Analyzing Land Cover Change and Urban Growth Trajectories of the Mega-Urban Region of Dhaka Using Remotely Sensed Data and an Ensemble Classifier

摘要: Accurate information on, and human interpretation of, urban land cover using satellite-derived sensor imagery is critical given the intricate nature niches of socioeconomic, demographic, environmental factors occurring at multiple temporal spatial scales. Detailed knowledge their changing pattern over time periods associated with ecological risk is, however, required for best use its resources. Interest in this topic has increased recently, driven by a surge open-source computing software, imagery, improved classification algorithms. Using machine learning algorithm Random Forest, combined multi-date Landsat we classified eight maps up-to-date between period 1972 2015 mega-urban region greater Dhaka Bangladesh. Forest—a non-parametric ensemble classifier—has shown quantum increase image accuracy due to outperformance traditional approaches, e.g., Maximum Likelihood. Employing Forest as an approach study independent cross-validation techniques, obtained high accuracy, user producer accuracy. Our overall ranges were 85% 97% kappa values 0.81 0.94. The area statistics derived from thematic map show that built-up 43-year expanded quickly, 35 km2 378 2015, net rate approximately 980% average annual growth 6%. This rate, was higher peripheral areas, 2903% expansion 8%, compared 460% 4% core city (Dhaka City Corporation). huge took place north, northwest, southwest regions Dhaka, transforming areas previously agricultural land, vegetation cover, wetland, water bodies. main driving towards northern corridors include flood-free availability transportation network, agglomeration manufacturing-based employment centers. resulting produced therefore serve facilitate detailed understanding dynamics change patterns