1
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
12-Year Trends Derived from Radiances:Intercomparisons to AIRS L3 and ERA
Reanalysis
L. Larrabee Strow, Sergio De Souza-Machado, and StevenBuczkowski
UMBCDepartment of Physics and
Joint Center for Earth Systems Technology
October 13, 2015
2
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Overview
AIRS mission length entering climate scalesAIRS stability: ∼ 0.002 K/year (using CO2 and SST astruth)Clear path to connect CrIS to AIRS radiometricallyWe believe we can convert AIRS radiance to CrIS ILS,building an homogeneous radiance recordCrIS very stable: at least 0.01K/year for first three years
How Can We Impact Climate Community?
Products need error traceability!Products need to be reproducible (by others)Community needs to be trained on utility of theseproducts.
3
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Path to Achieve Goal of Climate Level Products
Decrease data volume: Random subsetAverage data (L1b gridding and binning)This leads to constant reprocessing by anyone!Adopt OEM framework with scattering RTADecrease sensitivity to unknown variables by producingL3 trends and anomalies from time derivatives of L1cradiances.This is not a replacement for 3x3 or single-FOVretrievals.
This Work: A First Look
No radiance binning, average it all.Examine error characteristicsExamine ability to do anomaly retrievalsIt takes a few minutes or so to create L3 trend andanomaly retrievals for the whole mission!
4
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Data Set
AIRS V7 random data (implemented by UMBC)∼ 1-2% of data, using only 3x3 FORs next to nadirZonal averages (for now): 40 equal area latitude binsEach channel time series fit for offset, linear rate, and 4sinusoidal terms (yearly and fractional harmonics).Matched to ERA for each scene (FOV)No L1c or frequency corrections!
Data Set Sizes
Days Latitudes Channels4480 40 2378
Variables
B(T) ObservedB(T) ERA ClearB(T) ERA All-Sky
Anomalies generally quite Gaussian except for poles.
5
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
B(T) Time Series: Arctic Latitudes
2003 2005 2007 2009 2011 2013 2015
B(T
) of 795 c
m-1
(K
)
235
240
245
250
255
260
265
270
275
Biggest change of all latitudes, generally warmer winters (if cloud forcingis constant).
6
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Rates vs Latitude
Wavenumber (cm-1)
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
∆ B
(T)
in K
/year
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
0.08
0.1
-200 -150 -100 -50 0 50 100 150 200-100
-80
-60
-40
-20
0
20
40
60
80
100
Hash near 900 cm−1 is M8 A-only versus B-only drifts.
7
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Rates vs Latitude
Wavenumber (cm-1)
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
∆ B
(T)
in K
/year
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
-200 -150 -100 -50 0 50 100 150 200-100
-80
-60
-40
-20
0
20
40
60
80
100
Note: cold trend in window region.
8
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Rates vs Latitude
Wavenumber (cm-1)
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
∆ B
(T)
in K
/year
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
0.08
-200 -150 -100 -50 0 50 100 150 200-100
-80
-60
-40
-20
0
20
40
60
80
100
3 latitude bins combined, now more land.
9
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Rates vs Latitude
Wavenumber (cm-1)
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
∆ B
(T)
in K
/year
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
-200 -150 -100 -50 0 50 100 150 200-100
-80
-60
-40
-20
0
20
40
60
80
100
Northern temperature latitudes.
10
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Rates vs Latitude
Wavenumber (cm-1)
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
∆ B
(T)
in K
/year
-0.06
-0.05
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
-200 -150 -100 -50 0 50 100 150 200-100
-80
-60
-40
-20
0
20
40
60
80
100
Possible cooing in higher NH latitude, or increased clouds.
11
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Rates vs Latitude
Wavenumber (cm-1)
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
∆ B
(T)
in K
/year
-0.1
-0.05
0
0.05
0.1
0.15
0.2
-200 -150 -100 -50 0 50 100 150 200-100
-80
-60
-40
-20
0
20
40
60
80
100
Large increase in Arctic, warmer winters? Shown in earliertime series.
