Embed Size (px)
Citation preview
Climate Data Records: A MODIS Perspective
S. Platnick1, N. Amarasinghe2, and the MODIS Atmosphere Team
1 NASA GSFC, 2 SSAI
Workshop on Observations & Modeling of Aerosol/Cloud Properties
for Climate Studies Paris, 12-14 Sept 2011
https://ntrs.nasa.gov/search.jsp?R=20120002318 2020-07-06T19:04:49+00:00Z
Outline: ▼ Global Analyses
▶ Trends ▶ ENSO Correlations
▶ Some Lessons-Learned
Platnick, Paris Wkshop, 12 Sept 2011
Outline: ▼ Global Analyses
▼ Trends ▶ Time-to-Detection vs. grid size
▶ Example trends, instrument artifacts
▶ ENSO Correlations ▶ Some Lessons-Learned
Platnick, Paris Wkshop, 12 Sept 2011
Motivation ! Trends
– For observed MODIS temporal variability on regional scales, what is the expected “time to detection” for a given trend?
– Consistency between Terra and Aqua MODIS? Lack of consistency traced to instrument differences? Statistically significant regional trends found?
! Sensitivity of retrievals to interannual (low frequency) climate variability, e.g., ENSO – Correlation of atmosphere properties to ENSO useful for climate model
evaluation (e.g., GFDL AM3 cloud fields) – To what extent can ENSO responses alias into trend observations?
! Datasets and Analysis – Monthly mean anomalies derived from archived MODIS Atmosphere
Team Terra and Aqua monthly Level-3 aggregations – Analyses for various equal-angle grids, from 1° (native Level-3 resolution)
to regional and zonal – All analysis includes effects of data set temporal correlations (“order 1”
autocorrelation for trend calculations)
Platnick, Paris Wkshop, 12 Sept 2011
0 5 10 15 20!y
0
2
4
6
8
10
% t
ren
d/d
ec
5 10
10
20
20
30
30
40
40
50
50
n*(yrs) required to detect the trend
Weatherhead Equation (constant = 3.3)
example : n* = 16.3 yrs, with ! = 10%, and "=5%/dec
Number of Years Required to Detect a Trend (90% prob. of detecting a trend to a 0.05 statistical level, no autocorrelation:
Tiao et al., 1990; Weatherhead et al., 1998) Tr
end/
deca
de (%
)
~17yrs
Variability in monthly residual time series !y/⟨y⟩ (%) ^
Platnick, Paris Wkshop, 12 Sept 2011
Time Required for Detection of 5%/decade Trend (90% prob. of detecting a 0.05 statistical significance,
from 10 yr monthly anomaly variability w/autocorrelation correction)
0 40 yrs 20 30 10
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
[ N_STAR (omega = 5%/decade ]
Cloud_Optical_Thickness_Liquid N* (! = 5%/dec)
Cloud_Optical_Thickness_Liquid N* (! = 5%/dec)
Time Series, Monthly :JUL, 2000 to JUN, 2010
5% significance level
smoothing(NO), autocorrelaion(YES)
10° bins
Platnick, Paris Wkshop, 12 Sept 2011
Time Required for Detection of 5%/decade Trend (90% prob. of detecting a 0.05 statistical significance,
from 10 yr monthly anomaly variability w/autocorrelation correction)
0 40 yrs 20 30 10
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
[ N_STAR (omega = 5%/decade ]
Cloud_Optical_Thickness_Liquid N* (! = 5%/dec)
