Universal Patterns in the Long-term Growth of Urban Infrastructure in U.S. Cities from 1900 to 2015

Abstract: Despite the rapid growth of cities in the past century, our quantitative, in-depth understanding of how cities grow remains limited due to a consistent lack of historical data. Thus, the scaling laws between a city's features and its population as they evolve over time, known as temporal city scaling, is under-explored, especially for time periods spanning multiple decades. In this paper, we leverage novel data sources such as the Historical Settlement Data Compilation for the U.S. (HISDAC-US), and analyze the temporal scaling laws of developed area, building indoor area, building footprint area, and road length and other road network statistics for nearly all metropolitan areas in the U.S. from 1900 to 2015. We find that scaling exponents vary dramatically between cities as a function of their size and location. Three notable patterns emerge. First, scaling law exponents imply many, but not all, metropolitan areas are becoming less dense and indoor area per capita increases as cities grow, in contrast to expectations. Second, larger cities tend to have a smaller scaling exponent than smaller cities. Third, scaling exponents (and growth patterns) are similar between nearby cities. These results show a long-term trend that could harm urban sustainability as previously dense populations are rapidly spreading out into undeveloped land. Moreover, the regional similarity of long-term urban growth patterns implies that city evolution and sustainability patterns are more interconnected than prior research has suggested. These results help urban planners and scientists understand universal, long-term patterns of city growth across the US.