INTRODUCTION Points to make: - Aerosols harmful for health - Air quality in Australia is generally good, but may be getting worse in cities - AOD offers us a long-term spatial average and allows us to work out trends, spatial hot-spots, incidents, etc. - AOD can provide spatial validation of the models Aims of paper: - Test whether AOD is a decent proxy for PM2.5 in the GMR - Test models’ spatial skill for aerosols via AOD, throughout each campaign - Identify some high AOD days, and hot-spots during the period - relate them to case studies - Look at long-term spatial averages and tends across the GMR METHODS AND RESULTS Surface PM and remotely-sensed AOD Introduce: - Satellite AOD datasets + any processing - Surface AOD datasets - Surface PM2.5 datasets - Also, potentially, the region (topographic + population maps) Examine: - Relationship between surface-based AOD and PM2.5 at any collocated sites - Scatter-plots at sites, possibly examining the relationship for different seasons, wind-sectors, wind-speed bins - Relationship between surface-based AOD and satellite-based AOD - Same as for surface-based AOD vs. PM2.5 - Relationship between satellite-based AOD and surface PM2.5 - Same as for surface-based AOD vs. PM2.5 For this I will need: - All surface PM2.5 obs in the region - All surface-based AOD obs in the region - Collocated satellite-based AOD time-series Model skill for AOD during campaigns Introduce: - The models - The campaign periods - Any averaging procedures Examine: - Spatially averaged AOD for each campaign - Gridded modelled, gridded retrieved maps - Look at bias, correlation, standard deviation ratio (other metrics ??) - Modelled surface PM vs AOD - Is the relationship strong? Scatterplots modelled AOD vs modelled PM - Is the midday/snapshot AOD representative for daily AOD - for daily PM? Scatterplots of snapshot vs daily average. - Under what conditions is the relationship strong? ie. wind sectors, wind speed bins, times of day? Plots of correlations of the above conditional on the wind speed/direction, hour-of-day. - Incidents/episodes - Any smoke events? Plots of MODIS fires, plots of surface PM and AOD. - Representative examples of on-shore or off-shore flow. Plots of PM and AOD for these times. For this I will need: - Gridded hourly AOD from each model - Gridded AOD from the satellite products - Gridded hourly surface PM from each model - MODIS fire product for the region Trends and patterns in AOD across the region Examine: - Long-term averages - Maps of averages - Trends - Maps of trends - Time-series of trends at specific locations/regions - Possibly averaging across the area using an exposure proxy (e.g. weighting by population/gridcell) - Seasonal patterns - Time-series: monthly median 10%, 90% quantiles for specific locations/regions For this I will need: - Gridded AOD from the satellite products, over as long a time-series as possible DISCUSSION Points to make: - The models were able to reproduce these trends... but not these ... - The models detected some events (...) but not others (...) - Relationship between AOD and surface PM is weak/strong in the GMR - Gridded AOD shows an increasing/decreasing trend across the GMR