Previous research using a diverse set of methods focused largely on large metropolitan regions or cities has shown that household carbon footprints (HCFs) vary considerably, with energy, transportation, or consumption comprising a larger share of the total and with households in some locations contributing far more emissions than others. For example, motor vehicles in California account for 30% of HCF, compared to 6% for household electricity, while electricity is frequently the largest single source of emissions in locations with predominantly coal-fired electricity. Income, household size, and social factors have been shown to affect total HCF, while a large number of factors have been shown to contribute to household energy and transportation-related emissions.

A number of studies suggest that geographic differences in emissions are in part explained by population density. Population-dense municipalities tend to be urban centers with employment, housing, and services closely colocated, reducing travel distances, increasing demand for public transit, and with less space for larger homes. … These earlier studies have been limited to analyzing a small set of case studies, and the resulting conclusions are difficult to generalize beyond those included in the studies themselves. A large, nationwide data set of all locations at fine geographic resolution holds potential to reassess the urban form hypothesis to more accurately describe the relationship between population, policy, urban form, and emissions. Our primary research questions are (1) how much variability exists in the size and composition of household carbon footprints across all U.S. locations and (2) how much of this variability can be explained by population density, income, home size, and other factors contributing to carbon footprints in urban, suburban, and rural areas?

—Jones and Kammen

Using the nationwide data set, they found a “more nuanced relationship” between population density and household carbon footprints than determined by previous research using much smaller data sets. Those studies suggested a negative correlation between population density and emissions—i.e., as population density increases, emissions decrease.

In contrast, Jones and Kammen found that the mean, standard deviation and range of emissions increase until a population density of about 3,000 persons per square mile is reached, after which mean HCF declines logarithmically, leveling out at a lower limit of about 30 tCO₂ per household (35% below average) at densities over 50,000 persons per square mile.

The net effect of this “inverted-U” relationship is that there is no overall correlation between population density and HCF when considering all US zip codes and cities. However, there is a strong negative log−linear correlation between population density and HCF if only considering the most populous cities—findings consistent with previous studies. However, when considering entire metropolitan statistical areas, the inverted-U relationship disappears and the correlation appears to be slightly positive.

• More populous metropolitan areas tend to have somewhat higher net HCF due to the influence of more extensive suburbs, which are on average 25% higher than urban cores.

• The two largest metropolitan areas, New York and Los Angeles, are exceptions with somewhat lower net carbon footprints, suggesting the inverted-U relationship may hold when including extremely population-dense metropolitan areas, or megacities.

Metropolitan areas look like carbon footprint hurricanes, with dark green, low-carbon urban cores surrounded by red, high-carbon suburbs. Unfortunately, while the most populous metropolitan areas tend to have the lowest carbon footprint centers, they also tend to have the most extensive high-carbon footprint suburbs.

—Christopher Jones

The UC Berkeley researchers found that the primary drivers of carbon footprints are household income, vehicle ownership and home size, all of which are considerably higher in suburbs. Other important factors include population density, the carbon intensity of electricity production, energy prices and weather.

As a policy measure to reduce GHG emissions, increasing population density appears to have severe limitations and unexpected trade-offs. In suburbs, we find more population-dense suburbs actually have noticeably higher HCF, largely because of income effects. Population density does correlate with lower HCF when controlling for income and household size; however, in practice population density measures may have little control over income of residents. Increasing rents would also likely further contribute to pressures to suburbanize the suburbs, leading to a possible net increase in emissions. As a policy measure for urban cores, any such strategy should consider the larger impact on surrounding areas, not just the residents of population dense communities themselves. The relationship is also log−linear, with a 10-fold increase in population density yielding only a 25% decrease in HCF. Generally, we find no evidence for net GHG benefits of population density in urban cores or suburbs when considering effects on entire metropolitan areas.

Given these limitations of urban planning our data suggest that an entirely new approach of highly tailored, community-scale carbon management is urgently needed. Regions with high energy-related emissions, such as the Midwest, the South, and parts of the Northeast, should focus more on reducing household energy consumption than regions with relatively clean sources of energy, such as California. However, if household energy were the sole focus of residential GHG mitigation programs, then between two-thirds and 85% of household carbon footprints would be left unaddressed in most locations; the full carbon footprint of households should be considered in community GHG inventories and management plans. Suburbs, which account for 50% of total U.S. HCF, tend to have high motor vehicle emissions, large homes, and high incomes. These locations are ideal candidates for a combination of energy efficient technologies, including whole home energy upgrades and solar photovoltaic systems combined with electric vehicles. Food tends to be a much larger share of emissions in urban cores, where transportation and energy emissions tend to be lower, and in rural areas, where household size tends to be higher and consumption relatively low.

—Jones and Kammen

The project website includes a tool that calculates carbon footprints for essentially every populated US zipcode, city, county and U.S. state (31,531 zipcodes, 10,093 cities and towns, 3,124 counties, 276 metropolitan regions and 50 states) as well as an interactive online map allowing users to zoom in and out of different locations. Households and cities can calculate their own carbon footprints to see how they compare to their neighbors and create customized climate action plan from over 40 mitigation options.

The research was funded by the National Science Foundation and the California Air Resources Board.

Resources

• Christopher Jones and Daniel M. Kammen (2014) “Spatial Distribution of U.S. Household Carbon Footprints Reveals Suburbanization Undermines Greenhouse Gas Benefits of Urban Population Density,” Environmental Science & Technology doi: 10.1021/es4034364