the accuracy of aerosol satellites' aod products over

12th Annual Symposium on New Generation Operational Environmental Satellite Systems

The accuracy of aerosol satellites‘ AOD

y

The accuracy of aerosol satellites‘ AOD

products over Chinaproducts over China

Dr. Jinyuan Xin; L. Wang, Y. Wang

LAPC, Institute of Atmospheric Physics, CAS2016-01-10

Institute of Atmospheric physics, CAS

2016 01 10

ContentsContents

1 Th h b k d1. The research background

2. The observation network

3. Research results and progress

4. Summary

Institute of Atmospheric Physics, CAS

1. The research background1. The research backgroundThe effect of aerosol

Observation Modeling

IPCC 2001 IPCC 2013

NCAR CESM MISR+AERONET

Ob d AODSimulated AOD:0.2-0.3 in East China

Observed AOD: 0.4-0.8 in East China

Institute of Atmospheric Physics, CASNCAR CESM simulated and offered aerosol data in IPCC AR5.

Y. Feng 2011, CESM NCAR, report

Aerosol optical parameters: Bridging the ground observation, satellite, and modelg g g , ,

Climate effect of aerosol

观测实验 ModelingObservationObservation

IPCC 2013IPCC 2001The AThe A--TrainTrain

NCAR CESM MISR+AERONET

观测中国东部模拟中国东部

AOD为0.2-0.3！

观测中国东部

AOD为0.4-0.8

Institute of Atmospheric Physics, CASNCAR CESM 为参加IPCC AR5评估的模式提供气溶胶数据IMPROVE in U.S.IMPROVE in U.S. EMEP in EuropeEMEP in EuropeNASA AERONETNASA AERONET

2. The observation network in China2. The observation network in China

The Chinese Sun Hazemeter Network, 2004-2007

The CARE-China Network_Sunphotometer, 2011-Now

FukangChangbai Mountain

SanjiangHailun

Shenyang

CERN station (19)

The Chinese Sun Hazemeter Network

Typical city site (4)Calibration site (2)

Beijing City

The Chinese Sun Hazemeter Network2004.08-2007.07 (23 sites)

Haibei

LhasaYanting

Taoyuan

Tai lake

Jiaozhou bay

Shapotou

Shenyang

Beijing Forest

Ansai Fengqiu

Shanghai City

Lanzhou City

Xianghe

j g yEerduosi

The CARE-China Network

Sanya

Taoyuan

Xishuangbanna

SanyaTaipei CityDinghu Mountain

e C C a Netwo_ Sunphotometer

2012.01-Now(36sites)

Microtops II

LED LED hazemeterhazemeterLED LED hazemeterhazemeterSolar Light Co.U.S. Forest Service

The observing group of the networkThe observing group of the network

SNJ SYB CLDWXZ GGSSNJ CSC

NJCCQC YCAHLADHD BJC CBM

LSZ SJZSPTFKZ QYZ XLZGZC

SYC JZB KMCTJC YTZBNZ DHM

Calibration and Quality ControlCalibration and Quality ControlTo build the observation and instrument calibration standard

T b ild h d li l d f h kTo build the data quality control procedure for the network

10.0

10.5

(a) ID023 meter, at the Lhasa site, 6 November 2004

Langley plot 1.2

14983 14986 16329 16328 16335

8 5

9.0

9.5

10.0

ln(V

-Vda

rk)

880nmR2=0.998

R2=0.996

650nm

R2=0.965

500CIMEL Com.

Langley plot

0 4

0.6

0.8

1.0 AOD at 500 nmBeijing2011.1.11-14

Lhasa2011.10.01

Langley plot at the Lhasa site0 2 4 6 8 10 12

8.0

8.5

Relative air mass

R2=0.998500nm405nm

CIMEL Com.

