Cellular Automata approach in dynamic modelling of land cover changes using RapidEye images in Dhaka, Bangladesh

Abstract

Satellite images have been used extensively to identify the land use/land cover (LULC) changes in Bangladesh. However, no study has been conducted to classify LULC changes in the Dhaka Metropolitan Development Plan (DMDP) area using high-resolution commercial satellite images. The study aimed to simulate future LULC scenarios using RapidEye commercial images in the fastest-growing DMDP area. Support Vector Machine algorithm was applied to estimate the LULC scenarios for years 2012, 2015, and 2018. Cellular Automata machine learning algorithm was used to simulate the future LULC scenarios for 2025. The study result revealed that a significant net increase in the urban areas (UAs) by 15.52%, a remarkable decrease in sparse vegetation (SV) by 12.48%, and a transformation of 17.83% green cover (SV and dense vegetation) areas by 14.95% (8.9%/year) UAs were found from 2012 to 2018. Prediction results demonstrated that UAs would likely to be expanded by 53% and SV will be reduced by 13% (28% was in 2012) in 2025. The outcomes of this study will help the city authorities of DMDP in preparing a comprehensive micro-level urban development plan, where planned infrastructural development and supervision, land use planning, natural resource conservation, and environmental sustainability will be ensured.