Urban Spatial Growth Modeling and Prediction using an Inte-grated Model of Artificial Neural Network and Cellular Au-tomata
Year: 2026
Type: preprint
Abstract: The fast urbanisation and dynamic urban transformations resulting from population growth and migration require the identification and detection of such changes for effective urban management and sustainable development. Urban growth modelling is an important analytical tool to understand and address the complex challenges of urbanisation. A review of existing literature shows several research gaps. Most of the existing urban growth prediction models consider only geographical, topographical or socio-economic factors separately. Existing Cellular Automata (CA)-based models usually take into account only the current state of the neighbouring cells and the accuracy of their predictions is mainly based on the accessibility index or suitability index of the cells. To address these limitations, this paper proposes an integrated model that combines the strengths of CA and ANN to predict urban growth. The developed model first generates four-class LULC maps consisting of vegetation cover, surface water, bare soil, and urban land from the corrected imagery. These classes are then reclassified into two categories namely urban and non-urban. Geographical factors such as proximity from the urban center, roads, railways, water resources, and town centres, along with neighbourhood development intensity, estimates accessibility. Topographical factors like aspect, soil type, groundwater quality, and flood inundation determines suitability and socio-economic factors including population density, proximity to economic centres, and land value defines zoning. The model performance is evaluated using Percent Correct Match, Moran’s I, and Shannon’s Entropy. A confusion matrix, overall accuracy, and kappa statistics were used to evaluate classification accuracy. In addition, the model assessed land-use transitions through a transition matrix, conducted sensitivity analysis, and quantified urban area changes using statistical measures. Satellite images of Thoothukudi District, Tamil Nadu, India were pre-processed and classified into LULC (land use land cover) categories using the developed integrated model. Change detection techniques were used to identify changes such as urbanisation, loss of vegetation and changes in water bodies over time. These patterns are used to predict future LULC distribution for the year 2030 to help sustainable planning and environmental conservation. Clinical trial registration Not applicable. This research does not involve any clinical trials.
Source: Research Square
Language: English
Subfield: Global and Planetary Change
Field: Environmental Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7215082895