Transfer Cal.14983 14986 16329 16328 16335

0.0

0.2

0.4

1 11 21 31 41 51 61 71 81 91 101 111 121 131

1.0

1.5

1.0

1.5

AO

D y=x

650nm y=0.9770x-0.0184 (R2=0.941)

Test Stability 1.0

1.5

2.0

14983 14986 16329 16328 16335

Angstrom exponent

0.0

0.5

0.0

0.5

CIM

EL's

A

880nm y=0.8811x-0.0133 (R2=0.926)

(d)

In Rotation1 0

-0.5

0.0

0.5

1 11 21 31 41 51 61 71 81 91 101 111 121 131

Beijing Lhasa

Xin et al. JGR, 2007

0.0 0.5 1.0 1.50.0 0.5 1.0 1.5

ID005 hazemeter's AODData from Feb to Aug 2005 at Xianghe site

Comparison with AERONET siteTransfer calibration, test stability;

deal with the cloud pollutionCalibration Process

-1.0 j g

3. Research results and progress3. Research results and progress

3 1 The distribution of aerosol optical properties 3.1 The distribution of aerosol optical properties

3.2 The application to MODIS AOD upgrade in China

3.3 The accuracy of MODIS AOD C4, C5, C6 products

3 4 The acc rac of VIIRS IP AOD o er east China 3.4 The accuracy of VIIRS IP AOD over east China

3.5 The accuracy of Chinese FY-3A and GF-1 AOD

3.6 To find the relationship between PM2.5 and AOD


3.1 3.1 The distribution of aerosol optical propertiesThe distribution of aerosol optical properties

There were similar monthly and seasonal variations of AODs andaerosol types over northern and southern China.yp

Dust aerosol in spring, smoke and soot in autumn and winter. And dustover the Tibet Plateau, smoke and soot in Southeast Asia.

The annual The annual

Spring Summer

Dust Dust mean AOD at 500nm is 0.5-0.9 over the

l

mean AOD at 500nm is 0.1-0.2 over the

SpringFrom Aug 2004 t o Jul 2007

1.5

SummerFrom Aug 2004 to Jul 2007

1.5

AOD (500 )

Dust Dust

Dust Vapour

Smoke Smoke central, eastern, southern and eastern coastal

Tibet Plateau, remote northeast, and

AOD (500nm)Angstrom Exponent


Autumn Winter

Smoke Smoke

areas.2004.8-2007.7

Hainan island.2004.8-2007.7

WinterFrom Aug 2004 to Jul 2007

AutumnFrom Aug 2004 to Jul 2007

Dust Dust

Smoke Smoke

Xin et al. JGR, 2007, 2011; AE, 2008; AAS, 2010Institute of Atmospheric Physics, CAS

1.5


1.5


SmokeSmoke

The spatial and temporal distribution in 2012The spatial and temporal distribution in 2012

Comparing with 2004-2007, there was increasing of AOD in Chinaexcept the Tibet Plateau, similar variation of aerosol types as before.But many sites had large changes in the remote areas.


3.2 The application to MODIS AOD upgrade3.2 The application to MODIS AOD upgrade

In collaboration with UMD and U.S. Forest Service, we traced toevaluate the accuracy of MODIS AOD from C4 to C5 products.y p

The accuracy of MODIS C5 product increased by 10% on average inChina, but the ratio of available data decreased by 15%.

Available data: 60%Within NASA error: 50%Overestimate: 20%

Available data: 45%Within NASA error: 60%Underestimate: 15%

Valuation of MODIS Products with CSHNET2004.08-2006.06

12004.08-2007.07

Valuation of MODIS Products with CSHNETFrom Aug 2004 to Jul 2007

Evaluate MODIS C4 Product Evaluate MODIS C5 Product

R1: Ratio of MODIS to Site's ResultR2: Percentage within NASA Error

From Aug 2004 to Jul 20071

R1: Ratio of MODIS to Site's ResultR2: Percentage within NASA Error

Li, et al.

Wang, Xin, Li, et al., JGR, 2007; AE, 2007; CSB, 2007; GRL, 2008; AR, 2010Evaluate MODIS C4 Product Evaluate MODIS C5 Product


MODIS C5 AOD revised by the ground observation

MODIS C5 product also has large seasonal systematic errors in theurban region.urban region.

Overestimate by 20% in summer and autumn; underestimate by 20% inspring and winter.

