Global Air Pollution Inferred from Satellite Remote Sensing Randall Martin, Dalhousie and Harvard-Smithsonian with contributions from Aaron van Donkelaar, Dalhousie University Lok Lamsal, Dalhousie U NASA Goddard Rob Levy, Ralph Kahn NASA Michael Brauer, UBC Michal Krzyzanowski, WHO Aaron Cohen, HEI Workshop on Atmospheric Chemistry and Health: current knowledge and future directions 12 October 2011
Aerosol Remote Sensing: Analogy with Visibility Effects of Aerosol Loading 7.6 ug m ug m -3 Pollution haze over East Coast Waterton Lakes/Glacier National Park
Combined AOD from MODIS and MISR Rejected Retrievals for Land Types with Monthly Error vs AERONET >0.1 or 20% MODIS r = 0.40 (vs. in-situ PM 2.5 ) MISR r = 0.54 (vs. in-situ PM 2.5 ) Combined MODIS/MISR r = 0.63 (vs. in-situ PM 2.5 ) AOD [unitless] van Donkelaar et al., EHP, 2010
Calculate Coincident PM 2.5 /AOD with Chemical Transport Model (GEOS-Chem) Aaron van Donkelaar
Significant Agreement with Coincident In situ Measurements Satellite Derived In-situ Satellite-Derived [ μ g/m3] In-situ PM 2.5 [μg/m 3 ] Annual Mean PM 2.5 [ μ g/m 3 ] ( ) r MODIS τ 0.40 MISR τ 0.54 Combined τ 0.63 Combined PM van Donkelaar et al., EHP, 2010
Evaluation with measurements outside Canada/US Global Climatology ( ) of PM 2.5 Better than in situ vs model (GEOS-Chem): r= , slope = 0.63 – 0.71 Number sitesCorrelationSlopeOffset (ug/m 3 ) Including Europe Excluding Europe van Donkelaar et al., EHP, 2010
Error in Satellite-Derived PM 2.5 has Three Primary Sources Satellite Error limited to % by AERONET filter Implication for satellite PM 2.5 determined by η Satellite-derived PM 2.5 = AOD Model Affected by aerosol optical properties, concentrations, vertical profile, relative humidity Most sensitive to vertical profile [van Donkelaar et al., 2006] Sampling Biases Satellite retrievals are at specific time of day for cloud-free conditions
τ a (z)/τ a (z=0) Altitude [km] Evaluate GEOS-Chem Vertical Profile with CALIPSO Observations Coincidently sample model and CALIPSO extinction profiles –Jun-Dec 2006 Compare % within boundary layer Model (GC) CALIPSO (CAL) Optical depth above altitude z Total column optical depth
Error Estimate Estimate error from bias in profile and AOD ±(1 μg/m %) Contains 68% (1 SD) of North American data Total uncertainty 25% (with sampling) Global population-weighted mean uncertainty 7 μg/m 3 van Donkelaar et al., EHP, 2010 Satellite-Derived [ μ g/m3] In-situ PM 2.5 [μg/m 3 ]
van Donkelaar et al., EHP, 2010
Wildfires near Moscow in Summer 2010 MODIS/Aqua: 7 Aug 2010
Spatial and Temporal Variation in Satellite-Based PM 2.5 during Moscow 2010 Fires van Donkelaar et al., AE, 2011
Satellite-based Estimates of PM 2.5 in Moscow Before Fires During Fires van Donkelaar et al., 2011 MODIS-based In Situ PM 2.5 In Situ from PM 10 r 2 =0.85, slope=1.06
Similar Technique to Infer Ground-Level NO 2 from OMI Lamsal et al., JGR, 2008
Challenges Remote Sensing: Improved algorithms to increase accuracy and resolution Modeling: Develop representation of processes Develop assimilation capability to inform AOD/PM 2.5 Measurements: More needed for evaluation throughout the world Encouraging Prospects for Satellite Remote Sensing of Air Pollutants Acknowledgements: Health Canada NSERC NASA