Global Estimates of Ambient PM 2.5 Concentrations from Satellite AOD Aaron van Donkelaar with contributions from: Randall Martin, Brian Boys, Crystal Weagle,

Global Estimates of Ambient PM2.5 Concentrations from Satellite AOD

Aaron van Donkelaar

with contributions from:

Randall Martin, Brian Boys, Crystal Weagle, and Graydon Snider

VIIRS Aerosol Science and Operational Users Workshop

November 21-22, 2013

PM2.5 affects human health and longevity

Particulate Matter

Ground-level Ozone

Unsafe Water Supplyand Sanitation

Indoor Air Pollution

Malaria

0.0 1.0 2.0 3.0 4.0Deaths (millions of people) OECD, 2012

201020302050

• Life expectancy increases 7 months per 10 μg/m3 decrease in long-term exposure

• PM2.5 regulation achieved largest gains of all U.S. Federal Regulation

• Dominates global environmental mortality by 2030

Vast Regions Have Insufficient Measurements for Air Pollution Exposure Assessment

Locations of Publicly Accessible Long-Term PM2.5 Monitoring Sites

Monitor locations can be driven by compliance objectives

~1 site / 10,000 km2 in continental US & southern Canada

Emerging Network

Previous Global Burden of Disease Report for 2000Impaired by Insufficient Global Observations of Fine

Particulate Matter (PM2.5)

Estimate air pollution in urban areas causes:• ~1 million deaths (~1%)Could not estimate effects outside of urban areas for 2000 reportCould this be improved with satellite remote sensing?

Cohen et al., 2005

AOD is related to PM2.5…but how to separate surface aerosol from column concentrations?

Multiple approaches:• surface monitors calibration• empirical relations• model output

20 80 140

PM2.5 / AOD[μg/m3] GEOS-Chem v8-01-04

PM2.5 / AOD ratio is a function of:•vertical structure•aerosol type/hygroscopicity•meteorology

Significant agreement with coincident ground measurements over NA

SatelliteDerived

In-situ

Sat

ellit

e-D

eriv

ed [μ

g/m

3]

In-situ PM2.5 [μg/m3]

Ann

ual M

ean

PM

2.5 [

μg/

m3] (

2001

-200

6)

r

MODIS AOD 0.39

MISR AOD 0.39

Combined AOD 0.61

Combined PM2.5 0.77

18

16

14

12

10

8

6

4

2

0

van Donkelaar et al., EHP, 2010

Satellite-derived PM2.5 shows global agreement

Outside Canada/USN = 244 (84 non-EU)r = 0.83 (0.83)Slope = 0.86 (0.91)Bias = 1.15 (-2.64) μg/m3

CTM Agreement:r = 0.63 (0.72)slope = 0.54 (0.60)bias = 8.89 (2.45) μg/m3

van Donkelaar et al., EHP, 2010

Satellite-Derived PM2.5 (2001-2006) [μg/m3]0 5 10 15 20 50 80

Inform Epidemiological Studies:• Global childhood asthma (Anderson et al., 2012)

• Lung cancer in Canada (Hystad et al., 2012)

• Mortality in California (Jerrett et al., 2013)

• Diabetes (Brook et al., 2013; Chen et al., 2013)

• Global adverse birth outcomes (Fleisher et al., in press)

• Hypertension (Chen et al., in press)

• Low PM2.5 cardiovascular mortality (Crouse et al., 2012)

Global impact of global data

Global Burden of Disease 2010• 488 authors from 303 institutions in 50

countries• Inclusion of rural populations• PM2.5 causal role in 3 million deaths per year

Lim et al., Lancet, 2012

Crouse et al., EHP, 2012

Global in situ monitors are sparse...and far from AERONET

• Difficult to evaluate error sources• SPARTAN network collocates

PM2.5 and AOD measurements– Satellite-derived PM2.5 evaluation,

analysis,

Airphoton PM2.5 (Vanderlei Martins)

Filter PM2.5

Filter PM10

3-λ Nephelometer

AOD from CIMEL

Sunphotometer (e.g. AERONET)

www.spartan-network.org

• PM2.5 estimates accuracy impacted by– AOD retrieval– Simulated aerosol vertical profile– Mass to extinction conversion

• Sampling (clouds/snow)

Determine profile accuracy:• Compare with CALIOP

space-borne lidar• Match optical properties• Long-term (2006-2011)

comparisons

CALIOP allows vertical profile correction GEOS-Chem

CALIOP

Altit

ude

[km

]5

4

3

2

1

00 2 4 6 8 0 0.5 1 1.5 1.9Relative Profile Ratio Relative Profile

van Donkelaar et al., JGR, 2013

(𝑒𝑥𝑡𝑠𝑢𝑟𝑓𝐴𝑂𝐷 )𝐺𝐶

(𝑒𝑥𝑡𝑠𝑢𝑟𝑓𝐴𝑂𝐷 )𝐶𝐴𝐿

2

1.5

1

0.5

June-August 2006-2011

van Donkelaar et al., in prep.

