Global Fine Particulate Matter Concentrations and Trends Inferred from Satellite Observations, Modeling, and Ground-Based Measurements
Randall Martin
with contributions from
Aaron van Donkelaar, Brian Boys, Matthew Cooper, Colin Lee, Ryan MacDonell, Graydon Snider, Crystal Weagle, Mark Gibson
EGU 30 April 2014
Michael Brauer (UBC), Aaron Cohen (HEI), Daven Henze (CU Boulder), Christina Hsu (NASA), Yang Liu (Emory), Zifeng Lu (Argonne), Vanderlei Martins (AirPhoton),
David Streets (Argonne), Siwen Wang (Tsinghua), Qiang Zhang (Tsinghua)
Vast Regions Have Insufficient Measurements for Fine Particulate Matter (PM2.5) Exposure Assessment
Locations of Publicly Accessible Long-Term PM2.5 Monitoring Sites
Emerging Network
Previous Global Burden of Disease Project for the Year 2000Impaired by Insufficient Global Observations of PM2.5
General Approach to Estimate Daily PM2.5 Concentration
Daily Satellite(MODIS, MISR, SeaWifs) Column of AOD Coincident Model
(GEOS-Chem) Profile
2.5,sat2.5,model
modelsat
PMA
PM OD
A DO
Alti
tude
Concentration
Accounts for • relation of “dry” PM2.5 with ambient extinction• relation of aerosol during satellite-observation vs continuous
Climatology (2001-2006) of MODIS- and MISR-Derived PM2.5
van Donkelaar et al., EHP, 2010EHP Paper of the Year
Evaluation in North America:r=0.77slope = 1.07N=1057
Outside Canada/USN = 244 (84 non-EU)r = 0.83 (0.83)Slope = 0.86 (0.91)Bias = 1.15 (-2.64) μg/m3
Used in Global Burden of Disease Study 2010
Lim et al., Lancet, 2012
PM2.5 Causal Role in
70 Million Disability Adjusted Life Years (~3%)>3 Million Excess Deaths (~5%)
Similar Conclusions Reached by WHO in 2014
Significant Association of Long-term PM2.5 Exposure with Cardiovascular Mortality at Low PM2.5
Three-fold increase in premature mortality rate over previous GBD study for 2000
Crouse et al., EHP, 2012
Benefits from Large Statistical Power
2
a2
aa2 2a ε
dρρ AODdAODAOD AOD
J(AOD)σ σ
Observed TOAreflectance
a priori AODa posteriori
AOD
a priori errorobservational
error
Optimal Estimation allows:• Error-constrained AOD solution• Consistent optical properties• Local reflectance information
CALIOPSpace-borne LIDAR
• Optimal Estimation AOD• CALIOP-adjusted AOD/PM2.5
MODISImaging Spectroradiometer
Optimal Estimation constrains AOD retrieval by error:
Enhanced Algorithm to Infer PM2.5 from MODIS
van Donkelaar et al., JGR, 2013
Optimal Estimation (OE) Can Improve Global AOD Retrieval
van Donkelaar et al., JGR, 2013
Opt
imal
Est
imat
ion
AO
D (U
nitle
ss)
2005 slope correlation
Region Operational
OE Simulated
Operational
OE Simulated
E NA 1.3 0.9 0.7 0.9 0.9 0.7
W NA 1.5 0.8 0.6 0.7 0.8 0.5
EU 1.2 1.1 1.4 0.8 0.8 0.6
India 1.2 0.7 0.7 0.8 0.8 0.5
E Asia 1.1 0.8 0.7 0.9 0.8 0.7
Africa 0.9 0.8 1.0 0.9 0.9 0.8
Western North America
Eastern North America
Num
ber o
f O
bser
vatio
ns
Best agreement
slope=1.47r=0.65
slope=0.81r=0.75
slope=0.56r=0.53
slope=1.25r=0.85
slope=0.94r=0.87
slope=0.71r=0.71
Operational
Operational = NASA MODIS Collection 5
Use CALIOP Observations (2006-2011) to Correct Seasonal Bias in Simulated Aerosol Extinction
van Donkelaar et al., JGR, 2013
η =
PM2.
5 / A
OD
Eastern US East China
Satellite-Derived PM2.5 Trends Inferred from SeaWifs & MISR AOD and GEOS-Chem AOD/PM2.5
Boys et al., submitted
MISR2000 -2012
SeaWiFS1998 -2010
PM2.5 Trend [μg m-3 yr-1]
Combine SeaWifs & MISR to Calculate 15-Year PM2.5 Timeseries (1998-2012)
Boys et al., submitted
Eastern North America
Middle East South Asia
East AsiaPM
2.5 (μ
g m
-3)
PM2.
