A Data-Driven Model of Urban Carbon Emissions at the Human Scale

The measurement of localized greenhouse gas (GHG) emissions is a necessary step toward the goal
of citywide and global reduction targets (Gurney et al., 2015). Understanding local patterns of
GHG emissions provides a foundation for policymakers to identify significant sources of emissions,
target policies and interventions, and evaluate changes in emissions over time (Larsen and Hertwich,
2009). However, numerous hurdles remain to accurate, transparent, and consistent GHG emissions
monitoring at the scale of buildings, city blocks, and neighborhoods (Gately and Hutyra, 2017;
Lin et al., 2017; Mitchell et al., 2018). These include the lack of cross-city standards for energy
consumption data and transportation networks, data access and sharing, and models that estimate
localized emissions with reasonable degrees of confidence (Dhakal, 2010; Kates et al., 1998; Ibrahim
et al., 2012). Beyond this are questions of scale: What is local? How can cross-jurisdictional
emissions be accounted for? What is the appropriate geography to measure and monitor emissions
from different sources?

This work integrates numerous urban data sources, including public and private administrative
datasets, detailed building-level energy use data, vehicular traffic, and resident-generated data, to
develop a high resolution spatiotemporal model of urban carbon emissions. We combine data-driven
models with physical (engineering) models of energy use and carbon emissions to estimate total
city-wide carbon emissions from buildings and transportation networks. We examine localized
emissions, down to a 500m hexagonal grid and by hour, to analyze emissions patterns, neighborhood
disparities in emissions intensity, and opportunities to take action to reduce energy consumption
and emissions. Model outputs are visualized through a web-based, interactive dashboard that aims
to support city leaders and urban policymakers with an unprecedented, hyperlocal view of carbon
emissions to enable data-driven and evidenced-based climate action based on rigorous scientific
models.