Monitoring air quality from space: fast retrievals of short-lived atmospheric pollutants and precursors

Climate change and air quality are the most pressing environmental issues of our lifetime, and they are inextricably linked.
Nitrogen oxides (primarily produced through combustion processes) and volatile organic compounds are precursors of tropospheric ozone, a strong greenhouse gas and one of the largest components of the radiative forcing of climate (after carbon dioxide and methane). They are also precursors of secondary aerosols, also known as particulate matter, the most harmful form of air pollution, particularly if smaller than 2.5 micron in diameter (PM2.5) and are able to penetrate deep into the lungs, blood streams and brain, leading to ill health and premature death. These aerosols are reflective, scattering solar radiation back to space, and tend to have a cooling effect on climate. Therefore, efforts to improve air quality will likely lead to further increases in temperature. On the other hand, these temperature increases will lead to changes in the chemistry associated with tropospheric ozone formation. Increases in temperature will also lead to an increase in the VOCs emitted from vegetation, providing more ozone and PM2.5 precursor.
A number of atmospheric sounders, with low radiometric noise and high spectral resolution, measure upwelling radiances in the thermal infrared spectral region; from these measurements we can determine the concentrations of many short-lived pollutants and precursors, e.g. ammonia, largely arising from agricultural sources (primarily ammonia-based fertilizers and animal manure), which significantly contributes to the formation of PM2.5; isoprene, the dominant biogenic volatile organic compound emitted by vegetation, which is chemically reactive and leads to the production of tropospheric ozone and secondary aerosols; as well as a wealth of pollutants produced from fires, such as methanol, formic acid, peroxyacetyl nitrate, ethene, and ethyne.
We live in an age when satellite-based measurements of atmospheric composition are becoming more and more ubiquitous. The ever increasing amounts of data available, which are only going to increase in the future, will require smarter and faster computational methods to extract meaningful information from these observations. Traditional retrieval methods based on line-by-line radiative transfer models are slow but accurate. The challenge is to maintain this accuracy whilst significantly improving the speed.