5 Methods and data sources
5.1 Methods
The general workflow for quantifying electricity emissions is to identify all of the electric utilities that operate within our study area, collect any reporting they provide to the states of Minnesota and Wisconsin about the amount of energy delivered to all their customers (with reference to federal reporting sources where state-level reporting gaps exist), and apply EPA-provided emissions factors to the reported activity/energy deliveries to calculate estimated emissions. Methodologies for allocating utility activity reports to counties varies across MN and WI and are further described in the following section. Most inputs we use in the construction of our electricity emissions data set are of the highest quality rank (Table B.2), as they are either direct government-created data (e.g., emissions factors) or data reported to state/federal authorities (e.g., regulatory filings). However, for two Minnesota electric utilities – Elk River Municipal Utilities (Elk River Municipal Utilities 2022) and New Prague Utilities Commission (New Prague Utilities Commission 2022) – where regulatory filing data could not be sourced to quantify electricity deliveries, we referred to financial reporting documents published by the utilities.
Total regional emissions (Emissionsr) represents the sum of all recorded energy deliveries by utility i within county j, where i refers to each of the electric utilities operating across our region, and j refers to the eleven counties included in this inventory. Our regional total therefore represents an aggregation of electricity deliveries for all distinct utility-county records.
\[Emissions_r = \Sigma (mWhDelivered_iCounty_j \times {Emissions Factor}) \]
Our inventory takes a “demand-side” approach to emissions quantification and seeks to aggregate all reported delivery of energy to ALL customers served by utilities (meaning all customer types, inclusive of residential, commercial, industrial, and government accounts). This means that energy loss and use experienced by utilities in the process of energy generation and transmission, and delivery and resale to utilities operating outside of our study area, are not directly reflected in the numbers attributed to counties. The U.S. Energy Information Administration (EIA) estimates that annual electricity transmission and distribution (T&D) losses averaged about 5% of the electricity transmitted and distributed in the United States in 2018 through 2022. (Administration 2023)
While our primary data collection does not include a breakout of electricity deliveries by sector, we do leverage year 2021 NREL SLOPE forecasts of electricity consumption (built from a base year of 2016 observed data) by sector (residential, commercial, industrial) at the county level to calculate modeled proportions of consumption by sector, which we then apply to our aggregate numbers to calculate estimated emissions by sector NREL (2017).
5.1.1 Identifying utilities in scope
To identify the electric utilities that operate within our 11-county study area, we referred to maps and geospatial datasets capturing utility service areas in Minnesota and Wisconsin. To identify Wisconsin electric utilities, we downloaded the Electric Service Territory map maintained by the Wisconsin Public Service Commission (Wisconsin Public Service Commission and Tomaszewski 2024). To identify Minnesota electric utilities, we downloaded the Electric Service Territory map maintained by the Minnesota Public Utilities Commission and the Minnesota IT Geospatial Information Office(Office 2023).
5.1.2 Collecting and aggregating activity data from utilities
County-level activity data
After identifying which utilities operate within our study area within each state, we collect reporting submitted by these utilities to the relevant state and federal authorities and use a variety of approaches, depending on data availability, to allocate utility activity/energy deliveries to specific counties. For the state of Minnesota, we collected reports provided by all in-scope utilities for years 2014-2023. For year 2021, additional manual data collection efforts were completed to guarantee a complete data set, since reports for all in-scope utilities were unfortunately not available for all years. As needed, data for other county-years could be finalized in a similar fashion.
Minnesota
All electric utilities authorized to do business in Minnesota are required to file an annual data report pursuant to MN Rules Chapter 7610. The Minnesota Public Utilities Commission makes these reports searchable through an eFiling Site, and downloadable as Excel workbooks (Commerce 2005). For each utility identified in distinct_electricity_util_type_MN.RDS (a data product of minnesota_electricUtilities.R, a script that looks for intersections between electric utility service areas and our Minnesota counties), we downloaded the relevant 2021 annual reports from this site (see note about Great River Energy in the previous section for caveats), except for North Branch Municipal Water and Light, which did not submit a 2021 report (we used their 2022 report as a substitution). Elk River Municipal Utilities and New Prague Utilities Commission did not file reports for 2021 or 2022, and so we used financial reports to identify their total electricity delivered; both utilities operated within only one county in our study area, which meant no estimation/allocation was necessary.
We wrote code to extract the county-level data reported in the report section titled “ITS DELIVERIES TO ULTIMATE CONSUMERS BY COUNTY FOR THE LAST CALENDAR YEAR” on the relevant annual data report Excel workbooks compiled into a folder directory, and created a table with three columns: county, utility, and mWh_delivered (megawatt-hours). By compiling this data for all utilities found to operate within our study area, aggregating all electricity deliveries at the county level becomes possible.
