Predicting land use land cover changes and impact on urban wetlands using cellular automata and artificial neural networks approach, a case study in Greater Accra, Ghana

Abstract

Satellite imagery was used to map and predict potential future land use and land cover (LULC) changes and impact on wetlands in the Greater Accra Metropolitan Area (GAMA), Ghana for establishing appropriate urban planning policies and management methods. The research used classification and regression tree (CART) analysis with Landsat imagery from 2000, 2011, and 2020 to identify LULC changes and to project potential scenarios in 2030 and 2040. In the integrated cellular automata (CA) and artificial neural networks (ANN) (CA-ANN) framework, the transition potential was computed by ancillary driver variables that influence change including elevation, slope, NDVI, annual precipitation, distance from roads, and population density. The validation of the simulated LULC maps for 2020 produced an overall agreement of 86.26 % and a Kappa of 0.78. Future projections indicate that urban development and sprawl are expected to increase at an annual rate of up to 0.9 %, while wetlands and vegetation are projected to decline at annual rates of up to 2.6 % and 2.9 %, respectively. Results from the driver variables suggest that while the major road network in GAMA promoted the spread of urban expansion, the topographic constraints (i.e., slope and elevation) hindered the urban expansion. Urbanization is likely to have a more detrimental effect on wetlands in the near future, less so in the distant future.

Keywords

• Land use/cover change
• Wetlands
• Artificial neural networks-cellular automata
• Prediction
• Urban sprawl