Informing urban climate planning with high resolution data: the Hestia fossil fuel CO2 emissions for Baltimore, Maryland

Background Cities contribute more than 70% of global anthropogenic carbon dioxide (CO2) emissions and are leading the effort to reduce greenhouse gas (GHG) emissions through sustainable planning and development. However, urban greenhouse gas mitigation often relies on self-reported emissions estimates that may be incomplete and unverifiable via atmospheric monitoring of GHGs. We present the Hestia Scope 1 fossil fuel CO2 (FFCO2) emissions for the city of Baltimore, Maryland—a gridded annual and hourly emissions data product for 2010 through 2015 (Hestia-Baltimore v1.6). We also compare the Hestia-Baltimore emissions to overlapping Scope 1 FFCO2 emissions in Baltimore’s self-reported inventory for 2014. Results The Hestia-Baltimore emissions in 2014 totaled 1487.3 kt C (95% confidence interval of 1158.9–1944.9 kt C), with the largest emissions coming from onroad (34.2% of total city emissions), commercial (19.9%), residential (19.0%), and industrial (11.8%) sectors. Scope 1 electricity production and marine shipping were each generally less than 10% of the city’s total emissions. Baltimore’s self-reported Scope 1 FFCO2 emissions included onroad, natural gas consumption in buildings, and some electricity generating facilities within city limits. The self-reported Scope 1 FFCO2 total of 1182.6 kt C was similar to the sum of matching emission sectors and fuels in Hestia-Baltimore v1.6. However, 20.5% of Hestia-Baltimore’s emissions were in sectors and fuels that were not included in the self-reported inventory. Petroleum use in buildings were omitted and all Scope 1 emissions from industrial point sources, marine shipping, nonroad vehicles, rail, and aircraft were categorically excluded. Conclusions The omission of petroleum combustion in buildings and categorical exclusions of several sectors resulted in an underestimate of total Scope 1 FFCO2 emissions in Baltimore’s self-reported inventory. Accurate Scope 1 FFCO2 emissions, along with Scope 2 and 3 emissions, are needed to inform effective urban policymaking for system-wide GHG mitigation. We emphasize the need for comprehensive Scope 1 emissions estimates for emissions verification and measuring progress towards Scope 1 GHG mitigation goals using atmospheric monitoring.


Contents
Text S1: Detailed methods for the spatial distribution of nonpoint building emissions to building and parcel footprints Text S2: Detailed methods for the temporal distribution of onroad emissions to individual road segments Figure S1: Annual average temporal distribution for particular days of the week as derived from Baltimore Metropolitan Council traffic data.