Cloud_Optical_Thickness_Liquid N* (! = 5%/dec)
Time Series, Monthly :JUL, 2000 to JUN, 2010
5% significance level
smoothing(NO), autocorrelaion(YES)
20° bins
Platnick, Paris Wkshop, 12 Sept 2011
Time Required for Detection of 5%/decade Trend (90% prob. of detecting a 0.05 statistical significance,
from 10 yr monthly anomaly variability w/autocorrelation correction)
0 40 yrs 20 30 10
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
[ N_STAR (omega = 5%/decade ]
Cloud_Optical_Thickness_Liquid N* (! = 5%/dec)
Cloud_Optical_Thickness_Liquid N* (! = 5%/dec)
Time Series, Monthly :JUL, 2000 to JUN, 2010
5% significance level
smoothing(NO), autocorrelaion(YES)
30° bins
Platnick, Paris Wkshop, 12 Sept 2011
Time Required for Detection of 5%/decade Trend (90% prob. of detecting a 0.05 statistical significance,
from 10 yr monthly anomaly variability w/autocorrelation correction)
0 40 yrs 20 30 10
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
[ N_STAR (omega = 5%/decade ]
Cloud_Fraction_Day N* (! = 5%/dec)
Cloud_Fraction_Day N* (! = 5%/dec)
Time Series, Monthly :JUL, 2000 to JUN, 2010
5% significance level
smoothing(NO), autocorrelaion(YES)
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
[ N_STAR (omega = 5%/decade ]
Cloud_Optical_Thickness_Liquid N* (! = 5%/dec)
Cloud_Optical_Thickness_Liquid N* (! = 5%/dec)
Time Series, Monthly :JUL, 2000 to JUN, 2010
5% significance level
smoothing(NO), autocorrelaion(YES)
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
[ N_STAR (omega = 5%/decade ]
Cloud_Optical_Thickness_Liquid N* (! = 5%/dec)
Cloud_Optical_Thickness_Liquid N* (! = 5%/dec)
Time Series, Monthly :JUL, 2000 to JUN, 2010
5% significance level
smoothing(NO), autocorrelaion(YES)
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
[ N_STAR (omega = 5%/decade ]
Cloud_Optical_Thickness_Liquid N* (! = 5%/dec)
Cloud_Optical_Thickness_Liquid N* (! = 5%/dec)
Time Series, Monthly :JUL, 2000 to JUN, 2010
5% significance level
smoothing(NO), autocorrelaion(YES)
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
0 90E 180 90W 0
-90
-45
0
45
90
0
10
20
30
40
[ N_STAR (omega = 5%/decade ]
AOD_Land_And_Ocean N* (! = 5%/dec)
AOD_Land_And_Ocean N* (! = 5%/dec)
Time Series, Yearly :JUL, 2000 to JUN, 2010
5% significance level
30° bins
Platnick, Paris Wkshop, 12 Sept 2011
0 90E 180 90W 0
-90
-45
0
45
90
-20.0
-10.0
0.0
10.0
20.0
0 90E 180 90W 0
-90
-45
0
45
90
-20.0
-10.0
0.0
10.0
20.0
[ %b0 (per decade) ]
TERRA CF (C51) anomaly (CLOUD MASK) TREND (TOTAL)
Time Series, Monthly :JUL, 2000 to JUN, 2010
5% significance level
0 1.0 0.5
Cloud Fraction from MODIS mask, Terra
(10° binning, daytime observations only)
!20 "20 0
Annual Mean Fraction (July 2000 – June 2001)
Cloud Fraction Trends (monthly anomalies, July 2000 – June 2010)
%/decade
Trends Masked by Significance Level ! 0.05 (w/autocorrelation correction in t-test)
0 90E 180 90W 0
-90
-45
0
45
90
0.0
0.5