Jan JanJanJanJanJan2004 2005 2006 2007 2008 2009 20102.0(d)

Observed AOD MODIS AOD

(d)

Jan JanJanJanJanJan2004 2005 2006 2007 2008 2009 20102.0 Observed AOD MODIS AOD

(d)


(d)


0 5

1.0

1.5( )

Val

ue

(d)

0 5

1.0

1.5

Val

ue

( )

0 5

1.0

1.5

Val

ue

( )

0 5

1.0

1.5

Val

ue

Jan FebMarAprMayJun Jul AugSep OctNovDec0.0

0.5


0.5


0.5


0.5

B k d it B iji Cit Sh A Ji h BBackground site Beijing City Shenyang Agr. Jiaozhou Bay

Xin et al., AE, 2011

MODIS C5 AOD revised by the ground observationMODIS C5 AOD revised by the ground observation

The available ratio of the revised MODIS product reached 60-85%,raised by ~20% on average; the relative error declined to ~15%.raised by 20% on average; the relative error declined to 15%.

The network was serviced to evaluate the Chinese satellites: HJ-1A ,GF-1 and FY-3A. The GF-1 is getting close to MODIS.

1.0

1.5

2.0

S A

OD

Beijing Forest (669 days)

0 94 0 041.0

1.5

2.0

1.0

1.5

2.0 The mean AOD for 0.1-interval observed AOD

60

80

100ag

e (%

)(e)

1.5

2.0

2.5

3.0

S A

OD

Jiaozhou Bay (649 days)

0 97 0 021.5

2.0

2.5

3.0

1.5

2.0

2.5


60

80

100

tage

(%)

(e)

1.0

1.5

2.0

S A

OD

Shenyang Agr. (644 days)

0 83 0 061.0

1.5

2.0

1.0

1.5


60

80

100

tage

(%)

(e)

1.5

2.0

2.5

3.0

S A

OD

Beijing City (650 days)

1.5

2.0

2.5

3.0

1.5

2.0

2.5


60

80

100

tage

(%)

(e)

0.0 0.5 1.0 1.5 2.00.0

0.5

1.0

Observed AOD

MO

DI

<1.0 >1.0

y=0.94x-0.04R2=0.81

0.0 0.5 1.0 1.5 2.00.0

0.5

1.0

0.0 0.5 1.0 1.5 2.00.0

0.5

1.0

0

20

40

Percentage

Perc

enta

0.0 0.5 1.0 1.5 2.0 2.5 3.00.0

0.5

1.0

1.5

Observed AOD

MO

DI

<1.0 >1.0

y=0.97x-0.02R2=0.73

0.0 0.5 1.0 1.5 2.0 2.5 3.00.0

0.5

1.0

1.5

0.0 0.5 1.0 1.5 2.0 2.5 3.00.0

0.5

1.0

1.5

0

20

40

Perc

ent

0.0 0.5 1.0 1.5 2.00.0

0.5

1.0

Observed AOD

MO

DI

<1.0 >1.0

y=0.83x-0.06R2=0.70

0.0 0.5 1.0 1.5 2.00.0

0.5

1.0

0.0 0.5 1.0 1.5 2.00.0

0.5

1.0

0

20

40

Perc

ent

0.0 0.5 1.0 1.5 2.0 2.5 3.00.0

0.5

1.0

1.5

Observed AOD

MO

DI

<1.0 >1.0

y=1.07x+0.11R2=0.83

0.0 0.5 1.0 1.5 2.0 2.5 3.00.0

0.5

1.0

1.5

0.0 0.5 1.0 1.5 2.0 2.5 3.00.0

0.5

1.0

1.5

0

20

40

Perc

ent

1 5

2.0

OD

Beijing Forest (669 days)

1 5

2.0

1 5


80

100(f)

2.5

3.0

OD

Jiaozhou Bay (649 days) (f)2.5

3.0

2.5


80

100

1 5

2.0

OD

Shenyang Agr. (644 days)

1 5

2.0

1 5


80

100(f)

