Air quality and climate impacts of current and future energy strategies
Robust assessment of the wide range of energy policies considered for meeting air quality and climate related objectives requires a rapid approach to analyzing the impact of emission control strategies on the chemical and physical state of the atmosphere. The feasibility of exploring such a wide parameter space is hindered by the complexity of models of atmospheric chemistry and physics.
Our research focuses on a specialized approach to evaluating the sensitivity of atmospheric models to changes in energy usage and emissions. Borrowing techniques from variational calculus and optimal control theory, we use adjoint modeling to rapidly calculate the sensitivity of a single model response (ci) with respect to numerous (millions) of parameters (pj).
How is this used? Consider the following example where the parameters are emissions estimates of ozone precursors, and the model response of interest is the amount by which simulated ozone concentrations exceeded air quality standards in the U.S. The next figure shows the influence of different emissions on U.S. ozone air quality exceedances.
How will these influences change as energy policies and emissions levels around the world evolve? How will these influences respond to a changing global climate? We use adjoint sensitivity modeling to tackle these questions, and other issues such as:
Our research focuses on a specialized approach to evaluating the sensitivity of atmospheric models to changes in energy usage and emissions. Borrowing techniques from variational calculus and optimal control theory, we use adjoint modeling to rapidly calculate the sensitivity of a single model response (ci) with respect to numerous (millions) of parameters (pj).
Fig: The illustration on the left shows how sensitivities are traditionally calculated in a forward model by successive perturbation of individual parameters (pj). In contrast, the adjoint approach depicted on the right is used to trace a perturbation in a model response (ci) back to variations in all the model parameters.
How is this used? Consider the following example where the parameters are emissions estimates of ozone precursors, and the model response of interest is the amount by which simulated ozone concentrations exceeded air quality standards in the U.S. The next figure shows the influence of different emissions on U.S. ozone air quality exceedances.
Fig: The sensitivity of the SOMO35 ozone air quality index in the U.S. during April 2001 with respect to emission from various ozone precursors. Note that these results, calculated in a few hours with an adjoint model, would have taken several months to evaluate using a traditional forward modeling to evaluation sensitivities.
How will these influences change as energy policies and emissions levels around the world evolve? How will these influences respond to a changing global climate? We use adjoint sensitivity modeling to tackle these questions, and other issues such as:
- What are the most effective emissions control strategies for meeting the National Ambient Air Quality Standards for PM2.5 in the U.S.?
- What is the influence of specific emissions from different energy sectors on the direct radiative forcing of aerosols and ozone?
Remote sensing of atmospheric trace gases
In the last decade the launch of several new satellite instruments dedicated to detection of atmospheric trace gases has revolutionized how we approach atmospheric chemistry. These instruments provide an unprecedented amount of spatial and vertical information about the chemical content of the atmosphere, continually challenging our ability to understand and model our environment.
Our research focuses on exploring the robustness, validity and utility of satellite measurements of reactive trace gas species. Satellite measurements from the TES and SCIAMACHY instruments are used in combination with atmospheric models to provide insight into species such as ozone (O3), nitrogen dioxide (NO2), and ammonia (NH3).
The objective is to better understand the sources and sinks of these species and the way in which they affect air quality and influence climate change.
Movie (courtesy of NASA): shows several satellites and how their instruments are scanning the atmosphere.
Our research focuses on exploring the robustness, validity and utility of satellite measurements of reactive trace gas species. Satellite measurements from the TES and SCIAMACHY instruments are used in combination with atmospheric models to provide insight into species such as ozone (O3), nitrogen dioxide (NO2), and ammonia (NH3).
The objective is to better understand the sources and sinks of these species and the way in which they affect air quality and influence climate change.
Inverse modeling
Inverse modeling is the process by which a range of feasible model parameters is reduced to only those that give model estimates consistent with observations.
Our approach to inverse modeling is to use an adjoint model to propagate information about the discrepancy, J, between the model estimates and the observations, back to the model parameters. This information is used iteratively with an optimization algorithm to refine the allowed range of model parameters.
One recent inverse modeling application was using surface measurements of aerosol concentrations to estimate the sources of the gas-phase aerosol precursors. The findings reveal an important discrepancy in our current understanding of NH3 emissions regarding the season of maximal emissions, with implications for modeling and controlling PM2.5 air pollution in the U.S.
Other inverse modeling applications include:
Our approach to inverse modeling is to use an adjoint model to propagate information about the discrepancy, J, between the model estimates and the observations, back to the model parameters. This information is used iteratively with an optimization algorithm to refine the allowed range of model parameters.
Fig: schematic of the iterative use of forward and adjoint models for inverse modeling.
One recent inverse modeling application was using surface measurements of aerosol concentrations to estimate the sources of the gas-phase aerosol precursors. The findings reveal an important discrepancy in our current understanding of NH3 emissions regarding the season of maximal emissions, with implications for modeling and controlling PM2.5 air pollution in the U.S.
Fig: results of inverse modeling compared to previous studies -- when should emissions peak?
Other inverse modeling applications include:
- Constraining uncertainties in chemical mechanisms of Titan's atmosphere
- Estimating aerosol growth parameters based on time series of size-distribution measurements
Secondary Organic Aerosol
Nearly half of the average mass of fine particulate matter is secondary organic aerosol (SOA) -- material that has been formed via chemical processing of volatile organic compounds to form non-volatile species that condense or remain in the aerosol phase. This is in contrast to primary organic aerosol, which is emitted directly as non-volatile particulate matter.
The chemical processes by which secondary organic aerosol forms in the atmosphere are very complicated and not entirely understood. Large uncertainties in our present grasp of the sources, identities and fates of SOA pose significant challenges for air quality and climate models.
Of particular interest is understanding the roles of the following species and issues when considering the air quality and climate impacts of SOA:
The chemical processes by which secondary organic aerosol forms in the atmosphere are very complicated and not entirely understood. Large uncertainties in our present grasp of the sources, identities and fates of SOA pose significant challenges for air quality and climate models.
Of particular interest is understanding the roles of the following species and issues when considering the air quality and climate impacts of SOA:
- Isoprene, a substantial biogenic source of organic carbon in the atmosphere with small, yet appreciable, SOA yields via a variety of pathways.
- Aromatic hydrocarbons, which have large man-made sources.
- The local oxidative environment, and how it determines the secondary organic aerosol yield
- Estimating how changing emissions and global climate in future decades will affect the distribution of organic aerosols.