Enhancing urban growth prediction with the Spatio-Temporal Matrix: Case studies from Vietnam, India, and Ivory Coast

Kurzfassung

Urban growth prediction is essential for sustainable urban planning, requiring accurate and reliable models. Satellite-based Earth observation (EO) time series data offer valuable insights into past and allow to conclude on future trends of urban development. However, existing models often struggle to incorporate detailed local information, leading to inaccuracies in growth predictions, or require plenty additional datasets. To address this challenge, we utilize the Spatio-Temporal Matrix (STM) approach, which leverages EO data to generate spatial and temporal characteristics of neighborhoods. In this study, we applied the STM-based model coupled with a multi-layer perceptron (MLP) for settlement growth prediction in Huê (Vietnam), Surat (India), Ho-Chi-Minh City (Vietnam), and Abidjan (Ivory Coast). Our research aims to assess the model's ability to predict settlement growth while avoiding restricted or intra-urban areas without incorporating additional datasets besides multitemporal settlement layers (World Settlement Footprint – WSF). Using the WSF-evolution dataset, we achieved promising results, indicating the STM-based model's effectiveness in settlement growth prediction. Compared to baseline results of the SLEUTH model, our approach produced more realistic growth patterns, minimizing predictions in restricted areas without the need for additional layers. This study highlights the potential of the STM-based model as a reliable tool for urban growth prediction based on EO information products, offering enhanced accuracy and independence from external datasets. By providing insights into future urban development while respecting local constraints, our approach contributes to more sustainable urban planning practices.