5 (μ
g m
-3)
PM2.
5 (μ
g m
-3)
00.25
-0.25 1-1 1.5
-1.5 2-2
0.010.05
0.1
P- v
alue
PM2.5 Trend [µg m-3 yr-1]
-0.5
0.5
Consistent Trends in Satellite-Derived and In Situ PM2.5
Boys et al., submitted
Sate
llite
-D
eriv
edIn
Situ
In Situ (1999-2012)0.37 ± 0.06 μg m-3 yr-1
Satellite-Derived (1999-2012)0.36 ± 0.13 μg m-3 yr-1
Eastern US
PM2.
5 A
nom
aly
(ug
m-3)
SeaWifs & MISR
In Situ
1999-2012
Interpret Satellite-derived PM2.5 Trends with GEOS-Chem
Boys et al., submitted
Eastern North America Middle East
South Asia East Asia
Year Year
GEOS-Chem Secondary Inorganic -0.4 μg m-3 yr-1 GEOS-Chem Mineral Dust 0.7 μg m-3 yr-1
GEOS-Chem Secondary Inorganic 0.7 μg m-3 yr-1
GEOS-Chem Secondary Inorganic 0.8 μg m-3 yr-1
GEOS-Chem Organic 0.04 μg m-3 yr-1 GEOS-Chem Organic 0.2 μg m-3 yr-1
SeaWifs & MISR -0.39±0.10 μg m-3 yr-1 SeaWifs & MISR 0.81±0.21 μg m-3 yr-1
SeaWifs & MISR 0.93±0.22 μg m-3 yr-1 SeaWifs & MISR 0.79±0.27 μg m-3 yr-1
PM2.
5 [ug
/m3 ]
PM2.
5 [ug
/m3 ]
van Donkelaar et al., submitted
Changes in Long-term Population-Weighted Ambient PM2.5
Clean Areas are Improving; High PM2.5 Areas are DegradingWHO Guideline & Interim Targets
van Donkelaar et al., submitted
Changes in Long-term Population-Weighted Ambient PM2.5
Clean Areas are Improving; High PM2.5 Areas are DegradingWHO Guideline & Interim Targets
Exceedance of WHO AQG drops from 62% to 19%
Exceedance of WHO IT1 increases from 51% to 70%
1998
2012
1998 (51%)
2012 (70%)
WHO AQG
WHO IT1
SPARTAN: An Emerging Global Network to Evaluate and Enhance Satellite-Based Estimates of PM2.5
Measures PM2.5 Mass & Composition at Sites Measuring AOD
Semi-Autonomous PM2.5 & PM10 Impaction Sampling Station (AirPhoton)Ions & metals
3-λ Nephelometer
AOD from CIMEL Sunphotometer (e.g. AERONET)
www.spartan-network.orgSnider et al., in prep
Testing
Deployed
Committed
Prospective
Nonlinear Relation Between PM2.5 and SourcesWhich Local Sources Should be Reduced to
Decrease Mortality from PM2.5?
PM2.5Primary Chemistry Precursors
Sulfur Dioxide (SO2)Nitrogen Oxides (NOx)
Ammonia (NH3)
Adjoint Model: Calculate Sensitivity of Global Premature Mortality to Local Emissions
Emissions Chemistry & Transport Concentrations
Health Impact
Function
∂∂Emissions
Chemistry & Transport ∂
∂Concentrations
Global Mortality
GEOS-Chem
GEOS-Chem Adjoint
Colin Lee
Sensitivity of Global Premature Mortality to SO2 Emissions
Lee et al., in prep
PM2.5 subgrid variability resolved using satellite AODExposure-response function from Global Burden of Disease Project
ΔMortalityglobal / 10% ΔEmissions
Sensitivity of Global Premature Mortality to:
SO2 Emissions
NH3 Emissions
Lee et al., in prep
ΔMortalityglobal / 10% ΔEmissions
Insight into Global PM2.5 through Satellite Remote Sensing Modeling, and Ground-based Instruments
Acknowledgements:NSERC, Health Canada, Environment Canada
• Particulate matter is major risk factor for global premature mortality
• Regions with high PM2.5 have increasing concentrations
• Regions with low PM2.5 have decreasing concentrations
• Asian PM2.5 increasing by 1-2 ug/m3/yr
• SPARTAN and CALIOP evaluate AOD/PM2.5 simulation
• Adjoint allows efficient calculation of sensitivity of premature
mortality to emissions changes
• Controls in South Asia on SO2 much more effective than on NH3