moderately from 8.31% to 15.14%, and urban areas expanded from 0.07% to 2.20%. The analysis of landscape metrics across eight urban zones (1994, 2004, 2014, and 2024) reveals significant growth of informal settlements. Edge density increased by 15-25% across all zones, with patches in Zone 1 and Zone 7 growing by 40-50%. The spatial complexity, measured by the shape index, rose by 10 20% in Zones 3 and 6. Zones 1, 6, 7, and 8 experienced rapid expansion. From 2024 to 2054, Dire Dawa will experience significant land use changes. Built-up areas will increase from 2.20% to 6.88%, and agricultural land will grow substantially from 15.14% to 38.94%. This study highlights how geospatial technologies, including GIS and CA-Markov modeling, effectively analyze and predict the expansion of informal settlements. The integration of satellite imagery, machine learning, and spatial metrics provided detailed insights into land use dynamics in Dire Dawa over the past and future decades. The 2054 simulation predicts more urban and agricultural growth, along with further shrubland loss. The trained CA-Markov model offers reliable future projections, highlighting the need for sustainable land management and conservation strategies.