2.5

3.0

OD

Beijing City (650 days)2.5

3.0

2.5


80

100(f)

Comparison with MODIS C5 AOD in the urban

0.5

1.0

1.5

Rev

ised

MO

DIS

AO

<1.0 >1.0

y= x + 9E-6R2=0.84

0.5

1.0

1.5

0.5

1.0

1.5

20

40

60

Percentage

Perc

enta

ge (%

)

0.5

1.0

1.5

2.0

Rev

ised

MO

DIS

AO

<1.0 >1.0

y= x - 4E-05R2=0.77

0.5

1.0

1.5

2.0

0.5

1.0

1.5

2.0

20

40

60

Perc

enta

ge (%

)

0.5

1.0

1.5R

evis

ed M

OD

IS A

O

<1.0 >1.0

y= x + 8E-06R2=0.72

0.5

1.0

1.5

0.5

1.0

1.5

20

40

60

Perc

enta

ge (%

)

0.5

1.0

1.5

2.0

Rev

ised

MO

DIS

AO

<1.0 >1.0

y= x + 2E-05R2=0.85

0.5

1.0

1.5

2.0

0.5

1.0

1.5

2.0

20

40

60

Perc

enta

ge (%

)

0.0 0.5 1.0 1.5 2.00.0

Observed AOD0.0 0.5 1.0 1.5 2.0

0.00.0 0.5 1.0 1.5 2.0

0.0 0 Percentage

0.0 0.5 1.0 1.5 2.0 2.5 3.00.0

Observed AOD0.0 0.5 1.0 1.5 2.0 2.5 3.0

0.00.0 0.5 1.0 1.5 2.0 2.5 3.0

0.0 00.0 0.5 1.0 1.5 2.0

0.0

Observed AOD0.0 0.5 1.0 1.5 2.0

0.00.0 0.5 1.0 1.5 2.0

0.0 0

Xin et al., AE, 2011

0.0 0.5 1.0 1.5 2.0 2.5 3.00.0

Observed AOD0.0 0.5 1.0 1.5 2.0 2.5 3.0

0.00.0 0.5 1.0 1.5 2.0 2.5 3.0

0.0 0

Comparison with the revised MODIS AOD

3.3 The accuracy of MODIS C4, C5, C6 products

65% of C6 products filling in the NASA errors over China.But C6 products underestimated AOD by 20% (DB and theby 20% (DB and the Merge) and 14% (DT).In the northeast, the northwestern desert

The correlation between MODIS-C6 AOD and observed AODnorthwestern desert and the Tibet Plateau, the accuracy of C6 DT was higher than C4 and C5 DT products.C6 DB was superior to DT and Merge in the

th t th B ijinortheast, the Beijing-Tianjin-Hebei and the Yangtze River delta. But DT and Merge

The rates of C6 products filling in the NASA errors

gwere more superior to DB in other regions.

3.4 The accuracy of VIIRS IP AOD over east China

BJC Beijig

The Bohai Sea

Sea

Yucheng

Jiaozhou Bay

The

Yello

w

Spring Summer Autumn Winterp g

Meng and Xin et al., RS, 2015, submitted

3.5 The accuracy of Chinese FY-3A and GF-1 AOD

1 6

2.0

0 8

1.2

1.6

演AO

D

0 0

0.4

0.8

GF-1

反 N=7R=0.81SD=0.38y=0.59+0.86*x

0.0 0.4 0.8 1.2 1.6 2.00.0

地基观测 AOD

王中挺, 辛金元等, 遥感学报, 2015.

3.6 The relationship between PM3.6 The relationship between PM2.52.5 and AODand AOD

There were certain correlations between PM2.5 and AOD.

Th l ti diff t t diff t i d th f The correlations are different at different regions and among the fourseasons due to the aerosol types, components and weather conditions.

200

200

200

200

100

150

PM2.5; Daily mean in the dry season y=34.28x+40.54 (R2=0.29)

M2.

5 (g/

m3 )

1 21.41.61.82.0

100

150

M2.

