The Dark Side of the Earth: Benchmarking Lighting Access for All Cities on Earth and the CityNet dataset

In this paper, we present an analysis of urban form, defined as the spatial distribution of macroeconomic quantities that characterize a city such as population, built environment, and energy use. In particular, we develop a framework to study the question of “mismatch” between the spatial distribution of lighting levels observed in a city (which was previously shown to be a proxy for energy access and wealth levels) and that city’s population density and built area distribution. This allows us to rank cities globally by their ability to, intuitively, “match people with lighting/energy access”. For this, we develop and make available a derived dataset we call CityNet that is based on best-available open-source remote sensing data products for the world’s largest 30,000 cities. We first describe how cities may be grouped into a few classes by the scale magnitude of these key quantities. Then we introduce simple quantities to measure their spatial distributions such as the average radial profile, the discrepancy between the radial profile of population and that of other quantities, and the effort of transforming the distribution of a given quantity to match that of the population density. To compare a given city against its “peers”, we define a simple benchmark model of urban form. We use this model to rank cities by the relative magnitude of the lack of access to built and energy infrastructure. The key observation we make is that in many parts of the world, development (including built area density and energy access and use) does not follow the spatial population distribution.