Urban sprawl analysis in Kutupalong Refugee Camp, Bangladesh

Abstract

Urban sprawling is a common phenomenon associated with geographical and political challenges such as refugee settlements and environmental extremes. Urban sprawl related to refugee or habitation settlement has been an area of active interest because of humanitarian and environmental problems. For example, higher rates of urban sprawling are positively correlated with higher rates of deforestation. The present study explored the viability and reproducibility of different classification techniques in assessing urban sprawl among Rohingya refugees in the Kutupalong refugee camp in South-Eastern Bangladesh. Two classification techniques were used to assess the urban sprawl among the study population. These classifications include the Support Vector Machine and Maximum Likelihood Classifier. The sprawl was measured based on the classification of urban ad non-urban classes, according to the topography of the camps. The study showed that urban class exhibited exponential growth from 2.01 km2 to 5.37 km2 within nine months based on Support Vector Machine Classifier, while Maximum Likelihood Classification detected 3.2 km2 to 7.8 km2 of urbanization. On the contrary, the non-urban class shrunk from 12.58 km2 to 9.95 km2 during the same period with Support Vector Machine and 11.3 km2 to 6.7 km2 with Maximum Likelihood Classification. The Support Vector Machine yielded better overall accuracy performance compared to Maximum Likelihood Classification.