5 (g/

m3 )

100

150

Daily means at Xinglong y=60.83x+15.45 (R2=0.47)

M2.

5 (g/

m3 )

100

150

Daily means at Beijing City y=56.21x+23.31 (R2=0.46)

M2.

5 (g/

m3 )

Dunhuang Desert Xinglong Mt. Beijing City Dinghu Mt.

0.0 0.5 1.0 1.5 2.00

50

PM

Observed AOD

00.20.40.60.81.01.2

0.0 0.5 1.0 1.5 2.00

50 Daily means at Dunhuang y=87.00x+32.18 (R2=0.40)

PM

Observed AOD0.0 0.5 1.0 1.5 2.0

0

50

PM

Observed AOD0.0 0.5 1.0 1.5 2.0

0

50

PM

Observed AOD

Dunhuang Desert Xinglong Mt. Beijing City Dinghu Mt.

150

200(a) Daily mean in spring

y = 99.72x + 4.48 (R2 = 0.60, N=109)

m3 )

150

200 Daily mean in winter y = 72.85x + 10.44 (R2 = 0.34, N=108)

m3 )

(d)

150

200 Daily mean in autumn y = 138.00x +1.10 (R2 = 63, N=116)

m3 )

(c)

150

200(b) Daily mean in summer

y = 92.10x + 7.61 (R2 = 55, N=71)

m3 )

0.0 0.5 1.0 1.50

50

100

PM2.

5 (g/

m

0.0 0.5 1.0 1.50

50

100

PM2.

5 (g/

m

-1.0-0.500.51.01.52.02.53.0

Spring Summer WinterSpring0.0 0.5 1.0 1.5

0

50

100

PM2.

5 (g/

m

0.0 0.5 1.0 1.50

50

100

PM2.

5 (g/

m

Beijing Forest background station in North China

Xin et al., Atmospheric Research, 2014

Observed AOD Observed AODObserved AODObserved AOD

An example: PM2.5 retrieved from the satelliteAn example: PM2.5 retrieved from the satellite

100

150

200 Daily mean in spring y = 96.06x + 11.02 (R2 = 0.60)

g/m

3 )

3 0

(a)

100

150

200 Daily mean in summer y = 91.31x + 16.32 (R2 = 0.56)

g/m

3 )

3 0

(b)

Beijing Forest background station

0.0 0.5 1.0 1.50

50

100

PM2.

5 (g

MODIS AOD

-1.0-0.500.51.01.52.02.53.0

0.0 0.5 1.0 1.50

50

100

PM2.

5 (g

MODIS AOD

-1.0-0.500.51.01.52.02.53.0

200200

Spring SummerMODIS AOD MODIS AOD

100

150

Daily mean in winter y = 26.95x + 14.71 (R2 = 0.01)

M2.

5 (g/

m3 )

1 52.02.53.0

(d)

100

150

Daily mean in autumn y = 161.37x + 7.80 (R2 = 0.57)

M2.

5 (g/

m3 )

1 52.02.53.0

(c)

The annual PM2.5 on sunny days (2009-2011): 40-55 μg/m3

0.0 0.5 1.0 1.50

50

PM

MODIS AOD

-1.0-0.500.51.01.5

0.0 0.5 1.0 1.50

50

PM

MODIS AOD

-1.0-0.500.51.01.5

The relationships between PM2.5 and MODIS AOD

Autumn Winter

The annual PM2 5 (2001 2006):

days (2009 2011): 40 55 μg/mp

The annual PM2.5 (2001-2006): 60-80 μg/m3

There may be large error, if ith t th l l i lwithout the local revisal.

Xin et al., Atmospheric Research, 2014

Donkelaar et al., 2010

4. Summary4. Summary1. Since 2004, we began a network to investigate the spatial and

t l di t ib ti f l ti l ti i Chitemporal distribution of aerosol optical properties in China, and try to track the trends.

2. The network has been used to revise the satellite products, aerosol model results, and to obtain the relationship between PM2 5 d AODPM2.5 and AOD.

3. The accuracy of MODIS AOD increased by 20% from C4 to C6 products. 65% of C6 product filled in the NASA errors over China, but the product underestimated AOD in China.

