Analysis and prediction of land use land cover dynamics in the Kpeshie Lagoon Basin of Ghana using satellite remote sensing

Abstract

Accurate and current Land Use and Land Cover (LULC) maps are important for planning purposes and to monitor the alterations to the environment mostly caused by humans’ activities. The increased utilization of land resources due to population growth have led to loss of biodiversity and urban planning issues such as flooding and pollution. This study analysed the LULC changes within the Kpeshie lagoon Basin of the Greater Accra Region of Ghana and made prediction to the year 2030. Random Forest (RF) classifier was employed to classify the LULC within the study area using Landsat image for four different time-steps (1991, 2002, 2013 and 2020). LULC change analysis was performed for consecutive years (1991 – 2002, 2002 – 2013 and 2013 – 2020) and for the entire period (1991 – 2020). Subsequently, a prediction LULC was made to the year 2030 using a combination of artificial neural network (ANN) and cellular automata (CA) simulations. The LULC classification produced 92.68 %, 84.35 %, 84.41 % and 89.93 % overall accuracies and kappa statistics of 0.87, 0.87, 0.84 and 0.91 for the time-steps respectively. Over the study period, significant LULC changes were observed, as the Kpeshie Lagoon Basin which was predominantly covered by vegetation (69.33 %) in 1991 had transformed into a major built-up area (50.50 %) in 2020. The spatial prediction estimated built-up to cover 60.15 % in 2030, followed by bare land, 32.39 %, vegetation 6.97 % and waterbody 0.49 %. The study revealed that LULC within the Kpeshie Lagoon Basin has been hugely impacted due to urbanization and non-enforcement of regulations.