Abstract: Data Assimilation and Inverse Modeling of the Atmospheric Composition
The increasing availability of measurements of chemical (e.g., NO2, CO, CO2, CH4) and aerosol (e.g., PM2.5 PM10, AOD) constituents, along with data on meteorological states (e.g., T, H2O, U/V) and land-surface characteristics, offers an opportunity to study changes in atmospheric composition through the integration of these measurements with predictions from regional to global chemical transport models within the Earth system framework. Central to this integration is a chemical data assimilation (DA) and/or inverse modeling system that is reasonably efficient, effective, and flexible in assimilating measurements spanning multiple spatiotemporal scales and multiple chemical/aerosol species. This lecture introduces the development and application of data assimilation and inverse modeling approaches in Earth system science particularly towards prediction and inference problems in atmospheric chemistry and physics. I will present three key scientific/technical problems that this community is attempting to address with these DA approaches. These are: 1) estimating sources and sinks of trace gases and aerosols, 2) assimilating multi-species and/or multi-platform and/or coupled chemical and weather data assimilation research; and 3) conducting observing system simulation experiments (or OSSEs) to support future satellite observations of global atmospheric composition and/or deployment of measurement network. I will end this lecture by posing a question regarding the complementary roles of machine learning and data assimilation in atmospheric composition and related studies.