Comparative evaluation of LULC classification methods and urban expansion simulation using CA-Markov in Hyderabad, Sindh (1990–2050)

Abstract

Urban expansion poses significant challenges in developing countries, affecting land-use patterns, environmental sustainability, and urban management. Hyderabad, the second-largest city in Sindh, has experienced considerable urban growth during the past three decades, highlighting the need for a quantitative assessment of its spatial dynamics. This study analyzed built-up area changes in Hyderabad from 1990 to 2022 using Landsat imagery (TM, ETM+, OLI and TIRS) and explored potential urban growth patterns up to 2050 using Cellular Automata–Markov modeling.

Six classification methods—Maximum Likelihood Classification (MLC), Support Vector Machine (SVM), Artificial Neural Network (ANN), Minimum Distance (MD), Mahalanobis Distance (MHD), and Spectral Angle Mapper (SAM)—were applied to map built-up areas, and classification performance was evaluated using overall accuracy and the Kappa coefficient. The results show variations among classification outputs, reflecting differences in algorithm design and sensitivity to spectral characteristics. Several classifiers, including MLC, ANN, and SVM, achieved relatively high classification accuracy in mapping urban areas.

Directional analysis indicates that urban expansion has predominantly occurred toward the north, north-east, and south-east of the city. Shannon entropy analysis was used to examine the spatial dispersion of built-up areas, indicating a gradual shift from relatively concentrated growth patterns to more dispersed urban expansion over time.

CA–Markov simulations were used to explore potential future urban expansion scenarios for Hyderabad up to 2050 based on historical land-use transitions. The projections suggest continued outward expansion of built-up areas into surrounding peri-urban regions. Overall, the study highlights how differences in classification approaches may influence the representation of urban growth patterns and provides insights that may support sustainable urban planning and land-management strategies in Hyderabad.