Ecological Network Analysis in Urban Areas Through Graph Theory and Remote Sensing

Urban green areas play a crucial role in supporting the provision of ecosystem services, and are pivotal in urban landscape planning and in enhancing the ecological connectivity of cities. However, urban sprawl poses challenges to urban vegetation connectivity and biodiversity, being one of the main causes of landscape fragmentation and degradation. Our paper aimed to introduce a standardized and comparable approach for evaluating the connectivity of green urban areas across Europe. The goal was to evaluate the landscape connectivity of urban areas in 28 European Capital cities and to establish rankings based on two key criteria: the percentage of vegetation and overall landscape connectivity. First, we implemented data from the European Union Earth Observation program—Copernicus—to construct the European Urban Vegetation Maps (EUVMs). EUVMs referred to 2018, had a spatial resolution of 10 m, and categorized dominant vegetation cover into trees, shrubs, and grass across 28 European Capital cities at two official geographical levels: Local Administrative Unit (LAU) and Urban Core areas (UC). Based on the EUVMs, the percentage of vegetation was calculated per each Capital city in Europe. The accuracy assessment against field surveys conducted under the Land Use and Coverage Area Frame Survey confirmed the EUVMs’ potential, with an overall accuracy of 83.57%. Further, to evaluate the connectivity of green urban areas we conducted a landscape network analysis using Graphab software. Our analysis examined three global metrics—Probability of Connectivity (PC), Integral Index of Connectivity (IIC), and Equivalent Connected Area (ECA)—as well as networks’ nodes and links features within each city. Results showed a mean urban vegetation percentage of 49.43% within the LAU areas, with Southeast European Capitals, such as Sofia, Ljubljana, and Zagreb, exhibiting over 75%. The network analysis showed how Stockholm, despite having the highest average patch capacity and ECA, exhibited an uneven distribution of patch capacity (Gini coefficient 0.96), while Ljubljana recorded the largest values in PC and IIC metrics (0.131 and 0.098, respectively). In conclusion, future research should (i) consider species-specific habitat requirements to develop effective conservation strategies, (ii) investigate public accessibility and urban green area utilization, and (iii) identify standardized methodologies for city boundaries identification to improve international comparability. In this context, Copernicus data can be leveraged for urban planning and ecological connectivity analysis.