Wisconsin
All municipal and investor-owned utilities authorized to do business in Wisconsin are required to file an annual report with financial and operational information pursuant to Wis. Stat. § 196.07. The Public Services Commission of Wisconsin makes these reports searchable through an E-Services Portal, and downloadable as either PDFs or Excel workbooks, with options to export only specific portions of the reports as spreadsheets (Public Service Commission of Wisconsin 2022). For each utility identified in distinct_electricity_util_type_WI.RDS (a data product of wisconsin_electricUtilities.R, a script that looks for intersections between electric utility service areas and our Wisconsin counties), we downloaded the relevant 2021 annual reports from this site.
A similar process was followed for the four Wisconsin cooperative utilities for which we referenced federal regulatory filings data to populate our dataset of electricity deliveries.
Because of the small amount of data, and the different data structures of the reports for municipally-, cooperatively-, and investor-owned utilities in Wisconsin, we hard-coded observed information into data frames rather than extracting data through a web scraper or document analyzer (see processed_wi_electricUtil_activityData.R).
City-level activity data (seven-county metro in MN only)
No regulatory requirements in Minnesota or Wisconsin directly require utilities make city-level reports on electrical or natural gas usage data activity publicly available in a manner similar MN Rules Chapter 7610’s requirement for public county-level reporting. In alignment with climate planning statute, the Met Council is leading an effort to collect this information on behalf of cities and townships in the seven-county region as a constituent part of the greenhouse gas inventories being developed by the Council to support communities in their own local climate planning as required by state statute and Imagine 2050 minimum requirements. This data collection effort entails 1) direct processing of real, city-sector level data provided by utilities in our region, as well as 2) a series of modeling efforts geared towards addressing missingness in our city/township dataset, as well as disaggregation of coarser numbers provided by utilities (e.g., combined data for commercial/industrial sectors) into residential, commercial, and industrial numbers.
Processing utility-provided data
Xcel, the largest electric utility in our region, publishes Community Energy Reports (Energy, n.d.) that document energy deliveries to specific communities in their service area. We gathered this data for years 2015-2022 and compiled it into the final .RDS file xcel_activityData_NREL_2015_2022_process.RDS.
minnesota_xcelCommunityReports_electricity.R includes all data processing steps necessary to process the nearly 1000 files (one for each city-year combination) made available on the Community Energy Reports site. The function get_files identifies and extracts relevant info (e.g., city name) from all data files, while process_file reads detailed electricity consumption data from sections titled “Standard Community Report” in each Excel workbook. Key metrics, such as kilowatt-hours delivered (kWh_delivered) and utility-reported carbon emissions (util_reported_co2e), are then transformed into aggregated datasets (Xcel_activityData_2015_2023) at the utility-year-county level.
Ambiguities in sector classifications, particularly the “Business” category, are addressed through proportional disaggregation using NREL-modeled emissions proportions. This process incorporates additional datasets, such as nrel_slope_city_emission_proportions.RDS, to allocate energy and emissions between commercial and industrial sectors. Geographic integration (i.e., disaggregating CTUs with footprint in multiple counties) is achieved using cprg_ctu and ctu_population, enabling population-based allocation of activity/utility-reported emissions to constituent COCTU units.
Final outputs of minnesota_xcelCommunityReports_electricity.R include clean, joined datasets (xcel_activityData_NREL_2015_2022_process.RDS) and visualizations that compare utility-reported sector proportions to modeled values. More to say here before modeling missingness section?
Modeling missingness
Content to come
Emissions factors
To transform electricity deliveries (recorded in mWh) into emissions in metric tons CO2e, we referenced Emissions & Generation Resource Integrated Database (eGRID) summary data for the MROW subregion (Midwest Reliability Organization - West) (USEPA 2021).
This dataset provides estimates in lbs/mWh for CO2, CH4, N2O, and CO2e emissions based on MROW’s sub-regional electricity generation grid mix; converting this factor to metric tons per mWh and multiplying the county-level mWh estimates yields an estimate for CO2e. To generate this sub-regional estimate, eGRID first calculates estimated emissions at the plant level, and assigns plants to regions. By using an emissions factor that is localized, our inventory accounts for the specific grid mix in our study area (see the grid mix linked to the eGRID MROW emissions factor used in this inventory below). Per the eGRID technical guide, “the subregions were defined to limit the import and export of electricity in order to establish an aggregated area where the determined emission rates most accurately matched the generation and emissions from the plants within that subregion.”
