Assessing the fragmentation of the green infrastructure in Romanian cities using fractal models and numerical taxonomy

As the share of urban population increases globally each year, man-dominated systems tend to sprawl over the natural ones, substituting and fragmenting them. Urban sprawl is the main cause of many environmental issues, in tight connection with pollution and loss of biodiversity. One of the main consequences is a decrease of the ecosystem services provided by the urban green infrastructure. However, the extent of urban sprawl is spatially uneven due to the spatial structure of human settlements. Among the methods used to pinpoint sprawl, fractal analyses have a good potential for analyzing fragmentation, especially if used in conjunction with statistical methods. This study aimed to assess, in an exploratory perspective, the level of fragmentation in the Romanian cities covered by the Urban Atlas data, and determine its correlation with parameters related to their demographical and physical characteristics. In addition, taxonomical analyses were used to find whether cities or specific components of the green infrastructure can be grouped. The results did not reveal a general trend, although it seems that the green infrastructure consisted of agricultural/ semi-natural/ wetland areas, forests, green areas, sports and leisure facilities and water bodies in all of them, in different shares; with respect to their distribution, the numerical taxonomy analysis indicated that they form classes matching the types of 'nature in the city' previously described by ecologists, despite the particular historical evolution of each city and its particular influence on urban planning. The correlation analysis revealed that the population and its density and the share of the green infrastructure within the total area are significantly correlated with most fractal parameters. Similarly, the fractal dimension of the area, computed using Interactive Quantitative Morphology, seems to correlate with most morphological parameters. However, the taxonomical analysis of cities did not find very relevant groups due to the fact that many large Romanian cities lack Urban Atlas data. The results suggest that the degree of urban fragmentation is correlated especially with the population of cities and its density, reclaiming planning measures aimed at controlling the densification processes (sprawl, gentrification, location of specific activities etc.) (C) 2016 The Authors. Published by Elsevier B.V.