Land use land cover change detection and urban sprawl prediction for Kuwait metropolitan region, using multi-layer perceptron neural networks (MLPNN)

With the rapid expansion of cities, monitoring urban sprawl is recognized as a vital tool by many researchers who use this information in several applications like urban planning, microclimate modelling, policy development, etc. However, accurate land cover (LC) prediction is still challenging, even with technological advancements. Machine learning (ML) and artificial intelligence (AI) have gained a reputation amongst diverse science applications, including their popularity in monitoring land cover. Therefore, the present study investigates the performance of the ML-based classification algorithm random forest (RF) in monitoring LC classes for 2016 and 2021 for the metropolitan region of Kuwait City, Kuwait. The accuracy assessment for the derived land use maps achieved an overall accuracy of 93.6% and 95.3% and kappa coefficient values of 0.86 and 0.93 for 2016 and 2021, respectively. The results show an increase in built-up cover by ∼11 %. The land use maps for 2016 and 2021 were further used to predict the urban built-up for 2026 using an artificial neural network (ANN) based on multi-layer perceptron neural networks (MLPNNs). It was predicted with an overall accuracy of 83.6%. The built-up was predicted to increase by 15% in 2021–2026, and mostly expansion was observed on the western and southern sides. The outcomes exhibit that MLPNN techniques combined with Remote sensing and Geographic Information Systems (RS and GIS) can be adopted to derive the land cover and predict the urban sprawl with fair accuracy and precision. Such studies would prove valuable to city governments and urban planners to improve future sustainable development strategies.