Global air pollution modelling: uncertainties due to chemistry modelling

30 april 2019

Air pollution has important consequences on human health, while atmospheric composition aspects also play an important role in climate change and weather forecasts, mainly due to its radiative effects. A good representation of atmospheric composition is also important to quantitatively interpret satellite retrievals of air pollution, such as has become available from the TropOMI instrument.

These have been a major motivations for KNMI to invest in atmospheric chemistry modelling efforts, as part of the Copernicus Atmosphere Monitoring Service (CAMS). CAMS provides daily analyses and forecasts of atmospheric composition, analogue to the well-known daily weather forecasts.

“An important step has been reached recently by developing three fully functioning, independent modules to describe the chemical conversion of trace gases which describe tropospheric and stratospheric ozone, and related trace gases”, says lead author of a recent paper, Dr Vincent Huijnen.

By having these modules, overall uncertainties due to atmospheric chemistry modelling on resulting chemistry fields can now be quantified. Examples of causes of differences are different choices in selecting the set of main chemical trace gases and reactions, modelling assumptions on the impact of solar radiation and its cloud interaction on photolysis rates, choices in chemistry-aerosol interaction, differences in numerical methods to solve the set of equations. 
By comparing these model simulations to observations (see Figure 1), we learn about how much of discrepancies can be explained by uncertainties in such chemistry aspects, as compared to other uncertainties, such as emissions and transport.

A figure showing the differences between model output and observations
Figure 1: a) Observed carbon monoxide (CO) total column retrieval for April 2011, and corresponding model biases as simulated by three chemistry models; b) IFS(CBA), c) IFS(MOZ), and d) IFS(MOC).

V. Huijnen, et al. (2019): Quantifying uncertainties due to chemistry modelling – evaluation of tropospheric composition simulations in the CAMS model (cycle 43R1). Geoscientific Model Development. In press.