jae kim 1, s. m. kim 1, and mike newchurch 2 1. pusan national university, korea 2. university of...
TRANSCRIPT
Jae Kim1, S. M. Kim1, and Mike Newchurch2
1.Pusan National University, Korea2.University of Alabama in Huntsville, USA
The analyses and intercomparison of
satellite-derived HCHO measurements
with statistical approaches
AURA Science Workshop, 14- 18 September, Leiden, Netherland
Global climate change is currently the biggest issue.
Palmer et al., 2003; 2006Because global temperature has been
increased, isoprene from biogenic activity must be increased. Expect to see an increasing tendency in HCHO trend.
Motivation
.
Data
Satellite HCHO data
Period
GOMEApril 1996 - June
2003
SCIAMACHYJan 2003-Dec
2007
OMIOct 2004-Dec
2008
HCHO trend on tropical rain forests
Amazon
Africa rain forest
Western Pacific
GOME overPacific withall time
GOME overPacific tillYear 2001
6.9%/year
1.2%/year
GOME P2
SCIAMACHY
OMI
Africarainforest
0.7%/year
0.5%/year
7.5%/year
GOME P2
SCIAMACHY
OMI
Amazon
-2.3%/year
1.7%/year
7.9%/year
GOME P2
SCIAMACHY
OMI
Western Pacific
-1.0%/year
1.0%/year
10.1%/year
GOME P2
SCIAMACHY
OMI
Central Pacific1.2%/year
0.0%/year
7.0%/year
/yearPerio
d
Amazon[75W-50W,
10S-5N]
Africa[15W-25E
,5S-10N]
WesternPacific[90E-150E,
10S-10N]
Central Pacific[180W-160W ,
10S-10N]
GOME P2Apr96-Dec01
-2.3% 0.7% -1.0% 1.2%
SCIAMACHY
Jan03-Dec07
1.7% 0.5% 1.00.0%
OMIOct04-Dec08
7.9% 7.5% 10.1% 7.0%
HCHO Trend analyses
Satellite data have an intrinsic problem, ill-posed problem, that comes from the fact that a number of various physical parameters can have a similar effect on measured radiance.
Most of the previous evaluations of satellite performance have relied on point-by-point comparisons with limited spatial and temporal coverage of in-situ measurements
The levels of agreement from these comparisons vary according to location and season, so there is not a clear superior method for various satellite tropospheric gas products.
Difficulty in satellite measurement validation comes from large uncertainties, especially HCHO vertical columns, whose error typically range from 40-105% [Palmer, et al., 2006; Kurosu, et al., 2008]. Inter-comparison between satellites HCHO measurements are challenging
Validation of satellite HCHO
.
Our approach is to validate the satellite measurements by analyzing spatial and temporal coherence between individual satellite products and a known source data set
MOPITT CO
A promising statistical tools for identifying these coupled relationships with spatial-temporal patterns are
individual parameters is Empirical Orthogonal Function (EOF)
combinations of two parameters, Singular Value Decomposition (SVD)
Power Spectrum analyses for cycle of the data sets
Statistical tools for validation
Tropical areas with biomass burning and biogenic activity in rain forests
South America
Africa
Western Pacific
EOF and SVD analyses of GOME, SCIAMACHY, and OMI HCHO measurements in conjunction with MOPITT CO.
Data periods
Data
Sensor Period
GOMEApril 1996 - June
2003
SCIAMACHYJan 2003-Dec
2007
OMIOct 2004-Dec
2008
MOPITT COMarch 2000-Dec
2008
EOFMode1
GOME P1HCHO
SCIAMACHYHCHO
OMIHCHO
MOPITTCO
sudden increasing tendency
January
September
Red: +Blue: -
AfricaAfrica
GOME HCHO
SCIAMACHYHCHO
OMIHCHO
MOPITTCO
Power Spectrum analysis
GOME P1HCHO
SCIAMACHYHCHO
OMIHCHO
MOPITTCOMay
September
GOME HCHO
SCIAMACHYHCHO
OMIHCHO
MOPITTCO
AmazonAmazon
GOME P1 HCHO
SCIAMACHYHCHO
OMIHCHO
MOPITTCO
March
October
GOME HCHO
SCIAMACHYHCHO
OMIHCHO
MOPITTCO
Central PacificCentral Pacific
western Pacificwestern Pacific
HCHO-CO SVD analysis
data Period
SCIAMACHY HCHO
Jan 2003-Dec 2007
OMI HCHOOct 2004-Dec
2008
MOPITT COMarch 2000-Dec
2008
SVD 1st mode of MOPITT CO and SCIAMACHY HCHO
August
February
SVD 1st mode of MOPITT CO and OMI HCHO
August
January
SVD 1st mode of MOPITT CO and SCIAMACHY HCHO
September
February
SVD 1st mode of MOPITT CO and OMI HCHO
September
January
SVD 1st mode of MOPITT CO and SCIAMACHY HCHO
SVD 1st mode of MOPITT CO and OMI HCHO
1. EOF analyses shows spatial and temporal distribution of GOME, SCIAMACHY HCHO, MOPITT CO match each other. However, OMI HCHO shows different spatial and temporal pattern compared with others.
2. SVD analyses shows GOME HCHO-MOPITT CO, SCIAMACHY HCHO – MOPITTCO shows consistent spatial and temporal coherence
3. However, OMI HCHO – MOPITT CO shows relatively low correlation.
Relationship between GOME (SCIAMACHY) HCHO and CO shows that biomass burning is most likely the major source of HCHO over Africa and South America.
However, relationship between OMI HCHO and CO suggests biomass burning is not as significant source as of HCHO.
4. GOME and SCIAMACHY HCHO trend is marginal, but OMI HCHO trend is as high as 10% climate change can not explain the big increase. It could be due to OMI instrument or calibration error.
5. EOF and SVD analyses can be another useful method for satellite data validation.
Conclusions