12
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Global Mean Rate
Wavenumber (cm-1)
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
∆ B
(T)
in K
/year
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
CO2, N2O and CH4 increases dominant.0.01K/year in window region, 2-σ error at limit of detectionMaybe slightly less water.Hash past 1400 cm−1 is due to lack of frequency calibration
13
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Global Mean B(T) Obs
Wavenumber (cm-1)
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
B(T
) in
K
210
220
230
240
250
260
270
280
290
AIRSERA: SARTA_cloudyERA SARTA_clear
Clear dominates, but still quite good agreement!
14
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Global Mean B(T) Obs
Wavenumber (cm-1)
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
Ob
s -
Ca
l
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
Clear dominates, but still quite good agreement!
15
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Global Mean Rate: ERA Simulation
Wavenumber (cm-1)
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
∆ B
(T)
in K
/year
-0.16
-0.14
-0.12
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
AIRSERA
ERA mis-match with AIRS in time, and the lack of accurate surfacetemperatures make models unsuitable for climate trending of spectralOLR. However, a direct trend on the model surface temperature is quitegood (see later slide), so this result is a bit curious.
16
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
B(T) Rates by Hemisphere
Wavenumber (cm-1)
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
∆ B
(T)
in K
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
SHNHStd NHStd SH
17
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
B(T) Rates by Hemisphere
Wavenumber (cm-1)
800 900 1000 1100 1200 1300 1400
∆ B
(T)
in K
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
SHNHStd NHStd SH
Very similar, although land masses very different!
18
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
ERA Surface Temperature versus 1231 cm−1 B(T) Rate
Latitude-80 -60 -40 -20 0 20 40 60 80
dB
T/d
t in
K/y
ea
r
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
ERA Surface T1231 cm-1 B(T)
Very good agreement?ERA land surface T inaccurate, but time dependence isn’t bad?However, 1231 cm−1 B(T) rate includes possible variations in waterand cloud amount
19
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Ocean Only: Approach for Cloud Forcing?
Wavenumber (cm-1)
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
∆ F
orc
ing in K
-0.15
-0.1
-0.05
0
0.05
0.1
ERA for clear scenes is very good, esp. for window regionsB(T) clear calculation minus B(T) observed = cloud forcingB(T) clear from ERA also has no minor gas forcingsResults suggest a slight decrease in cloud forcing, but sosmall also could be inaccurate SST used by ERA
20
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
All-Sky Rate Retrievals
Single Jacobian per latitude bin from kCARTAFor now: average all scenes in latitude bins. Future: 2Dbinning, and, bin observations by cloud forcing.Binning by cloud forcing will allow improved Jacobians inOEM retrievals.Jacobians from mean ERA fields (all-sky, with conversion of67 layer clouds to 2 layer slabs). Can be improved withretrievals on mean B(T) in future.Linear retrieval, no iteration. (Slight lien for anomaliesduring ENSO.)100 layers retrieved. Helpful for diagnostics.Retrieval constrained mostly by Tikhonov L1 regularization,not covariance estimates.Here we retrieve minor gas rates, but think it best to fixthem in the future.
21
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Temperature vs ERA-Interim, MERRA, AIRS L3
22
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Water vs ERA-Interim, MERRA, AIRS L3
23
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Water vs ERA-Interim, MERRA, AIRS L3: Improved
24
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
UncertaintiesTemperature Water
These include inter-annual variability and a 0.002 K/year estimateduncertainty in AIRS radiometric stability.
Nominal values from previous slides:
Temperature: Up to ±0.15KWater: Up to ± 0.03 fractional change
With more years, AIRS stability error will dominate over geophysicalvariability.
25
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Anomaly Retrievals
Climate community mostly interested in trends (and)AnomaliesWe create the smoothed anomalies in radiance space(and)Then perform OEM conversion to geophysical variables.Single jacobians still used, and a linear retrievalAnomalies very large compared to rates. Linear retrievalappears to be OK except during strong ENSO events.
