Proxy datasets to identify city boundaries

Abstract

City boundaries can be delineated using various proxy datasets, such as population, built-up area and night-time lights. Each proxy reflects different assumptions about what constitutes a city, and each has its own strengths and limitations. How, then, can we determine the most appropriate proxy for a given application? Here we propose a framework to support researchers in selecting a proxy, and encourage them to be explicit about the rationale for their choice. We argue that the choice of proxy should ultimately be purpose-driven, ensuring that the data align with the conceptualization of the city relevant to the application. City boundaries can be delineated using various proxy datasets, such as population and built-up area. The Perspective argues that the choice of proxy should align with the working conceptualization of the cities in question.