Ronald Carl Cohen

Addressing urban air pollution requires identifying, quantifying, and mitigating emission sources, yet bottom-up inventories often lack fine spatiotemporal precision and rarely capture unpermitted or informally operated sources. We combine dense mobile and fixed-site measurements in a Bayesian inverse modeling framework to produce hourly, hyperlocal black carbon (BC) emission maps (150 meters by 150 meters) that reveal how and where key sources are underrepresented. In West Oakland, CA—an overburdened community near major freight infrastructure—our approach identifies underestimated and unaccounted emission sources, increasing estimated BC emissions by ∼33%. These corrections quadruple the contribution of neighborhood and port-related sources to population-weighted exposures (from 5 to 20%), disproportionately affecting low-income neighborhoods adjacent to freight corridors. Our results demonstrate that combining dense observational data with inverse modeling enables detection and quantification of overlooked emissions, supporting targeted intervention. Sensitivity analyses indicate similar performance with a substantially reduced fixed-site network (∼10 monitors; ∼1 site per square kilometer, compared with 6.7 sites per square kilometer in the full campaign), paired with time-averaged mobile maps, supporting applicability in data-limited urban settings.

Airborne measurements over the Southeast US during the summer of 2013 revealed the evolution of nitrogen oxides and nitric acid that was markedly different from that over the Northeast US during the winter of 2015. We can reconcile the observed trends by invoking nitrogen oxide reservoir species formed under elevated NO _x environments that then convert to HNO ₃ . This process may represent a daytime loss pathway of NO _x that is faster by many factors than the oxidation of nitrogen dioxide by hydroxyl radicals. We present measurements of a suite of organic nitrates made using online iodide‐ionization time‐of‐flight mass spectrometry and simple box modeling that support their role in controlling the net conversion of NO _x to HNO ₃ . Rapid loss of organic nitrates derived from biogenic volatile organic compounds, such as what we infer from this analysis, has significant implications for our understanding of oxidant budgets and top‐down NO _x emissions assessments.

Addressing urban air pollution requires identifying, quantifying, and mitigating emission sources, yet bottom-up inventories often lack fine spatiotemporal precision and rarely capture unpermitted or informally operated sources. We combine dense mobile and fixed-site measurements in a Bayesian inverse modeling framework to produce hourly, hyperlocal black carbon (BC) emission maps (150 m × 150 m) that reveal how and where key sources are underrepresented. In West Oakland, CA — an overburdened community near major freight infrastructure — our approach identifies underestimated and unaccounted emission sources, increasing estimated BC emissions by ~33%. These corrections quadruple the contribution of neighborhood and port-related sources to population-weighted exposures (from 5 to 20%), disproportionately affecting low-income neighborhoods adjacent to freight corridors. Our results demonstrate that combining dense observational data with inverse modeling enables detection and quantification of overlooked emissions, supporting targeted intervention. Sensitivity analyses indicate similar performance with a substantially reduced fixed-site network (~10 monitors; ~1 site / km ² , compared with 6.7 sites/km ² in the full campaign), paired with time-averaged mobile maps, supporting applicability in data-limited urban settings.

Identifying and quantifying local methane emitters remains a major challenge for atmospheric monitoring. We present a novel top-down method to estimate both the upwind location and emission strength of an unknown atmospheric source from a time series of concentration observations. The approach employs backward trajectories from a Lagrangian Particle Dispersion Model (LPDM) to derive a characteristic transfer function for each potential source region. The transfer function that best reproduces the observed enhancement identifies the most likely source location. In a second step, the emission strength is inferred from the particle ensemble and its corresponding surface footprint. The method was developed and tested using data from a six-week measurement campaign in the San Francisco Bay Area, where six EM27/SUN near-infrared Fourier transform spectrometers were operated as part of a collaborative effort to quantify greenhouse gas emissions. At the UC Berkeley site, one instrument recorded a strictly periodic methane enhancement of approximately 10 ppb occurring every 12 minutes. Since co-emitted species showed no correlation with this pattern, the signal was attributed to a single, point-like, puff-emitting methane source. Favourable meteorological conditions enabled the analysis of several enhancement peaks. The retrieved average emission strength during the emission episodes was 0.8–78 g CH₄ s^-1 (equivalent to 2.1–190 metric tons yr^-1). Although the exact source could not be identified in the field, the emission characteristics are consistent with periodic natural-gas venting from a heating system with an installed power output of approximately 500–1000 kW installed power. The study demonstrates the potential of this approach for detecting and characterising local methane emitters from ground-based remote-sensing observations.

