I participated in the AGU Atmospheric Science poster session today in San Francisco. There was considerable interest in our research.
We assessed how satellite measurements of carbon monoxide can constrain boreal wildlife emissions. These fires impact the climate, air quality, and the boreal ecosystem carbon balance. The frequency and severity of these fires are expected to increase under global warming.
We employ a four-dimensional variational (4D-Var) data assimilation approach based on the GEOS-Chem transport model. A newly improved GEOS-Chem adjoint model is used to optimize prior emissions by separately assimilating satellite observations of CO from the Measurements of Pollution in the Troposphere (MOPITT), Infrared Atmospheric Sounding Interferometer (IASI), and Tropospheric Monitoring Instrument (TROPOMI).
We focus on a case study of the August 2018 Canadian wildfire event. Our inversion estimates are evaluated against ground-based measurements at
the Total Carbon Column Observing Network (TCCON) sites in East Trout
Lake (ETL), Canada, and Park Falls (PK), USA.
Our results show that MOPITT, IASI, and TROPOMI CO inversion estimates that use monthly assimilation windows were somewhat comparable to the prior and had a poor agreement with the TCCON measurements at ETL and PK.
However, shorter assimilation windows significantly increased the estimated CO emissions and improved the correlation with the TCCON measurements.
With the monthly assimilation window, Boreal North American a posteriori CO emissions show that the MOPITT inversion estimate in August is much higher than IASI and TROPOMI.
The CO estimates for all three assimilations are much more similar at shorter assimilation windows and are twice as large as the a priori.
Estimating episodic wildfire CO emissions requires sufficient observations and an effective inversion approach to constrain the emissions on short timescales.
