Seminar

Abstract

By altering the chemical composition of the atmosphere, humans have induced shifts in climate, reductions in human health, and perturbations in ecosystems. Atmospheric models are essential tools for evaluating the drivers of these changes, especially when the air quality or climate end points are well represented by the models. Nevertheless, typical approaches to evaluating the sensitivity of impacts with respect to emissions are computationally expensive or limited in accuracy. As an alternative, the adjoint of an atmospheric chemical transport model can be used to simultaneously calculate the relative influence of many model parameters, such as emissions, on select endpoints with analytical accuracy. In the context of environmental change, these results facilitate evaluation of tradeoffs between emissions reduction strategies. In cases where disagreement between models and observations exists, an adjoint facilitates adjustment of model parameters to statistically optimize the agreement between dense observations and model predictions in the four-dimensional variational data assimilation framework.

Here, adjoint-based sensitivity analysis and variational data assimilation are applied to domestic air quality and global climate questions. First, emissions influences on cloud droplet number concentration in select months of 2008 are evaluated using the global GEOS-Chem adjoint coupled with the Nenes and Seinfeld aerosol activation parameterization. Additionally, the method for assimilation of ammonia observations from the satellite-based Tropospheric Emissions Spectrometer over North America in summer of 2010 is described. Finally, the utility of an adjoint for environmental decision making is demonstrated by evaluating the influence of energy system emissions on potential human health and crop disbenefits due to ozone exposure. This assortment of applications will demonstrate the variety of questions that can be investigated with these modeling and sensitivity analysis tools.