Forecasting severe pollution episodes

Air pollution can be extremely harmful to the population, thus being able to anticipate the occurrence of pollution episodes is useful for people to plan their like around them (e.g., avoid exercising in bad air days). Also, in countries including Chile, France, and more recently China, the government has the capability to invoke temporary measures to reduce emissions that could lower the impact of such episodes. However, for these measures to have the desired impact, they need to be invoked before the episode unfolds, thus there is a growing need to have accurate air quality forecasts that could predict episodes a few days in advance.

Our research focus on developing these forecasting tools, find ways to improve them, and help in their transition to operations. We developed the system that is used by the Chilean government to declare pollution episodes in multiple cities (see our publication here) and are actively studying Chinese haze (see our publications).

The figure above shows an example of an episode unfolding in the city of Temuco, Chile, as a combination of stagnant conditions and intense emissions due to wood burning stoves for heating which in only 5 hours are able to rise particulate matter concentrations by over an order of magnitude and then slowly decrease overnight. The circles represent the air quality stations, while the color map the model prediction, which in this case was able to properly represent the episode.Windbarbs and topography are also represented in the figure.