5.2 Utility Activity Data
5.2.1 Minnesota
Under Minnesota Administrative Rules Chapter 7610 (Commerce 2005), utilities are required to file an annual data report that supports the identification of “emerging energy trends based on supply and demand, conservation and public health and safety factors, and to determine the level of statewide and service area needs” (Minnesota Department of Commerce 2022). This includes a report of county-level energy deliveries (reported in megawatt-hours, commonly written as mWh). Because the information is structured in this manner, electricity emissions at the county-level can be estimated as a direct function of energy deliveries to counties reported by utilities.
One utility operating within our study area, Great River Energy (GRE), is a non-profit wholesale electric power cooperative which provides wholesale power and delivery services to 28 Minnesota-based electric cooperatives, which collectively own GRE and provide retail electric service. They are the primary supplier of energy (>99%) to the cooperative utilities operating in our study area. Though most electricity suppliers having relationships with Minnesota electric utilities do not file these annual data reports, GRE does, given their unique hybrid wholesale/transmission structure and cooperative ownership model. As a result, while we keep only GRE’s reporting in our data collection for county-level electricity deliveries, we exclude reporting from these subsidiary retail electric cooperatives, in order to avoid double counting electric deliveries. We use GRE’s reported deliveries to our 9 Minnesota study area counties as a full substitution for the deliveries of these utilities, which may represent a marginal undercount given that marginal-to-negligible amounts of energy delivered by these retail cooperatives came from other other sources, per their annual reports.
5.2.2 Wisconsin
Under Wis. Stat. § 196.07, investor- and municipally-owned electric utilities operating within the state of Wisconsin must submit annual reports to the State which include an array of information related to utility finance and operations, including key figures leveraged in our data collection, such as total energy deliveries made (in units of kWh) and total number of customer accounts within each county (Wisconsin State Legislature 2024).
Of the seven in-scope electric utilities, only three (the investor- and municipally-owned utilities) were required to make these reports to the State in 2021 (four of the in-scope electric utilities are cooperative utilities); state data was leveraged in this case. For the four utilities (all cooperative utilities) not making these reports, we relied upon the detailed data files provided by the EIA, recording the responses of utilities to the Annual Electric Power Energy Report (Form EIA-861), which records retail sales by utilities and power marketers (U.S. Energy Information Administration 2023). Two utilities (Dunn Energy Cooperative and Polk-Burnett Electric Cooperative) filled the long version of the form (filled by larger entities), and two (St. Croix Electric Cooperative and Pierce-Pepin Electric Cooperative Services) filled the short form (filled by smaller entities). For our purposes, both the long and short form provided suitable activity information (total energy delivered, total customers) to allocate energy deliveries to counties in concert with Census population data, in the process outlined below.
Because Wisconsin utilities do not report energy deliveries at the county level, it was necessary to estimate energy deliveries by a given utility i within a particular county j. For those three utilities who reported county-level customer counts and total customer counts to the state, we estimated mWh delivered by utility i in county j by multiplying their total statewide energy deliveries (as reported to the relevant state authorities) by the proportion of their customers residing in each of our two study area counties.
\[mWhDelivered_iCounty_j = (mWhEnergyDeliveries_i \times {ProportionOfTotalUtilityCustomers_j}) \]
Note: This approach implicitly assumes that customer accounts across counties within the operations of a given utility have the same average per-account demand for energy, when this is influenced by land-use mix and relative magnitude/scale of residential and commercial/industrial utility accounts within a given county.
To calculate the estimated energy delivered by utility i in county j for the four cooperatively-owned utilities that did not report county-level customer counts (i.e., did not report to the State), we used population figures to estimate/allocate reported electricity deliveries to counties. We took the actual total energy delivered by utility i across Wisconsin (as reported to the relevant federal authorities) and multiplied this by the proportion of total population within each utility’s entire service area residing within county j at the 2020 decennial Census.
\[mWhDelivered_iCounty_j = (mWhDelivered_i \times {ProportionOfTotalUtilityPopulation_j}) \]
The factor ProportionOfTotalUtilityPopulationj was calculated by spatially joining Census block centroids containing population data to 1) polygons representing the entirety of our in-scope utilities’ service areas (including areas outside of St. Croix and Pierce counties) and 2) polygons representing only the portions of these utilities’ service areas within a) St. Croix county and b) Pierce County. These utility-county service areas are calculated separately, to facilitate an informed allocation of statewide energy use to each county in turn.
Comparison with federal inventories
NREL SLOPE
The NREL SLOPE (State and Local Planning for Energy) Platform provides yearly forecasted emissions tied to the user of electricity up to 2050 based on 2016 reported data at the county level. In comparing these figures to our own inventory, we observed that, where we estimated 0 metric tons of emissions linked to electricity deliveries in our study area in the year 2021, NREL SLOPE forecasted 355,545,687metrics tons in our study area.