Land use and density in the European city: monocentric analysis and scaling

In this work we study the evolution of land use and population density in European cities with respect to the distance to the city centre, and how land use and density curves scale with city population. The principal aim of this research is to provide stylized facts and generic formulas for the calibration of monocentric urban economic models to European cities. We use the GMES Urban Atlas database, providing a precise description of land use in the 305 major European larger urban zones (LUZ), which have more than 100.000 inhabitants. For the population density we use the Geostat population grid, which covers the whole of European Union (EU) with a 1km$^2$ grid dataset.

We combine the land use and population datasets by attributing the population of the Geostat grid to the geographic units of the (much more precise) land use dataset. This attribution is proportional to the surface of each unit and weighted thanks to the nomenclature of the GMES Urban Atlas, which gives classes of soil sealing and density of urban fabric. Then we analyse the evolution with distance to the city centre, which we define for convenience as the location of the city hall, of population density and of the share of land used for different purposes: housing, roads, railways, urban green, water, agriculture, forest. To this end, we define concentric rings of fixed width around the city centre, in which we average land use and population.

In order to compare results between different cities and to obtain a global picture ("average" or "standard" European city), we study simple scaling relationships for the obtained monocentric land use shares and density curves. We choose to study the scaling of these curves with respect to the city population, as has been done in the literature for different parameters (like income or road space). In our case city population is taken as the population of the LUZ, computed thanks to the population grid.
It turns out that land use curves, in particular housing or roads shares, tend to scale like the square root of city population.
Population curves have roughly exponential shapes, as widely modelled in the literature, and tend to scale like the city population to a power close to $1/3$.
These results allow us to propose a simple monocentric description of land use shares and population curves in a representative European city, whose size can be chosen based on the scaling relationships we obtain. This result is especially interesting, and of practical use, for the purpose of calibration of monocentric urban models.