Land imperviousness reflects settlement growth and urban sprawl. Grounded on a comparative approach, a set of multidimensional statistical techniques were adopted here to quantify the evolution of land imperviousness from Copernicus High-Resolution Layers (HRLs) in a representative case study of Southern Europe (Athens, Greece). A two-way data matrix reporting the percent share of the surface land exposed to different sealing levels (101 classes ranging continuously from 0% to 100%) in the total municipal area was computed for two years (2006 and 2018) individually for 115 municipalities in metropolitan Athens. This matrix represented the information base needed to derive place-specific urban footprints and a comprehensive (global) profile of land imperviousness. Results of a Detrended Correspondence Analysis (DCA) delineated a metropolitan structure still organized along the density gradient, moving from dense settlements in central locations with dominant land classes sealed for more than 90% of their surface area to completely pervious land (0%) typical of rural locations. While the density gradient became less steep between 2006 and 2018, it continued to aliment a socioeconomic polarization in urban and rural districts with distinctive profiles of land imperviousness. Intermediate locations had more mixed imperviousness profiles as a result of urban sprawl. Differential profiles reflect place-specific urban footprints with distinctive land take rates.

This study investigates the land-use/population mix over time as the base to derive an indicator of urban sprawl. Land-use individual patches (provided by Urban Atlas, hereafter UA, with a detailed spatial geometry at 1:10,000 scale) were associated with the total (resident) population based on official statistics (census enumeration districts and other public data sources), providing a comprehensive mapping of the spatial distribution of population density by land-use class in a representative case study for the Mediterranean region (metropolitan Athens, Greece). Data analysis adopted a mix of statistical techniques, such as descriptive statistics, non-parametric curve interpolation (smoothing splines), and exploratory multivariate statistics, namely hierarchical clustering, non-metric multi-dimensional scaling and confirmative factor analysis. The results of this study indicate a non-linear gradient of density decline from downtown (dominated by compact settlements) to peripheral locations (dominated by natural land). Population density in agricultural land was locally high and increasing over time; this result suggests how mixed land use may be the base of intense sprawl in large metropolitan regions. The methodology implemented in this study can be generalized over the whole sample of European cities included in Urban Atlas, providing a semi-automatic assessment of exurban development and population re-distribution over larger metropolitan regions.