4. 60% of VIIRS IP products filled in the NASA errors over east China. Besides, Chinese FY and GF satellites’ AOD products


showed large errors in the region.

Thank youThank you！！Thank youThank you！！

Any question?Any question?

Dr. Dr. JinyuanJinyuan XinXin; [email protected] ; [email protected]

LAPC & CERNLAPC & CERN SCAS IAPSCAS IAP CASCASLAPC & CERNLAPC & CERN--SCAS, IAPSCAS, IAP--CASCAS

CV of Dr. Jinyuan Xin

Dr. JINYUAN XIN was born in 1975. He is an associate professor and amaster tutor of the Institute of Atmospheric Physics, Chinese Academy ofSciences. He obtained the bachelor and master degrees in atmosphericphysics and atmospheric environment, the doctorate degrees in meteorologyin Lanzhou University in 1999, 2002 and 2007, respectively. The researchy , , p yfield is in atmospheric physics and atmospheric environment. He has a deepresearch in the field of aerosol optical properties and air pollution. He hastaken on 7 projects including two NSFC projects and 5 sub-projects fromtaken on 7 projects, including two NSFC projects, and 5 sub-projects fromthe 863 important plan, the key project of knowledge innovation of CAS, twostrategic special projects of CAS and the 973 project.He won six Chinese academic awards: the National Excellent Youth ScholarHe won six Chinese academic awards: the National Excellent Youth Scholarof NSFC in 2012, the "QIUSHI" Award in 2005, the "XUE DU FENGZHENG" Award in 2006, the "National Hundred Outstanding PhD Thesis"N i i A d i 2009 h Y h M l i l S i dNomination Award in 2009, the Youth Meteorological Science andTechnology Excellent Paper in 2014, Youth Meteorological Science andTechnology Award in 2014. He has published 70s academic papers (40s SCI


papers), which have been cited more than 1000 times.

500

600 Dailymean y=88.26x+44.07 (R2=0.63)

0.00.2

400

500

)

Dailymeany=123.42x+25.57 (R2=0.57)

0.00.2

240

300 Dailymeany=35.88x+29.96 (R2=0.55)

0.00.2

200

300

400

*PM

2.5(g

/m3 ) y=-298exp(-x/1.69)+320 (R2=0.67) 0.4

0.60.81.01.21 4

200

300

00

)*PM

2.5(g

/m3 ) y 3. 5.57 ( 0.57)

0.40.60.81.01.21 4

120

180

M 2

.5 (

g/m

3 )

y 35.88 9.96 ( 0.55)0.40.60.81.01.21 4

0 1 2 3 4 50

100

AOD *1000 / H( m-1 )

f(R

H)* 1.4

1.61.8

0.0 0.4 0.8 1.2 1.6 2.00

100

f(R

H)

AOD ( 500 nm)

1.41.61.8

0 1 2 3 4 50

60PM

AOD *1000 / H( m-1 )

1.41.61.8

600

800

1000

g/m

3 )

Dailymean y=202.58x+68.35 (R2=0.47)

0.00.20.40 6 300

400

5003 )

Dailymean y=55.08x+76.06 (R2=0.45)

0.00.20.40 6

( )

900

1200 Dailymean y=152.37x+117.52 (R2=0.56) y=-527exp(-x/1.65)+603 (R2=0.60)

0.00.20.40 6g/

m3 )

( )

200

400

600

(RH

)*PM

10(

g 0.60.81.01.21.41.6 100

200

300

PM 1

0 (

g/m

3 0.60.81.01.21.41.6

300

6000.60.81.01.21.41.6f(

RH

)*PM

10(

g

0.0 0.4 0.8 1.2 1.6 2.00

AOD ( 500 nm)

f( 1.8

0 1 2 3 4 50

AOD *1000 / H( m-1 )

1.8

0 1 2 3 4 50

AOD *1000 / H( m-1 )

1.8f林海峰, 辛金元*等, 环境科学, 2013.

the accuracy of aerosol satellites' aod products over

Documents