Mapping urban green structures using object-based analysis of satellite imagery: A review

Abstract

Urban green structures (UGS) play important roles in enhancing urban ecosystems by providing benefits such as mitigating the urban heat island effect, improving air quality, supporting biodiversity, and aiding in stormwater management. Accurately mapping UGS is important for sustainable urban planning and management. Traditional methods of mapping such as manual mapping, aerial photography interpretation and pixel-based classification have limitations in terms of coverage, accuracy, and efficiency. Object-based image analysis (OBIA) has gained prominence due to its ability to incorporate both spectral and spatial information making it particularly effective for classification of high-resolution satellite data. This paper reviews the application of OBIA on satellite images for UGS mapping, focusing on various data sources, popular segmentation methods, and classification techniques, highlighting their respective advantages and limitations. Key segmentation methodologies discussed include multi-resolution segmentation and watershed segmentation. For classification, the review covers machine learning techniques such as random forests, support vector machines, and convolutional neural networks, among others. Several case studies highlight the successful implementation of OBIA in diverse urban environments by demonstrating improvements in classification accuracy and detail. The review also addresses the challenges associated with OBIA, such as dealing with heterogenous urban landscapes, data sources and with OBIA methods itself. Future directions for UGS mapping include the integration of deep learning algorithms, advancements in satellite data technologies, and the development of standardized classification frameworks. By providing a detailed analysis of the current state-of-the-art in object-based UGS mapping, this review aims to guide future research and practical applications in UGS management.