Urban expansion modeling with a semi automated spatial Markov chain framework

Abstract

Urban growth modeling is essential for effective urban planning, yet conventional techniques often lack spatial precision and rely heavily on manual image processing. This study proposes a semi-automated methodology using Spatial Markov Chains (SMC) to enhance urban growth projection by incorporating a novel image pre-processing technique. This approach combines automated image treatment with SMC modeling to estimate not only the probability of urban expansion but also its spatial placement, addressing a critical gap in urban growth forecasting methods. Using Landsat 7 and 8 satellite imagery, six Mexican cities–Acapulco, Puebla, Querétaro, Tampico, Tijuana, and Toluca–were selected as case studies. Data from 2003 were used to project urban expansion for 2017 and 2031, with four performance metrics–Kappa index, Jaccard index, Shannon’s entropy, and fractal dimension–applied to evaluate model accuracy. Results indicate that this approach produces reliable projections with a spatial accuracy exceeding 85%, as validated by metrics such as the Kappa index (0.75–0.81) and the Jaccard index (0.68–0.92). These results highlight the method’s capability to outperform traditional models in terms of precision and spatial detail, making it a robust tool for urban planning. The findings demonstrate the potential of semi-automated SMC methods to improve the precision of urban growth models, offering a valuable resource for policymakers and urban planners seeking to manage sustainable development effectively.