Post on 04-Apr-2020
transcript
Evaluation of LTDR dataat
AERONET sites(For Year 1999)
Jyothy Nagol
TOA AVHRR(9 pixel matrix)
Atmosphericdata from
AERONET
Atmospheric correction
(6s)
REFERENCE DATA(At surface NDVI)
Comparison withLTDR, PAL and GIMMS
Reference Data
Reference Data: Comprehensive atmosphericcorrection of 9 pixels around AERONET sites.
Comparison of LTDR and other NDVI datasetswith reference data
Sites: 48 sites
Data points: 580
Comparison with Reference data
Daily data comparisons (all data)– TOA, PAL, LTDR-beta_release(Dec_06)
15 day composite comparison (site)– TOA, PAL, LTDR and GIMMS
Monthly composite comparison (site)– LTDR, TOA, GVI
NOTE: ALL data is constrained to View zenith angle < 42
Daily data comparisonsfor
LTDR, PAL, and TOA
0
0.2
0.4
0.6
0.8
1
0 0.2 0.4 0.6 0.8 1
Corrected Rho1
PA
L R
ho
1 a
nd
LT
DR
Rh
o1
1 to 1 line
PAL_rho1
LTDR_rho1
Linear (PAL_rho1)
Linear (LTDR_rho1)
0
0.2
0.4
0.6
0.8
1
0 0.2 0.4 0.6 0.8 1
Corrected Rho2
PA
L R
ho
1 a
nd
LT
DR
Rh
o2
1 to 1 linePAL_rho2
LTDR_rho2
Linear (PAL_rho2)
Linear (LTDR_rho2)
BAND 1
Under estimation
NDVI is more sensitive to Red.
Different Calibration: PAL: Stable desert target vicarious calibration (Rao and Chen, 1996)
LTDR: ocean-cloud vicarious calibration (Vermote and Kaufman, 1995)
BAND 2
PAL
and
LTD
R R
ho 2
PAL
and
LTD
R R
ho 1
NDVI
-0.2
0
0.2
0.4
0.6
0.8
1
-0.2 0 0.2 0.4 0.6 0.8 1
Reference Data
TO
A,
PA
L,
an
d L
TD
R d
ail
y d
ata
TOA_NDVI
PAL_NDVI
LTDR_NDVI
1 to 1 line
-0.2
0
0.2
0.4
0.6
0.8
1
-0.2 0 0.2 0.4 0.6 0.8 1
Reference Data
TO
A,
PA
L,
an
d L
TD
R d
ail
y d
ata
TOA_NDVI
PAL_NDVI
LTDR_NDVI
1 to 1 lineLinear (TOA_NDVI)
Linear (PAL_NDVI)
Linear (LTDR_NDVI)
0.09150.031210.0645LTDR0.10840.07070.0889PAL0.19740.08090.1432TOA
RMSE(> 0.3 NDVI)
RMSE(< 0.3 NDVI)RMSE
15 day compositefor
LTDR, PAL and GIMMS
0
0.1
0.2
0.3
0.4
0 50 100 150 200 250 300 350
Day of the Year
ND
VI
Da
ta
TOA
PAL
GIMMS
LTDR
Reference
NDVI
0
0.05
0.1
0.15
0.2
0 0.05 0.1 0.15 0.2
Reference Data
TO
A,
PA
L,
an
d L
TD
R d
aily
da
ta
TOA_NDVI
PAL_NDVI
LTDR_NDVIDesert site
Solar village, Saudi Arabia
0
0.1
0.2
0.3
0.4
0 50 100 150 200 250 300 350
Day of the Year
ND
VI D
ata
GIMMS
PAL
LTDR
Reference NDVI
Semiarid SiteSevilleta, USA
0
0.1
0.2
0.3
0 0.1 0.2 0.3
Reference Data
TO
A, P
AL
, an
d L
TD
R d
aily d
ata
toa_NDVI
PAL_NDVI
LTDR_NDVI
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 50 100 150 200 250 300 350
Day of the Year
ND
VI D
ata
GIM M S
PAL
LTDR
Reference NDVI
1
Agricultural SiteBondville, USA
0
0.2
0.4
0.6
0.8
0 0.2 0.4 0.6 0.8
Reference Data
TO
A, P
AL, and L
TD
R d
aily d
ata
toa_NDVI
PAL_NDVI
LTDR_NDVI
Monthly Compositefor
LTDR and GVI
Desert siteSolar village, Saudi Arabia
0
0.1
0.2
0.3
0.4
0 50 100 150 200 250 300 350
Day of the Year
ND
VI
Da
ta
GVI
LTDR
Reference NDVI
Agricultural SiteBondville, USA
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 50 100 150 200 250 300 350
Day of the Year
ND
VI D
ata
TOA
GVI
LTDR
Reference NDVI
Conclusions from Comparison withReference Data
LTDR dataset is closest to reference dataPAL underestimates both band1 and band2GIMMS data overestimates in high NDVI and
underestimates in low NDVI,– Possibly due to use of statistical orbital drift correction.
GIMMS shows abnormal flatness in desert sitesbecause it has been normalized to be flat there.
GVI reduces the amplitude of phenology– Possibly due to use of averaging in compositing system
This method of evaluation gives us a chance toisolate atmospheric issues.
LimitationsAssumes spherical aerosol modelThe sample is not representative of all biomesVery few operational AERONET sites in 1999.Only 580 points
– Few operational AERONET sites,– non-operational Instruments (data gaps),– Clouds.
Future plansExpand this analysis to other years and include
other sensors in LTDR product.
Thank you
Jyothy Nagol