1.0
0 90E 180 90W 0
-90
-45
0
45
90
0.0
0.5
1.0
[ Annual mean ]
TERRA CF (C51) CLOUD MASK
JUL, 2000 to JUN, 2001
0 90E 180 90W 0
-90
-45
0
45
90
-20.0
-10.0
0.0
10.0
20.0
0 90E 180 90W 0
-90
-45
0
45
90
-20.0
-10.0
0.0
10.0
20.0
[ %b0 (per decade) ]
TERRA CF (C51) anomaly (CLOUD MASK) TREND (TOTAL)
Time Series, Monthly :JUL, 2000 to JUN, 2010
5% significance level ~8% of grid boxes Platnick, Paris Wkshop, 12 Sept 2011
MODIS Cloud Optical Thickness Trends, Aqua vs. Terra (water clouds, 10° binning)
Platnick, Paris Wkshop, 12 Sept 2011
Trends Masked by Significance Level ! 0.05 (w/autocorrelation correction in t-test)
0 90E 180 90W 0
-90
-45
0
45
90
-20.0
-10.0
0.0
10.0
20.0
0 90E 180 90W 0
-90
-45
0
45
90
-20.0
-10.0
0.0
10.0
20.0
[ %b0 (per decade) ]
TERRA ! (C51) anomaly (LIQUID) TREND (TOTAL)
Time Series, Monthly :JUL, 2000 to JUN, 2010
5% significance level
!20 "20 0
%/decade
Terra (monthly anomalies, July 2000 – June 2010)
0 90E 180 90W 0
-90
-45
0
45
90
-20.0
-10.0
0.0
10.0
20.0
0 90E 180 90W 0
-90
-45
0
45
90
-20.0
-10.0
0.0
10.0
20.0
[ %b0 (per decade) ]
AQUA ! (C51) anomaly (LIQUID) TREND (TOTAL)
Time Series, Monthly :JUL, 2002 to JUN, 2010
5% significance level
Aqua (monthly anomalies, July 2002 – June 2010)
Outline: ▼ Global Analyses
▶ Trends ▼ ENSO Correlations
▶ Aliasing into trends?
▶ Some Lessons-Learned
Platnick, Paris Wkshop, 12 Sept 2011
ENSO3.4 SST Anomaly Index (avg. temperature in a box in east-central equatorial Pacific)
Aqua
Terra
Platnick, Paris Wkshop, 12 Sept 2011
0 90E 180 90W 0-90
-45
0
45
90
-6.0
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
0 90E 180 90W 0-90
-45
0
45
90
-6.0
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
[ LAG (months) ]
AQUACF-HIGH (C51) anomaly(CLOUD MASK) and Ninio 3.4 ENSO index
For extreme lag correlation
Time Series: July, 2002 - January, 2011
1% significance level
5% significance level
smoothing(NO), autocorrelaion(YES)
Lag (months) Modified Chen et al., 2007
[red => cloud response lags E3.4 index; blue => cloud response precedes index]
0 90E 180 90W 0-90
-45
0
45
90
-0.8
-0.5
-0.3
0.0
0.3
0.5
0.8
0 90E 180 90W 0-90
-45
0
45
90
-0.8
-0.5
-0.3
0.0
0.3
0.5
0.8
[ R ]
AQUACF-HIGH (C51) anomaly(CLOUD MASK) and Ninio 3.4 ENSO index
For extreme lag correlation
Time Series: July, 2002 - January, 2011
1% significance level
5% significance level
smoothing(NO), autocorrelaion(YES)
High Cloud Amount Correlation
High Cloud Amount
Example ENSO3.4 vs. MODIS Monthly Anomaly Correlation 1° bins, masked by 1% statistical sig., July 2002–Jan 2011
Platnick, Paris Wkshop, 12 Sept 2011
0 90E 180 90W 0-90
-45
0
45
90
-110.00
-55.00
0.00
55.00
110.00
0 90E 180 90W 0-90
-45
0
45
90
-110.00
-55.00
0.00
55.00
110.00
[ b = !P/!N3.4 (hPa/ °C) ]
AQUA CT PRESSURE (C51) anomaly and Ninio 3.4 ENSO index
For extreme lag correlation
Time Series: July, 2002 - January, 2011
1% significance level
5% significance level