Satellite measurements of are widely used for investigations of emissions, pollution exposure, and photochemical ozone production. The most commonly available data products compromise on accuracy and spatial resolution in favor of lower compute time. In addition to limited spatial resolution, inaccurate modeling of winds and misrepresentation of the lifetime contribute to large systematic errors in the satellite retrievals that depend on these modeled outputs. We introduce a signal‐derived retrieval (SDR) for Tropospheric Emissions: Monitoring of Pollution (TEMPO) which derives instrument‐resolution a priori profiles from measured slant column densities. This approach enables fast reprocessing of TEMPO data and removes the gradient smoothing caused by low resolution priors and systematic errors in models. The SDR product reduces the negative bias of TEMPO total against Pandora reference measurements by , with a larger impact on the tropospheric column. This bias correction is shown to have important effects on the application of TEMPO for relevant science questions. The disparity between low and high income areas in San Jose, CA increased by using the SDR. The maximum to upwind line density ratio increased by using the SDR, indicating higher emissions and a shorter lifetime. These results outline the magnitude of systematic bias introduced by modeled a priori profiles and reduced by the SDR.

Point sources are often major contributors of greenhouse gas and air pollutant emissions in urban areas. Dense air monitoring networks provide a unique avenue for studying point source emissions over long time periods. Here, we use the Berkeley Environmental Air‐quality and CO ₂ Network (BEACO ₂ N) to study CO ₂ and air pollutant emissions from an oil refinery in the city of Richmond, CA. We identify 266 plumes crossing one or more sites in the BEACO ₂ N network during 2022–2023 as having a source at the refinery and quantify CO ₂ emissions using the Gaussian plume model. The refinery is modeled as two point sources, and total CO ₂ emissions are found to be 61.3 ± 6.3 kg s ⁻¹ , in close agreement with the EPA Facility Level Information on GreenHouse gases Tool inventory and Carbon Mapper measurements. Additionally, plume composition was found to vary, with CO/CO ₂ enhancement ratios ranging from 0 to 5 ppb/ppm. Taking the average CO/CO ₂ ratio, CO emissions are estimated to be roughly 3,000 mt, a sizable amount compared to the city's estimated non‐refinery CO emissions of 6,800 mt. In addition, we show that dense monitoring networks provide a unique resource for point source emissions quantification with broad application to plumes of varying size and composition, and recommendations are laid out for future applications.

High-resolution Greenhouse Gas (GHG) estimation is critical for verifying emissions inventories and informing climate policy. Current state-of-the-art estimates rely on "bottom-up" inventories, which are expensive to maintain, subject to reporting lags, and sensitive to inconsistent data supply chains. Conversely, "top-down" global reanalysis products, such as CarbonTracker, offer high quality but lack the spatial resolution required for actionable local policy, and high accuracy estimation of individual large polluters.To bridge this gap, we present a deepC, a method that leverages high-resolution simulation data to inform a generative prior while assimilating diverse ground-truth observations. We learn a patch-based diffusion prior from multi-resolution simulations of regional and global carbon transport to model the joint distribution of winds, surface fluxes, column concentrations, and emissions. We then apply a Bayesian posterior formulation to guide the generation process using sparse observations from six satellite missions, ground stations, and coarse global reanalysis. To ensure consistency over large regions, we employ a novel spatio-temporal Markov blanket scheme during posterior sampling, producing carbon emissions estimates at 1km resolution.We demonstrate the model's efficacy in CONUS and Western Europe, achieving stable emissions trajectories with low error relative to high-quality ground sensor and TCCON data. Early experiments suggest that conditioning the prior on embeddings from remote sensing foundation models significantly improves generalization to unseen domains. Furthermore, the model is robust to distribution shifts -- maintaining coherence under simulated future background CO2 levels. Finally, our approach yields well-calibrated uncertainty quantification at high inference speeds with ensemble generation, highlighting its potential for rapid, transparent emissions stocktaking, and lag-free policymaking.

Identifying and quantifying local methane emitters remains a major challenge for atmospheric monitoring. We present a novel top-down method to estimate both the upwind location and emission strength of an unknown atmospheric source from a time series of concentration observations. The approach employs backward trajectories from a Lagrangian Particle Dispersion Model (LPDM) to derive a characteristic transfer function for each potential source region. The transfer function that best reproduces the observed enhancement identifies the most likely source location. In a second step, the emission strength is inferred from the particle ensemble and its corresponding surface footprint. The method was developed and tested using data from a six-week measurement campaign in the San Francisco Bay Area, where six EM27/SUN near-infrared Fourier transform spectrometers were operated as part of a collaborative effort to quantify greenhouse gas emissions. At the UC Berkeley site, one instrument recorded a strictly periodic methane enhancement of approximately 10 ppb occurring every 12 minutes. Since co-emitted species showed no correlation with this pattern, the signal was attributed to a single, point-like, puff-emitting methane source. Favourable meteorological conditions enabled the analysis of several enhancement peaks. The retrieved average emission strength during the emission episodes was 0.8–78 g CH₄ s^-1 (equivalent to 2.1–190 metric tons yr^-1). Although the exact source could not be identified in the field, the emission characteristics are consistent with periodic natural-gas venting from a heating system with an installed power output of approximately 500–1000 kW installed power. The study demonstrates the potential of this approach for detecting and characterising local methane emitters from ground-based remote-sensing observations.