Integration of Earth Observation into the UK Met Office Air Quality Forecasting System

Poor air quality (AQ) is one of the largest environmental stresses on human health. In the UK, poor AQ results in 28,000-36,000 premature deaths per year and annual socioeconomic costs of ~£20 billion. To help address this, the UK Met Office (UKMO) provides critical national daily AQ forecasts of key pollutants (e.g. ozone (O₃), nitrogen dioxide (NO₂) and aerosols) to provide the public and government bodies (e.g. Defra) with prior warning of hazardous AQ events.

To evaluate the skill of their forecast model (AQUM – Air Quality in the Unified Model), and to bias-correct the forecasts, the UKMO use AQ measurements from the UK Automated Urban and Rural Network (AURN) of surface sites. The AURN observations are used in the “Statistical Post Processing of Observations (SPPO)” step to correct the forecasts (known as “hybrid-forecasts”) before release. However, sparse surface monitoring sites are often unrepresentative of widespread pollution.

Satellite AQ data provides a powerful resource to help address this issue with daily UK spatial coverage, detection of pollution hotspots and transboundary pollution gradients. Therefore, this project described here (AIRSAT) aims to integrate key satellite AQ products (e.g. tropospheric NO₂ & O₃) into the UKMO’s SPPO framework to improve these “hybrid-forecasts”, thus benefiting the downstream users of this service.

Analysis of multiple satellite AQ products concludes that tropospheric column NO₂ (TCNO₂), total column ammonia (TCNH₃) and lower (0-6 km) tropospheric column O₃ (LTCO₃) from the TROPOspheric Monitoring Instrument (TropOMI), the Cross-track Infrared Sounder (CrIS) and the Infrared Atmospheric Sounding Interferometer (IASI), respectively, are the most suitable datasets for UK AQ monitoring and for comparison with AQUM.

AQUM showed good agreement with TropOMI TCNO₂ but has substantial (i.e. absolute model-satellite bias is greater than the satellite uncertainty) negative biases over London. Consistent with other studies, this indicates that the nitrogen oxide (NO_x) emissions from the official bottom-up inventory (National Atmospheric Emissions Inventory – NAEI) are too low over London. Comparisons between AQUM and CrIS in summer indicate that the model substantially underestimates NH₃ over broad rural regions of the UK. Primary NH₃ emissions are linked to agricultural processes and the model biases co-locate with the regions, which is supported by surface model-observational comparisons. These findings are consistent with previous research but building on it taking account of key factors like satellite vertical sensitivities and errors.

Investigation of the 2-week air pollution episode (24^th June – 7^th July 2018) shows widespread enhancements in NO₂ and O₃ surface concentrations and satellite integrated columns. By using AQUM, O₃ enhancements are detected throughout the lower-mid troposphere across the UK. This is a novel result as it confirms that IASI retrieved LTCO3 is detecting O₃ originating from the surface / boundary layer (i.e. suitable for AQ applications) despite having peak measurement capabilities in the mid-troposphere.

Overall, I will present these results from the AIRSAT project (i.e. Work Package (WP) 1) and new developments in WP2 utilising TropOMI TCNO₂ and AQUM to infer surface NO₂ concentrations for evaluation of the modelled forecasts suitable for the SPPO approach.