Improved satellite retrievals of NO₂ and SO₂ over the Canadian oil sands and comparisons with surface measurements

Abstract. Satellite remote sensing is increasingly being used to monitor air quality over localized sources such as the Canadian oil sands. Following an initial study, significantly low biases have been identified in current NO₂ and SO₂ retrieval products from the Ozone Monitoring Instrument (OMI) satellite sensor over this location resulting from a combination of its rapid development and small spatial scale. Air mass factors (AMFs) used to convert line-of-sight "slant" columns to vertical columns were re-calculated for this region based on updated and higher resolution input information including absorber profiles from a regional-scale (15 km × 15 km resolution) air quality model, higher spatial and temporal resolution surface reflectivity, and an improved treatment of snow. The overall impact of these new Environment Canada (EC) AMFs led to substantial increases in the peak NO₂ and SO₂ average vertical column density (VCD), occurring over an area of intensive surface mining, by factors of 2 and 1.4, respectively, relative to estimates made with previous AMFs. Comparisons are made with long-term averages of NO₂ and SO₂ (2005–2011) from in situ surface monitors by using the air quality model to map the OMI VCDs to surface concentrations. This new OMI-EC product is able to capture the spatial distribution of the in situ instruments (slopes of 0.65 to 1.0, correlation coefficients of >0.9). The concentration absolute values from surface network observations were in reasonable agreement, with OMI-EC NO₂ and SO₂ biased low by roughly 30%. Several complications were addressed including correction for the interference effect in the surface NO₂ instruments and smoothing and clear-sky biases in the OMI measurements. Overall these results highlight the importance of using input information that accounts for the spatial and temporal variability of the location of interest when performing retrievals.