Session 7 : Geophysical Trends Derived from 12 years of AIRS Infrared Radiances
S. De Souza-Machado1, L. L. Strow1,2, S. Buczkowski1
1
Joint Center for Earth System Technology, Univ. of Maryland Baltimore County 2Dept. of Physics, Univ. of Maryland Baltimore County
1 AIRS instrument
4. Example Zonal All-Sky Rates: North Polar
AIRS: Grating spectrometer sounder on EOS-AQUA, 2378 channels with spectral
resolution ∼0.5-2.0 cm−1 from 650-2665 cm−1. 1:30 am/pm orbit. AIRS
operational for 13+ years, intensively validated.
ñ AIRS mission length entering climate scales
ñ AIRS stability: ∼ 0.002 K/year (from CO2 , SST as truth)
ñ Clear path to connect CrIS to AIRS radiometrically (for long term climate studies)
ñ Climate products need error traceability
ñ Climate products need to be reproduced by others
ñ
0.2
7. T(z) and WV(z) All-Sky rate retrievals
T(z) rate from UMBC-AIRS
100
80
0.15
∆ B(T) in K/year
0.15
10
40
0.1
ERA*AK d(T)/dt K/yr
d(T)/dt K/yr
60
20
0.05
-20
0.1
0.1
0.05
0.05
0
-40
0
100
800
-100
-200
1000 1200 1400 1600 1800 2000 2200 2400 2600
-150
-100
-50
0
50
100
150
2 Approach to Climate Level Products
1000
1000
−50
0
50
T(z) rate from MERRA * avg kernel
5. (L) Global and (R) Hemispherical Rates
50
T(z) rate from AIRS L3
0.15
10
0.04
0.06
0.04
0.1
0.1
0.05
0.05
0.02
0
∆ B(T) in K
∆ B(T) in K/year
0.02
0
-0.02
0
100
SH
NH
Std NH
Std SH
-0.02
-0.06
800
-0.08
1000 1200 1400 1600 1800 2000 2200 2400 2600
1000
800
1000 1200 1400 1600 1800 2000 2200 2400 2600
-1
-1
Wavenumber (cm )
Wavenumber (cm )
ñ
CO2, N2O and CH4 increases dominant.
ñ 0.01K/year in window region, 2-σ error
detection
ñ Maybe slightly less water.
ñ
N. and S. Hemisphere rates very
similar! (land masses quite different)
ñ Did not get similar ERA rates (using
cloud fields) .. cloud problems?
ñ
−50
0
50
WV(z) rate from UMBC-AIRS
Climate community mostly interested in trends (and) Anomalies
ñ Single jacobians still used for a linear retrieval
ñ Anomalies rates. Linear retrieval OK (except ENSO??)
10
3
10
0.02
0.02
0.01
0.01
0
0
100
−0.01
−0.01
−0.02
1000
0
3
2
−50
0
50
WV(z) rate from MERRA * avg kernel
−0.02
1000
−50
0
10
-1
2
101
1
L3 d(W)/dt frac/yr
10
-1
2
-2
-3
3
10
200
2004
2006
2008
2010
2012
0.1
0.05
400
500
0
600
700
-0.05
800
0.02
0.01
0.01
0
100
2004
2006
2008
2010
2012
0
100
−0.01
−0.01
2014
0.1
300
Pressure (mbar)
Pressure (mbar)
300
0.02
-3
3
10
200
2014
10
10
0
-2
Days Latitudes Channels Variables
4480
40
2378 B(ν,T) observed
B(ν,T) ERA All-Sky
B(ν,T) ERA Clear
50
2
Pressure (mbar)
Pressure (mbar)
1
0
WV(z) rate from AIRS L3
MERRA*AK d(W)/dt frac/yr
101
50
ERA*AK d(W)/dt frac/yr
100
ERA
UMBC
0
WV(z) rate from ERA * avg kernel
d(W)/dt frac/yr
3 Data Set
ñ
−50
UMBC-AIRS (from radiance rates) and AIRS L3 (from rates of average AIRS L3
products) both see significant stratospheric cooling in Southern Hemisphere
include 0.002 K/yr instr drift → UMBC retrieved dT/dt uncertainty ≤ 0.018 K/yr
6. Example Temp (top) and Water (bottom) Anomaly -3 S to 0 S
0
−0.1
1000
10
10
−0.05
−0.1
-0.04
-0.06
0
100
−0.05
-0.04
-0.08
−0.02
0.05
400
500
0
600
700
-0.05
1000
−50
0
50
−0.02
1000
−50
0
50
AIRS L3 water vapor product cuts off at 100 mb. UMBC-AIRS rates are more
similar to re-analysis (ERA and MERRA) than AIRS L3
800
-0.1
900
-0.1
900
2004
http://asl.umbc.edu
0
L3 d(T)/dt K/yr
0.15
10
This Work: A First Look
Anomalies generally quite Gaussian except for poles.
−50
MERRA*AK d(T)/dt K/yr
ñ
Data Set Sizes
−0.1
−0.1
200
Wavenumber (cm )
AIRS V7 random data (implemented by UMBC)
ñ ∼ 1-2% of data, using only 3x3 FORs next to nadir
ñ Zonal averages (for now): 40 equal area latitude bins
ñ Each channel time series fit de-seasonalized by bi-weekly average, after which
linear rate and anomaly found.
ñ Matched to ERA for each scene (FOV) (for SARTA TwoSlab Cloud Calcs)
ñ No L1c or frequency corrections!
−0.05
−0.05
-80
-1
ñ
0
100
-60
-0.05
Decrease data volume: Random subset
ñ Average data (L1b gridding and binning)
ñ This leads to constant reprocessing by anyone!
ñ Adopt OEM framework with scattering RTA
ñ Decrease sensitivity to unknown variables by producing L3 trends and anomalies
from time derivatives of L1 radiances.
ñ This is not a replacement for 3x3 or single-FOV retrievals.
0.15
10
0
-0.1
No radiance binning, average it all.
ñ Examine error characteristics
ñ Examine ability to do anomaly retrievals
ñ It takes a few minutes or so to create L3 trend and anomaly
retrievals for the whole mission!
T(z) rate from ERA * avg kernel
2006
2008
2010
2012
2014
2004
2006
2008
2010
2012
UMBC retrieved dWV(frac)/dt uncertainty ≤ 0.005 /yr (includes instr drift)
2014
sergio@umbc.edu