smoothing(NO), autocorrelaion(YES)
0 90E 180 90W 0-90
-45
0
45
90
-0.20
-0.10
0.00
0.10
0.20
0 90E 180 90W 0-90
-45
0
45
90
-0.20
-0.10
0.00
0.10
0.20
[ b = !CF/!N3.4 (per °C) ]
AQUACF-HIGH (C51) anomaly(CLOUD MASK) and Ninio 3.4 ENSO index
For extreme lag correlation
Time Series: July, 2002 - January, 2011
1% significance level
5% significance level
smoothing(NO), autocorrelaion(YES)
High Cloud Amount dfc/dT (K-1)
Cloud-top Pressure dpc/dT (hPa-K-1)
High Cloud Amount and Pressure: Derived Regression Slopes
Example ENSO3.4 vs. MODIS Monthly Anomaly Correlation 1° bins, masked by 1% statistical sig., July 2002–Jan 2011
Platnick, Paris Wkshop, 12 Sept 2011
0 90E 180 90W 0-90
-45
0
45
90
-110.00
-55.00
0.00
55.00
110.00
0 90E 180 90W 0-90
-45
0
45
90
-110.00
-55.00
0.00
55.00
110.00
[ b = !P/!N3.4 (hPa/ °C) ]
AQUA CT PRESSURE (C51) anomaly and Ninio 3.4 ENSO index
For extreme lag correlation
Time Series: July, 2002 - January, 2011
1% significance level
5% significance level
smoothing(NO), autocorrelaion(YES)
0 90E 180 90W 0-90
-45
0
45
90
-0.20
-0.10
0.00
0.10
0.20
0 90E 180 90W 0-90
-45
0
45
90
-0.20
-0.10
0.00
0.10
0.20
[ b = !CF/!N3.4 (per °C) ]
AQUACF-HIGH (C51) anomaly(CLOUD MASK) and Ninio 3.4 ENSO index
For extreme lag correlation
Time Series: July, 2002 - January, 2011
1% significance level
5% significance level
smoothing(NO), autocorrelaion(YES)
High Cloud Amount dfc/dT (K-1)
Cloud-top Pressure dpc/dT (hPa-K-1)
High Cloud Amount and Pressure: Derived Regression Slopes
Example ENSO3.4 vs. MODIS Monthly Anomaly Correlation 1° bins, masked by 1% statistical sig., July 2002–Jan 2011
GFDL AM3 high cloud amount regressed to NINO3 (courtesy Andrew Wittenberg)
±0.24 K-1
Platnick, Paris Wkshop, 12 Sept 2011
ENSO component of trend derived from correlation
regression slope (masked by trend & ENSO3.4
correlation sig. < 0.05)
0 90E 180 90W 0
-90
-45
0
45
90
-50.00
-25.00
0.00
25.00
50.00
0 90E 180 90W 0
-90
-45
0
45
90
-50.00
-25.00
0.00
25.00
50.00
[ %b per decade ]
AQUA CF-HIGH (C51) anomaly (CLOUD MASK) TREND (ENSO COMPONENT)
For extreme lag correlation
Time Series, Monthly :JUL, 2002 to JUN, 2010
1% significance level
5% significance level (trend + r)
smoothing(NO), autocorrelaion(YES)
0 90E 180 90W 0
-90
-45
0
45
90
-50.0
-25.0
0.0
25.0
50.0
0 90E 180 90W 0
-90
-45
0
45
90
-50.0
-25.0
0.0
25.0
50.0
[ %b0 (per decade) ]
AQUA CF-HIGH (C51) anomaly (CLOUD MASK) TREND (TOTAL)
Time Series, Monthly :JUL, 2002 to JUN, 2010
5% significance level
0 90E 180 90W 0
-90
-45
0
45
90
-50.0
-25.0
0.0
25.0
50.0
0 90E 180 90W 0
-90
-45
0
45
90
-50.0
-25.0
0.0
25.0
50.0
[ %b0 (per decade) ]
AQUA CF-HIGH (C51) anomaly (CLOUD MASK) TREND (TOTAL)
Time Series, Monthly :JUL, 2002 to JUN, 2010
5% significance level(trend only)High Cloud Amount
Example ENSO3.4 Component of MODIS Trend (from Monthly Anomalies) Aqua, 1° bins, July 2002–Jun 2010
Platnick, Paris Wkshop, 12 Sept 2011
MODIS Aqua Trend (%/dec) July 2002–Jun 2010
