Figure 3
From: A model for simulating emergent patterns of cities and roads on real-world landscapes
Quantitative comparison between the population and traffic datasets for Italy and the LTP simulation results for different landscape scenarios. (a) Population distribution from the 2011 census data for Italy (red) and the LTP simulation (blue) when aggregated at the Regione level for the ’Water bodies’ scenario. (b) GPS-tracked traffic distribution from OpenStreetMap for Italy (red) and the LTP simulation (blue) when aggregated at the Regione level for the ’Water bodies’ scenario. (c–f) Comparison in population distribution as in (a), but at Provincia level for each scenario. Error bars depict the standard deviation that comes from the 64 simulation runs for each scenario. (g,h) The Kullback–Leibler (KL) distance from the simulation data to the census population (g) and the GPS-tracked traffic datasets (h) for Provincia-level distributions. The error bars are the same as in (c–f).
