Current Research Areas

We use satellite data, large-scale computing, and machine learning to quantify concentrations and emissions of atmospheric pollutants. This involves solving inverse problems in radiative transfer and chemical transport. Our goal is to advance understanding of atmospheric composition while producing actionable information for policy on climate and air quality.

Methane is a potent greenhouse gas responsible for one third of atmospheric radiative forcing. Anthropogenic emissions are from a range of activities including agriculture, waste management, and fossil fuel production. Satellites are a powerful resource for quantifying methane emissions across scales to inform global climate policy and mitigation efforts.

Large, transient methane point sources have an outsized impact on climate and represent low-hanging fruit for emission mitigation. We are developing novel methane-retrieval and computer-vision algorithms to more efficiently detect these sources with high-resolution broadband and hyperspectral satellite instruments.

Objectives:

• Improve satellite methane retrieval precision and plume detection limits;
• Enable near-real-time detection of large methane point sources from both low-Earth and geostationary orbit;
• Create a comprehensive global catalog of methane point sources to guide mitigation efforts and elucidate trends.

We are developing the Integrated Methane Inversion (IMI) as an open-access community cloud-computing facility for inferring regional methane emissions from satellite observations. The IMI ingests daily global observations of atmospheric methane concentrations from the Sentinel-5p Tropospheric Monitoring Instrument (TROPOMI) along with 3D meteorological fields from the NASA Goddard Earth Observing System (GEOS) to quantify gridded methane emissions at up to 12-km resolution using a Bayesian analytical inversion. It can be applied to estimate emissions for any region, period, and industrial sector of interest. We are currently using the IMI to quantify the magnitude and variability of emissions from methane hotspots around the world, with a focus on oil, gas, and coal production regions.

Upcoming IMI developments include incorporating observations from new satellite missions (e.g., GOSAT-GW, Sentinel-5p) and suborbital platforms, improving data visualization and access through the Integral Earth (IE) interface to the IMI, and accelerating the inversions using machine learning.

Atmospheric nitrogen oxides (NO_x ≡ NO + NO₂) play important roles in air quality, radiative forcing, and nitrogen deposition to the biosphere. Satellites have observed atmospheric NO₂ concentrations since the 1990s with pixel resolutions ranging from hundreds of km to just a few km for the highest-resolution TEMPO mission launched in 2023. This is generally too coarse to resolve individual point sources (e.g., power plants, steel mills, cement plants) except for large isolated facilities.

We are developing new satellite retrieval techniques to map NO₂ concentrations and NO_x emissions at hyperlocal (sub-km) scales. Our goal is to enable facility- and neighborhood-level monitoring and attribution of air pollution, particularly in dense urban environments and in regions where pollution controls and in-situ monitoring are lacking.

Objectives:

• Quantify urban NO_x emissions at sub-km resolution;
• Determine the contribution of large point sources to regional NO_x emissions;
• Quantify the impact of these sources on local air quality.