(masked by stat. sig. <0.05)
0 90E 180 90W 0
-90
-45
0
45
90
-50.00
-25.00
0.00
25.00
50.00
0 90E 180 90W 0
-90
-45
0
45
90
-50.00
-25.00
0.00
25.00
50.00
[ %b per decade ]
TERRA CF-HIGH (C51) anomaly (CLOUD MASK) TREND (ENSO COMPONENT)
For extreme lag correlation
Time Series, Monthly :JUL, 2002 to JUN, 2010
1% significance level
5% significance level (trend + r)
smoothing(NO), autocorrelaion(YES)
0 90E 180 90W 0
-90
-45
0
45
90
-50.0
-25.0
0.0
25.0
50.0
0 90E 180 90W 0
-90
-45
0
45
90
-50.0
-25.0
0.0
25.0
50.0
[ %b0 (per decade) ]
TERRA CF-HIGH (C51) anomaly (CLOUD MASK) TREND (TOTAL)
Time Series, Monthly :JUL, 2002 to JUN, 2010
5% significance level(trend only)High Cloud Amount
Example ENSO3.4 Component of MODIS Trend (from Monthly Anomalies) Terra, 1° bins, July 2002–Jun 2010
MODIS Terra Trend (%/dec) July 2002–Jun 2010
(masked by stat. sig. <0.05)
ENSO component of trend derived from correlation
regression slope (masked by trend & ENSO3.4
correlation sig. < 0.05)
Platnick, Paris Wkshop, 12 Sept 2011
0 90E 180 90W 0
-90
-45
0
45
90
-50.00
-25.00
0.00
25.00
50.00
0 90E 180 90W 0
-90
-45
0
45
90
-50.00
-25.00
0.00
25.00
50.00
[ %b per decade ]
TERRA CF-HIGH (C51) anomaly (CLOUD MASK) TREND (ENSO COMPONENT)
For extreme lag correlation
Time Series, Monthly :JUL, 2000 to JUN, 2010
1% significance level
5% significance level (trend + r)
smoothing(NO), autocorrelaion(YES)
0 90E 180 90W 0
-90
-45
0
45
90
-50.0
-25.0
0.0
25.0
50.0
0 90E 180 90W 0
-90
-45
0
45
90
-50.0
-25.0
0.0
25.0
50.0
[ %b0 (per decade) ]
TERRA CF-HIGH (C51) anomaly (CLOUD MASK) TREND (TOTAL)
Time Series, Monthly :JUL, 2000 to JUN, 2010
5% significance level(trend only)
ENSO component of trend derived from correlation
regression slope (masked by trend & ENSO3.4
correlation sig. < 0.05)
High Cloud Amount
MODIS Terra Trend (%/dec) July 2000–Jun 2010
(masked by stat. sig. <0.05)
Example ENSO3.4 Component of MODIS Trend (from Monthly Anomalies) Terra, 1° bins, July 2000–Jun 2010
Platnick, Paris Wkshop, 12 Sept 2011
ENSO3.4 Aliasing Into Trends Number of years for ENSO trend to decay to less than some value?
0 10 20 30 40
–2
–1
0
1
2
years since start of record
ENSO
3.4
trend
(K/d
ecad
e)cumulative by year from January in given year up to 2010
start: 1975start: 1970start: 1965
±0.3K/dec ⇒ 15-30 yrs
Platnick, Paris Wkshop, 12 Sept 2011
Outline: ▼ Global Analyses
▶ Trends ▶ ENSO Correlations
▶ Some Lessons-Learned
Platnick et al., CDR, 2 Aug 2011
Calibration/Validation (L1 and L2)
L2 (pixel, swath) & L3 (gridded statistics)
algorithms
Instrument prelaunch & on-orbit characterization,
L1 algorithms
(re-)processing and test system
Archive, Distribution,
Documentation users
A Coordinated System/Infrastructure of Elements is Required for Production and Sustainability of Climate Records
Ancillary data sets
Platnick, Paris Wkshop, 12 Sept 2011
Thank you!
Platnick, Paris Wkshop, 12 Sept 2011
