https://doi.org/10.5061/dryad.8kprr4z0p

This COBE_Anonymized_Emissions_v4.csv file contains anonymized emissions measurements and data on facilities scanned with no emissions, collected by aerial vendors across multiple campaigns for the Colorado Ongoing Basins Emission (COBE) study. The data has been processed to ensure confidentiality for site operators, while retaining as much metadata as possible necessary for analysis and modeling replications. Facility names and locations have been removed and replaced with an anonymized facility identifier. Each facility's identifier is consistent throughout the dataset to show instances when multiple emissions came from the same facility.

Emissions data: For Bridger, emissions are reported at the source level. If multiple emissions are detected from the same source within a single day, they are averaged to generate a single source-level emission rate for that day. A single facility may have multiple emission sources, and each source is treated as a separate emission event in the dataset. Multiple identical emission rates recorded at a facility suggest the sensor could not clearly distinguish the contributions from individual sources and instead divided the aggregate plume among the identified sources.

Non-detects: This data is at the facility-level and reports when a facility was scanned but no emissions were detected. If there was a non-detect and then later the same facility was scanned and there was an emission, the Anonymized_Facility_ID will remain the same and the Facility Scan Number column would show 2 passes in this case.

Note: A new column was added on 3/12/26 to distinguish first scans from rescans. This dataset does not contain any information on when a facility was scanned and there were no emissions detected. 

Maintenance emissions (which are treated differently when developing a predictive model) are included, while emissions that are determined to be from non-production sites are not. Measurements that do not align with reported oil/gas facilities are also excluded from this file.

The COBE report is published here: https://hdl.handle.net/10217/242813

Important fields:

• Anonymized_Facility_ID: 8-character identifier generated using a hash function
• Season: Flight campaigns are grouped by season of Summer, Spring, Fall, Winter, or blank. Blank values indicate that the data is taken from another study with different flight timing than those done specifically for COBE.
• Basin: Geographic basin in which the measurement was taken. For the COBE study, oil/gas basins were grouped into three categories, Denver-Julesberg (DJ) basin, Piceance basin, and 'Other', which includes everywhere else in Colorado.
• Equipment type: For Bridger, this reflects the equipment type reported in their dataset. For GHGSat and Insight M, the equipment type is listed as unknown unless it was specified by the operator.
• Emission rate (kg/h): Observed emission rate by aerial measurement, rounded to two decimal points. If no emissions were detected but the facility was scanned the emission rate would = 0.
• Emission rate (upper & lower) 95CI: 95% confidence interval bounds, both upper and lower.
• PS: Prototypical Site, classification assigned to facilities of different configurations (see COBE report main text)
• Aircraft company: Vendor of aerial instrument product used to record measurement, either Bridger, GHGSat, or Insight M.
• Sensor: Specific sensor product/technology used to detect emission.
• Partner? (Y/N): Indicates whether facility operators participated in the COBE study. When 'Yes', some insight into the cause analysis process should be available.
• Type of Emission: Emissions are classified into four categories for the modeling in this study: Normal Operations, Maintenance, and Fugitive/leak, and emissions with unclear or not classified causes are denoted 'Other' or 'Unknown'.
• Cause: Where information from participating operators is available, this column provides the likely determined cause of emission.
• Facility Scan Number: An integer indicating the scan sequence at a facility per aircraft company, when emissions were detected. A value of 1 denotes the first scan where emissions were detected; subsequent scans on different calendar days receive incrementing numbers (2, 3, etc.). Example: A facility scanned on 3/10 with detected emissions receives Facility Scan Number = 1. If rescanned on 3/11 with detected emissions, it receives Facility Scan Number = 2.

This dataset is published by the METEC (Methane Emissions Technology Evaluation Center) research group as part of the COBE (Colorado Ongoing Basins Emissions) study.
For questions or additional information about this dataset, please contact PI Anna Hodshire; Anna.Hodshire@colostate.edu