1998-2012

PM2.

5 [μg

/m3 ]Long-term retrievals allow trends

combined MODIS, MISR and SeaWiFS-based PM2.5

Global:+0.6

Perc

ent o

f Pop

ulati

on [%

]

100

80

60

40

20

05 10 15 25 35 50 100

Annual Mean Exposure [μg/m3]

2012

2005

1998

Year

IT1

IT3

IT2

AQG

1998-2012 exposure trend [μg/m3/yr]

Global exposure and trends vary by region

van Donkelaar et al., in prep

Europe,Central

-0.2

Asia,South+1.0

Europe,Western

-0.2

North America,High Income

-0.3

Asia,East+1.4

North Africa,Middle East

+0.4

1008060

4020

0

8060

4020

0

8060

4020

0

Perc

ent o

f Pop

ulati

on [%

]

5 10 15 25 35 50 100Annual Mean Exposure [μg/m3]

5 10 15 25 35 50 100

Some Ongoing & Future Activities

Chinese PM2.5 Events

Size Resolved AOD/PM2.5 Calculation

1 km Optimal Estimation Retrieval

PURE Air: Prospective Urban and Rural Epidemiological Study

Operational AIRNow Satellite Data Processor

PCW/PHEOSEnvironmental Performance Index

ESCAPE - EUROPEAN STUDY OF COHORTS FOR AIR POLLUTION EFFECTS

Extras

Short-term agreement during extreme events – Moscow 2010

van Donkelaar et al., AE, 2011

MODIS/AquaAug 7th, 2010

Moscow

50 km

MODIS-based

In Situ PM2.5

In Situ PM2.5 from PM10

r =0.92, slope=1.061000

800

600

400

200

0

PM2.

5 [μg

/m3 ]

Jul-14 Jul-19 Jul-24 Jul-29 Aug-03 Aug-08 Aug-13 Aug-18

65°N

60°N

55°N

50°N

120100

80

60

40

20

0

PM2.

5 [μg

/m3 ]

Before FiresJuly 7- July 29, 2010

During FiresJuly 7- July 29, 2010

35°E 45°E 55°E 35°E 45°E 55°E

Daily Satellite-based Estimates in North America Benefit from Spatial Smoothing & Bias Correction

Error in Original (2004, 2006,2008)

Error in Smoothed & Bias-corrected (2004, 2006, 2008)

van Donkelaar et al., ES&T, 2012

Bias Corrected & Smoothed, Jun 27, 2005

OE

PM2.

5 [μg

/m3 ]

In situ PM2.5 [μg/m3]0 10 20 30 40

40

30

20

10

0

15

10

5

0

OE

PM2.

5 [μ

g/m

3 ]

Simulated values can constrain retrievals

2005van Donkelaar et al., JGR, 2013

2ε

2

a

a

2a

2a

σ

AODdAOD

dρρ

σAODAOD

J(AOD)

observed TOAreflectance

a priori AODa posteriori

AOD

a priori errorobservational

error

Satellites provide:• High resolution (0.1° × 0.1°)• Observational constraint• Column-integrated value

GEOS-Chem provides:• AOD/PM2.5 relationship• A priori values

Optimal estimation constrains AOD retrieval by error:

Available from 2004-2011

Sampling Station Designed for Nearly Autonomous OperationFilter Cartridges Replaced Every Two Months

Nephelometer (1 Hz) Resolves Satellite Overpass vs 24-hr Solar Capability 8-slot Filter Cartridge Collect PM2.5 & PM10

Cartridge Protects Filters

SPARTAN Activities at DalhousiePre&Post Weigh Filters Analysis for Ions, Metals, Black Carbon

(Ion Chromotraphy; Inductively Coupled Plasma Mass Spectrometry; Absorption )

SPARTAN Activities

Expansion &Supersite Development

Beijing, China Dhaka, Bangladesh

Kanpur, India