References
Text S1. Nonpoint buildings FFCO2 emissions for residential, commercial, and industrial nonpoint buildings represent on-site combustion processes only and not associated with large emissions allocated to a stack (which are categorized as point emissions). Emissions associated with the consumption of electricity in buildings are spatially allocated to electricity generating units (i.e. power plants) and are not included in this sector. Annual emissions for the city followed Vulcan v3.0 methodology (1). The hourly temporal profiles followed the same methods as well, but hourly emissions were assigned to individual parcels and buildings instead of US Census block groups. Local data are used to distribute the city-level FFCO2 emissions from Vulcan v3.0 in space. Parcel and building data for the city of Baltimore were obtained from the Maryland Department of Planning (2). Attributes for parcels and buildings include the land-use sector (residential, commercial, industrial, and exempt), building style, building description, structure floor area, and year built. Some records include non-building descriptions (e.g. "AUTO parking lot"). Additional building data from the Baltimore City Government (3) were used to gap fill missing data -notably in the structure floor area and year built columns -though this dataset only extended to the year 2007, while the state-wide dataset extended to 2015. This city-specific dataset contained complete records for structure floor area, but was otherwise less descriptive. The combination of the state and city data were used for structure floor areas.
As building design and codes have changed over time, buildings have generally become more energy efficient in recent years (4). Therefore, building stock age plays a role in the spatial distribution of nonpoint emissions. The "year built" data in the parcel and building datasets from the state of Maryland and the city of Baltimore represent the date that the primary structure was built. For missing data, the sector-specific mean was used to estimate the year built (residential: 1949, commercial: 1951, industrial: 1968).
No data were available regarding retrofitting of older buildings to more recent building codes -hence, the year a building was built was used to define its energy use intensity.
Energy use in buildings was based on data from the EIA's Commercial Building Energy Consumption Survey (CBECS), Residential Energy Consumption Survey (RECS), and Manufacturing Energy Consumption Survey (MECS). The "Microdata" building survey data contain building characteristics for multiple survey years, with data closest to the core year of 2011 being used in Hestia. These survey data contain information used to calculate the non-electric energy usage intensity (NE-EUI) for building prototypes, specific to fuel type (natural gas and liquid "petroleum" fuels) and US Census Regions. Baltimore falls under the South Atlantic (Division 5). Note that the calculated NE-EUIs used in Hestia represent averages in the Census Region for fuel types, and that the actual NE-EUI will vary for individual buildings due to (e.g.) building design, decay, retrofitting, microclimate, and use by occupants. No data exist on the natural gas pipeline system in Baltimore at the building scale, nor the distribution of liquid fuels to individual buildings. Therefore, natural gas and petroleum usage is assumed to be distributed to all nonpoint buildings in Baltimore based on their NE-EUI.
Unlike other Hestia cities -Indianapolis (5), Salt Lake City (6), and Los Angeles (7)the building classification scheme of CBECS and RECS were used instead of a Hestia-specific building scheme. The CBECS and RECS building classifications are also matched to the building types in the eQUEST model, which was used for temporal emissions allocation. Table S1 shows a crosswalk between CBECS/RECS building types and state/city parcel and building data. Tables S2 and S3 show crosswalks between the eQUEST building types and the CBECS and RECS building types, respectively. Furthermore, the age of buildings in Baltimore were categorized into building vintages within the CBECS, RECS, and MECS data.
However, due to the sample sizes, building vintages in Hestia were represented by broad age ranges. For commercial and residential buildings, NE-EUIs were calculated for two vintages -pre-1980 and post-1970. No vintages were assigned to industrial buildings due to a lack of data.
The 2012 CBECS microdata were used for the commercial sector. The data contain the building floor area and energy consumption (natural gas and fuel oil/diesel/kerosene -collectively referred to as "petroleum" in Hestia), along with a weighting factor that converts individual building sampling into a population total. For each Census division, building type, and vintage, the NE-EUI for natural gas (NEEUI-NG) is achieved by summing the weighted natural gas use for each building and dividing by the sum of the weighted floor areas for each building: where i is the Census division, j is the building type, k is the vintage, m is the record in the dataset (within Census division i, building type j, and vintage k), Nm is the natural gas consumption for each building (kBTU), Am is the floor area of each building (ft 2 ), and wm is the weighting factor that converts individual sampling to population totals.
The same procedure is followed for fuel oil/diesel/kerosene consumption (NEEUIFK). A similar procedure is followed for residential buildings. In the RECS data, the 2009 survey data were used. An equation identical to Eq. 1 is used for the same census division and the same fuel categorization (natural gas and "petroleum") and the same vintages.
The MECS data for 2010 were used for industrial buildings. The MECS data represent the sum of sampled buildings specific to Census division and manufacturing sector, represented by North American Industry Classification System (NAICS) codes, as opposed to individually sampled buildings within building types for CBECS and RECS. Thus, the fuel-specific (natural gas and the sum of residual fuel oil and distillate fuel oil) were calculated as follows: where i is the Census division, j is the manufacturing sector, N is the natural gas consumption, and A is the floor space. The NE-EUI for "petroleum" fuels is calculated in the same manor.
Each parcel or building is assigned an annual FFCO2 emission, E(b), based on the NE-EUI, the structure floor area, and the county-wide emissions: where b represents each individual structure, f represents the fuel type (natural gas or "petroleum"), s represents the building sector, NEEUI represents the fuel-and buildingtype-specific NE-EUI, A represents the structure floor area, and Etot represents the total county FFCO2 emissions. For coal, however, the NE-EUI is assumed to be 1 for all buildings since no building survey data on coal consumption exist. Thus, FFCO2 from coal consumption is proportional to floor area.  figure S1 shows the hourly profiles for Friday through Monday. Figure S1. Annual average temporal distribution for particular days of the week as derived from Baltimore Metropolitan Council traffic data.