Novel techniques for evaluating air quality forecasts

Poor air quality has serious impacts on human health, in particular for those who suffer from respiratory illnesses. For example, during the heat wave of August 2003 it was estimated that up to 800 premature deaths in the UK were associated with poor air quality. Thus the provision of accurate air quality forecasts is very important to provide guidance to those who are vulnerable to high pollution levels. Air quality forecasts are produced using a combination of weather forecast models and chemistry models. The accuracy of air quality forecasts therefore depends on the accurate representation of the meteorology, the chemistry, and the interactions between the meteorology and the chemistry in these models. For example, during summer high pressure events, wind speeds are generally low and sunshine levels are generally high due to the absence of clouds. The abundance of sunshine results in an enhancement in ozone production. The low wind speeds mean that there is little transport of ozone away from the Earth's surface resulting in a build-up of ozone concentrations. This has consequent impacts on human health and crops. The ability to forecasts such poor air quality events can help improve the way that people prepare and behave during hazardous episodes.

In order to improve air quality forecasts, sources of error in the representation of the meteorology, the chemistry, and the interactions between the meteorology and the chemistry need to be identified. This project aims to make use of geo-statistical techniques to create interpolated maps of observations with which to evaluate the Met Office air quality forecasts. The student will explore new verification techniques for air quality forecasting and determine the most suitable methods. Finally, the student will establish statistical connections between the Met Office meteorological forecast scores and the skill of the air quality forecast. This will be useful to determine the degree to which the performance of the air quality forecasts is dependent on that of the meteorological model, or on the representativity of chemistry and emissions under different synoptic situations.

This research is essential if the Met Office is to evaluate high resolution air quality models in a consistent framework. The CASE studentship will provide improved observational datasets and novel evaluation for the operational verification of air quality forecasts. These techniques can then be applied to determine how subsequent changes in model resolution or in model parameterisations of chemical and physical processes impact the performance of air quality forecasts.