Reconciliation Through Integration: A Source-Level Bayesian Framework for Developing Measurement-Integrated Methane Inventories
5 affiliations
- 1Center for Energy and Environmental Systems Analysis (CEESA
- 2University of Texas at Austin
- 3Energy Emissions Modeling and Data Lab (EEMDL
- 4Department of Petroleum and Geosystems Engineering
- 5Williams Companies, Inc., One Williams Center, Tulsa, Oklahoma, 74172
Abstract
Bottom-up (BU) inventories and top-down (TD) aerial measurements at midstream natural gas facilities frequently diverge, and this discrepancy remains a central challenge for methane accounting in the sector. Previous approaches typically reconcile this divergence by treating BU and TD data as independent estimates of a single true quantity to be adjusted toward agreement, rather than as evidence that each capture only part of a more complex underlying system. This complexity arises from methane emissions being strongly governed by equipment operational state and therefore instantaneous aerial “snapshots” translate poorly to annual inventories. We argue instead that BU and TD are complementary sources of evidence that should be integrated. We present a source-level Bayesian framework that constructs BU prior distributions for each emission source category and updates them with attributed aerial detections through an empirical data-augmentation procedure. The resulting posterior distributions were propagated through unit-level activity data and source-specific information via Monte Carlo simulation to produce annual estimates. Across 64 compressor stations, measurement-integrated annual estimates were 84% higher than the BU inventory, a difference driven almost entirely by compressor vents (rod packing and seal vents), and leakage through blowdown and isolation valves, which together accounted for 90% of the asset-level revision. The revision was heterogeneous across stations and source categories, emphasizing the importance of source-level, station-specific integration when developing an emissions inventory.
Related
- Measurement-Informed Methane Inventories for Colorado’s Upstream Oil and Gas Sector: Two Independent Modeling Approaches
- Station-trained air-quality forecasts without regional emission inventories
- The R Package forestinventory: Design-Based Global and Small Area Estimations for Multiphase Forest Inventories
- Applying satellite observations to improve bottom-up national emission inventories for methane: application to Colombia
- Evaluating global emission inventories of biogenic bromocarbons
BibTeX
@misc{raji2026reconciliation,
title = {{Reconciliation Through Integration: A Source-Level Bayesian Framework for Developing Measurement-Integrated Methane Inventories}},
author = {Abdulahi Raji and Caitlin Barndt and Joshua Blaine and Scott Peters and Arvind P. Ravikumar},
howpublished = {ChemRxiv},
year = {2026},
month = sep,
doi = {10.26434/chemrxiv.15008933/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008933/v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Reconciliation Through Integration: A Source-Level Bayesian Framework for Developing Measurement-Integrated Methane Inventories
AU - Abdulahi Raji
AU - Caitlin Barndt
AU - Joshua Blaine
AU - Scott Peters
AU - Arvind P. Ravikumar
PY - 2026
DA - 2026/09/16
PB - ChemRxiv
DO - 10.26434/chemrxiv.15008933/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008933/v1
AB - Bottom-up (BU) inventories and top-down (TD) aerial measurements at midstream natural gas facilities frequently diverge, and this discrepancy remains a central challenge for methane accounting in the sector. Previous approaches typically reconcile this divergence by treating BU and TD data as independent estimates of a single true quantity to be adjusted toward agreement, rather than as evidence that each capture only part of a more complex underlying system. This complexity arises from methane emissions being strongly governed by equipment operational state and therefore instantaneous aerial “snapshots” translate poorly to annual inventories. We argue instead that BU and TD are complementary sources of evidence that should be integrated. We present a source-level Bayesian framework that constructs BU prior distributions for each emission source category and updates them with attributed aerial detections through an empirical data-augmentation procedure. The resulting posterior distributions were propagated through unit-level activity data and source-specific information via Monte Carlo simulation to produce annual estimates. Across 64 compressor stations, measurement-integrated annual estimates were 84% higher than the BU inventory, a difference driven almost entirely by compressor vents (rod packing and seal vents), and leakage through blowdown and isolation valves, which together accounted for 90% of the asset-level revision. The revision was heterogeneous across stations and source categories, emphasizing the importance of source-level, station-specific integration when developing an emissions inventory.
ER -