Low-Computation Strategies for Extracting CO2 Emission Trends from Surface-Level Mixing Ratio Observations

Global momentum is building for drastic, regulated reductions in greenhouse gas emissions over the coming decade. With this increasing regulation comes a clear need for increasingly sophisticated monitoring, reporting, and verification (MRV) strategies capable of enforcing and optimizing emissions-related policy, particularly as it applies to urban areas. Remote sensing and/or activity-based emission inventories can offer MRV insights for entire sectors or regions, but are not yet sophisticated enough to resolve unexpected trends in specific emitters. Urban surface monitors can offer the desired proximity to individual greenhouse gas sources, but due to the densely-packed nature of typical urban landscapes, surface observations are rarely representative of a single source. Most previous efforts to decompose these complex signals into their contributing emission processes have involved inverse atmospheric modeling techniques, which are computationally intensive and believed to depend heavily on poorly understood a priori estimates of error covariance. Here we present a number of transparent, low-computation approaches for extracting source-specific emissions estimates from signals with a variety of nearfield influences. Using observations from the first several years of the BErkeley Atmospheric CO2 Observation Network (BEACO2N), we demonstrate how to exploit strategic pairings of monitoring "nodes," anomalous wind conditions, and well-understood temporal variations to hone in on specific CO2 sources of interest. When evaluated against conventional, activity-based bottom-up emission inventories, these strategies are seen to generate quantitatively rigorous emission estimates. With continued application as the BEACO2N data set grows in time and space, these approaches offer a promising avenue for optimizing greenhouse gas mitigation strategies into the future.