Application of urban growth boundary delineation based on a neural network approach and landscape metrics for Khulna City, Bangladesh

Abstract

The rapid and unprecedented urban growth in Khulna, Bangladesh is making it difficult to implement measures to limit further expansion and define clear administrative boundaries, which is posing a significant threat to the environment and ecological sustainability. Using an Artificial Neural Network (ANN) based urban growth simulation model and landscape metrics, this study aims to evaluate the spatial extent and direction of urban growth and demarcate an Urban Growth Boundary (UGB) by examining the future contiguous expansion of the city for implementing effective land use provision. Utilizing data on biophysical, proximity, neighborhood, and market factors over the past twenty years, the neural network with Markov chain model allocates the land demand for buildup area by 2020 and 2030, concerning twelve explanatory variables. The simulated map of the urban area is further used by landscape metrics to quantify local-level urban patch information viz. landscape pattern, size, aggregation, etc. The compact patch characteristics are mostly found under the Kotwali thana, while, fragmented and unstructured patches are prevailing between urban–rural interfaces. Finally, there has around 95 km2 gap between the existing service provided by KCC and the future demand of Khulna city, creating an imbalance between the supply and demand of urban services. Hence, restricted urban growth would make government investment in service facilities cost-effective and enable planners and decision-makers to intend a feasible trade-off between future land demand and the protection of natural resources.