Assessment of Urban Changes at the Residential Neighbourhood Level Based on Satellite Imageries

Ongoing urban expansion leads to the steady loss of green spaces. Residential units' gardens and green open spaces are a vital part of city life, contributing considerably to urban green infrastructure and ecological services. However, these areas are diverse, making it difficult to assess their changes over time to take advantage of their benefits and contribution to sustainable urban development. This study proposes a new methodology that combines survey data with high-resolution image analysis to construct maps and statistics of change in two residential neighbourhood areas in the Iraqi city of Baqubah. Three change detection techniques utilising very high-resolution multispectral Pléiades images were used to evaluate the changes: pixel value differencing, band index differencing, and categorical change detection. Then, a unique strategy employing geo-processing processes by the ModelBuilder tool was applied to the evaluation outcomes to assess the changes in a final manner. In addition to survey data that supported the final change detection outcomes, study validation was conducted through field verification, and the mean accuracy was ۹۳%. The final results indicated that open or green spaces decreased over a period of seven years at rates of ۲۴% and ۱۴% of the total of both areas assessed. Policymakers and urban planners see such privately owned land as difficult to affect. However, reducing vegetative cover areas and turning them into impermeable surfaces may result in the areas becoming inefficient in the development of urban sustainability. Our developed method demonstrates the capability of utilising Very High Resolution (VHR) imagery with local survey data to accurately infer changes in urban vegetation within residential neighbourhood regions. Doi: ۱۰.۲۸۹۹۱/CEJ-۲۰۲۵-۰۱۱-۰۱-۰۵ Full Text: PDF

کلیدواژه ها:

Urban Change Detection ، Pléiades Satellite Images ، Machine Learning ، Residential Areas ، Urban Sustainability ، ArcGIS.