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

Keith Burghardt, Johannes H.Uhl, Kristina Lerman, Stefan Leyk · 2024

Read Universal Patterns in the Long-term Growth of Urban Infrastructure in U.S. Cities from 1900 to 2015 by Keith Burghardt, Johannes H.Uhl, Kristina Lerman, Stefan Leyk (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-13181.txt.

Introduction

Keith Burghardt and co-authors track how United States city footprints grew against population from nineteen hundred to two thousand fifteen, in a paper from two thousand twenty-four. A fresh history data set covers near all metro areas. Built area, indoor area, footprint area, and road length were scaled against city size over time.

Growth rules swung widely by city size and place. Many metros grew less dense with more indoor space per person, against old hopes. Larger cities tended to show smaller growth exponents than smaller ones. Nearby cities shared like growth shapes, so sprawl looks regional. Readers get a century-scale warning for land use planning.

Catalog record

CREATORKeith Burghardt, Johannes H.Uhl, Kristina Lerman, Stefan Leyk
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE41 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-13181

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Universal Patterns in the Long-term Growth of Urban Infrastructure in U.S. Cities from 1900 to 2015 Keith Burghardt, Johannes H.Uhl, Kristina Lerman, Stefan Leyk 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. Journal: npj Urban Sustainability 1 Introduction Cities appear to follow common patterns that suggest simple mechanisms underly the complexity of urban growth ( 52 ; 25 ; 31 ; 42 ; 43 ; 27 ; 6 ; 41 ). For example, city properties vary consistently with their size, a phenomenon known as city scaling ( 8 ; 7 ; 19 ). While these analyses are usually conducted on cross-sectional data (i.e., across all cities at one point in time), recent works have explored longitudinal data (i.e., describing a single city across different points in time) to study temporal scaling laws of growing cities ( 6 ; 39 ; 9 ; 32 ; 21 ). This seemingly subtle difference in analytical design reveals unexpectedly non-universal features, with scaling laws that deviate from traditional theory ( 7 ) and vary substantially between cities ( 9 ). In this paper, we explore these issues using vast, integrated geospatial datasets to map urban infrastructure in over 850 metropolitan areas in the conterminous US (CONUS) at fine spatial resolution and over a broad time span. These datasets include the Historical Settlement Compilation for the U.S. ( 36 ; 1 ), Microsoft building footprint data ( 40 ), and road network data from the National Transportation Dataset ( 51 ; 19 ), illustrated in Fig. 1 . In contrast to previous analysis ( 44 ; 45 ; 34 ; 8 ), we are able to study temporal scaling over more than a century (1900-2015), constituting an observational window that goes beyond recent decades to which most geospatial data are constrained. Specifically, we analyze developed area, which encompasses the total built up land of a city, including their buildings, other built infrastructure and impervious surfaces, building indoor area, i.e., the total indoor residential area, commercial and industrial space, building footprint area, the total length of the urban road network, and a number of other road network statistics. Figure 1: Data collected to analyze temporal scaling. (a) The growth of building indoor area over time for Atlanta, Georgia, and San Francisco, California. (b) Roads built at different ages (based on the ages of nearby houses). (c) Building footprints in different periods. Data from ( 36 ; 19 ). Our results imply that, not only are smaller cities less dense than larger cities ( 20 ), they are getting less dense more quickly over the course of urban expansion, which invades nearby undeveloped areas. Moreover, growing cities create ever larger buildings per person, which requires increasing energy for heating and cooling; this is an even greater problem in smaller cities. Finally, our analysis reveals strong correlations between the scaling laws of nearby cities which decrease very slowly with distance, meaning the urban development patterns of nearby cities are similar. Overall, these results offer new insights into the seemingly non-universal temporal patterns of city growth, with important implications regarding sustainability concerns in urban planning. 2 Results City statistic Description Developed area Total developed area in a city ( 20 ) Indoor area Total indoor area in a city in square meters ( 20 ) Building footprint area Total area of building footprints in square meters ( 20 ) Road length Total km of road built in kilometers ( 20 ) Number of deadends Number of cul-de-sacs and other ends of roads ( 19 ) Number of 4+ intersections Number of intersections where four or more edges (roads) meet ( 19 ) Number of intersections Number of road intersections of all forms ( 19 ) Number of edges Number of unique edges (defined to start at one intersection and end at another) ( 19 ) Population Population in urban regions of a CBSA ( 19 ) Table 1: Urban statistics and population used to create temporal scaling. 2.1 Temporal scaling In contrast to cross-sectional analysis, which measures scaling relationships at a particular time across all cities, regardless of their age, temporal analysis tracks individual cities as they grow. One challenge here is defining city boundaries over time, because urban patches eventually coalesce as cities grow, making patch-level evolution difficult to assess. Because the largest one or two patches typically dominate the statistics (see SI Figure S1), the definitions of city boundaries are less important for temporal analysis as the statistics ultimately tend to describe the largest city. Here instead, we define cities by their 2010 CBSA boundaries. We specifically analyze developed area, indoor area, building footprint area, road length, as well as the number of deadends, the number of intersections with four or more edges, the total number of intersections, and the total number of edges versus population (see definitions in Table 1 ). All results are metric (area in meters squared or length in kilometers) or counts. We see very different scaling relations using temporal analysis. Figure 2 shows network statistics as a function of growing population for a set of cities chosen at random from the data. We find scaling relations are often (but not always) superlinear for developed area, indoor area, building footprint area, and road length, with these statistics increasing faster than the population (linear scaling shown as a dashed line). These results imply that each new resident can be linked to increases in per-capita developed area, indoor area, building footprint area, and road length. In practice, this means that growing cities become less dense, and houses are getting larger. In contrast, the number of deadends, the number of road intersections with four or more edges, the total number of road intersections, and total number of edges often scales sub-linearly. These results suggest that

Read the complete text