Skip to main content
U.S. flag

An official website of the United States government

Dot gov

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Https

Secure .gov websites use HTTPS
A lock ( ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

ACMD Seminar: Tailoring atmospheric inversions for estimating greenhouse gas emissions in urban regions

Sharon Gourdji and Subhomoy Ghosh
Greenhouse Gas and Climate Science Measurements Program, NIST

Monday, January 13, 2020, 3:00–4:00 PM
Building 101, Lecture Room B
Gaithersburg

Monday, January 13, 2020, 1:00–2:00 PM
Building 1, Room 4072
Boulder

This talk will be broadcast on-line using BlueJeans.  Contact acmdseminar@nist.gov for details.

Abstract: The atmosphere contains the downwind signature of greenhouse gas sources and sinks, which includes human emissions, occurring near the earth surface.  Thus, given a set of atmospheric observations of greenhouse gas mole fractions, a surface flux distribution can be inferred using an inverse modeling framework with inputs from an atmospheric transport and dispersion model.  In NIST’s Greenhouse Gas Measurement program, we are currently focused on estimating CO2 and CH4 emissions in the Washington DC/ Baltimore metropolitan area using data collected from a recently expanded network of surface towers and dedicated aircraft campaigns in the region.  Atmospheric inverse models for estimating GHG’s were originally developed at the global scale, using towers sited in areas sampling well-mixed air, important for inferring continental-scale fluxes.  However, at the urban scale, many of the new observational towers are sited in areas with high near-field flux variability both in space and time.  This variability presents unique mathematical challenges for using atmospheric observations appropriately in an inversion framework.  In particular, the proper treatment of a priori covariances in a high spatiotemporal resolution Bayesian inversion is an ongoing research area.  In an urban region, the model-data mismatch covariance should account for correlated transport errors, and the prior flux covariance matrix should account for spatial correlation along roadways, diurnal cycle traffic patterns, and discrete point sources within the domain.  Other scientific challenges include the estimation of spatiotemporally varying mole fractions in incoming air to the domain, the proper estimation of uncertainty on the posterior fluxes, and computational issues associated with estimating fine-scale (e.g. 1km) fluxes at an hourly timescale for an entire year.  Here we will present an overview of the problem, our current approaches for handling it, and some preliminary results.

Bios:

Sharon Gourdji:  As a research scientist in NIST’s Greenhouse Gas Measurement Program, Sharon is currently focusing on the biological aspect of the carbon cycle and investigating approaches for separating biospheric and anthropogenic contributions to total CO2 surface fluxes in atmospheric inversions.  She did her Ph.D. at the University of Michigan in Environmental Engineering, with her dissertation focused on global and regional-scale CO2 inversions using a geostatistical inverse modeling framework with surface tower data.  Her undergraduate degree is in mathematics.

Subhomoy GhoshSubhomoy Ghosh is currently a guest researcher at NIST's Green House Gas Measurement program, with a joint appointment as a computational scientist at the University of Notre Dame.  His current research focuses on developing scalable statistical models and methods for flux inversion problems using techniques from spatio-temporal modeling in statistics, Bayesian methods, machine learning, & high-performance computing.  Previously, he was a postdoctoral scholar at New York University Abu Dhabi (NYUAD).  He received his Ph.D. in Statistics from Iowa State University.

Note: Visitors from outside NIST must contact Cathy Graham; (301) 975-3800; at least 24 hours in advance.

Created December 9, 2019