26
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
-3S to 0S Latitude Anomaly RetrievalsUMBC ERA
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103
-3
-2
-1
0
1
2
3
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103
-3
-2
-1
0
1
2
3
AIRS L3
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103
-3
-2
-1
0
1
2
3
Liens
UMBC probably over-damped.UMBC smoothing followinglayers, should vary in strat vstrop
27
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
-72S to -64S Latitude Anomaly RetrievalsUMBC ERA
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103
-10
-8
-6
-4
-2
0
2
4
6
8
10
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103
-10
-8
-6
-4
-2
0
2
4
6
8
10
AIRS L3
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103
-10
-8
-6
-4
-2
0
2
4
6
8
10
28
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
-33S to -30S Latitude Anomaly RetrievalsUMBC ERA
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103 -2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103 -2
-1.5
-1
-0.5
0
0.5
1
1.5
2
AIRS L3
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103 -2
-1.5
-1
-0.5
0
0.5
1
1.5
2
29
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
12N to 14N Latitude Anomaly RetrievalsUMBC ERA
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103 -2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103 -2
-1.5
-1
-0.5
0
0.5
1
1.5
2
AIRS L3
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103 -2
-1.5
-1
-0.5
0
0.5
1
1.5
2
30
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
27N to 30N Latitude Anomaly RetrievalsUMBC ERA
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103
-1.5
-1
-0.5
0
0.5
1
1.5
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103
-1.5
-1
-0.5
0
0.5
1
1.5
AIRS L3
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103
-1.5
-1
-0.5
0
0.5
1
1.5
31
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
44N to 48N Latitude Anomaly RetrievalsUMBC ERA
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
AIRS L3
2004 2006 2008 2010 2012 2014
Pre
ssure
(m
bar)
100
101
102
103
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
32
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
-3S to 0S Latitude Water Anomaly RetrievalsUMBC ERA
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900-0.1
-0.05
0
0.05
0.1
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900-0.1
-0.05
0
0.05
0.1
AIRS L3
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900-0.1
-0.05
0
0.05
0.1
Liens
UMBC probably under-damped.Increase tikonov smoother.
33
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
-72S to -64S Latitude Water Anomaly RetrievalsUMBC ERA
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900 -0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900 -0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
AIRS L3
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900 -0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
34
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
-33S to -30S Latitude Water Anomaly RetrievalsUMBC ERA
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900 -0.1
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
0.08
0.1
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900 -0.1
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
0.08
0.1
AIRS L3
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900 -0.1
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
0.08
0.1
35
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
27N to 30N Latitude Water Anomaly RetrievalsUMBC ERA
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900-0.15
-0.1
-0.05
0
0.05
0.1
0.15
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900-0.15
-0.1
-0.05
0
0.05
0.1
0.15
AIRS L3
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900-0.15
-0.1
-0.05
0
0.05
0.1
0.15
36
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
44N to 48N Latitude Water Anomaly RetrievalsUMBC ERA
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900-0.1
-0.05
0
0.05
0.1
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900-0.1
-0.05
0
0.05
0.1
AIRS L3
2004 2006 2008 2010 2012 2014
Pre
ssu
re (
mb
ar)
200
300
400
500
600
700
800
900-0.1
-0.05
0
0.05
0.1
37
Introduction Spectral Rates All-Sky Retrievals Anomaly Retrievals
Conclusions
Demonstrated
Ability to generate full-mission L3 trends and anomalies in afew minutesFirst shot at L3 errors based on AIRS stability and geophysicalvariability. Insertion of AIRS stability uncertainty will becomemore important for retrievals over time!Good performance for trends and anomalies versus ERA
Future?
Can future ROSES call entertain new retrieval approaches!Produce AIRS L1c at DAAC ASAP!!Modify our approach to bin radiances by cloud amount. Willimprove Jacobians and permit more accurate retrievals.
I would like to thank Dan Zhou (NASA Langley) for his emissivityclimatology, which was used here for the ERA simulations.
