Variational Retrieval of Eastern Pacific Marine Atmospheric Boundary Layer Parameters
using ATOVS with the COAMPS Mesoscale Forecast System
TM
++
Grant Cooper1 2 , Nancy Baker2, and Philip Durkee1
1Naval
Postgraduate School, Monterey, CA; 2Naval Research Laboratory, Monterey, CA
Motivation
{++Current affiliation U.S. Navy Fifth Fleet, Bahrain}
Part I: Information Content and Retrieval Error Sensitivity
To investigate the ability of an operational IR and MW atmospheric sounding system to contribute information to a mesoscale
NWP system within the summertime eastern Pacific (EPAC) environment.
Information matrix = Hessian of cost function J(x)
Alternate form of retrieval error covariance
0.5∇2x J(x) = S(x)-1 = B-1 + HT R-1 H
• Advanced TIROS Operational Vertical Sounder (ATOVS)
S = B – B HT [HBHT + R]-1 H B
• Coupled Ocean – Atmosphere Mesoscale Prediction System (COAMPS™ ) and the Naval Research Laboratory (NRL) Atmospheric
Variational Data Assimilation System (NAVDAS)
Profile T and Loge q Jacobians
(Column elements of H calculated using RTTOV-6)
Approach
HIRS/3 channel 13 (4.57 um)
AMSU−B channel 2 (150.0 GHz)
0
200
300
300
300
400
400
400
500
1DVAR profile temperature and humidity retrievals using both simulated and actual ATOVS observations constrained by the
COAMPS™ short-term forecasts and synoptically relevant background errors.
The time period of interest coincided with the Dynamics and Chemistry of Marine Stratocumulus (DYCOMS) Phase II field
study (July 2001).
900
ATOVS Instrument and RTTOV-6
Forward Model Errors (R = O + F)
9
H2Ov
8
10
O3
7
1
H2Ov
W
CO2
(Rii)
CH
W
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
wavenumber (cm-1) / wavelength (µm)
After Kidder and Vonder Haar (1995)
AMSU-B ch 2
AMSU-B ch 3 - 5
AMSU-A ch 15
AMSU-B ch 1
AMSU-A ch 1
AMSU-A ch 2
AMSU-A ch 3-14
19-16
HIRS/3
2.78
0.77
0.58
0.37
0.27
0.27
0.24
0.21
0.24
0.26
0.48
0.98
0.22
0.22
0.31
1/2
K
AMSU-A AMSU-B
2.13
2.43
1.24
0.31
0.32
0.32
0.31
0.31
0.37
0.40
0.40
0.49
0.58
0.95
1.93
100
400
Pressure hPa
400
500
600
COAMPSTM EPAC
profile T errors
700
500
600
700
COAMPSTM EPAC
profile loge q errors
800
800
900
900
1000
1000
0
0.25
0.5
0.75
1
1.25 1.5 1.75
2
2.25
Standard deviation of error K
2.5
2.75
3
3.25
3.5
1/2
NOGAPS
COAMPS™
Unit
1.80
2.15
K
T2m
300
300
0
0.1
0.2
0.3
0.04
0.06
800
900
1000
−6
−4
−2
0
2
Loge specific humidity Jacobian K
4
0.4
0.5
0.6
0.7
0.8
Standard deviation of error g/kg
0.9
1
1.1
1000
6
−6
The effective rank = number of singular values > ~ 1
Elements outside effective row space cannot be
measured.
λi2
loge q2m
0.18
0.18
g kg-1
Ps
3.38
3.38
hPa
u2m
0.00
2.00
ms-1
v2m
0.00
2.00
ms-1
Tskin
1.57
1.57
K
PCT
N/A
Various
hPa
CFC
N/A
Various
0-1
εm
0.00
0.02
0-1
1.2
∑ (1 + λ i2)
U.S. Standard Atmosphere
Diagonal B
Full B
Diagonal B
7.4
6.9
7.2
6.7
BN background errors
HIRS/3
AMSU-A
AMSU-B
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
100
Reciprocal of the diagonal elements of
A = KH = BNHT[HBN HT + R]-1 H
400
Vertical resolution ~ 100 hPa near clear sky
MABL
200
300
300
400
400
Pressure hPa
500
600
300
300
500
600
Improved temperature retrieval above cloud-top
700
700
800
800
900
220
230
240
250
260
Temperature K
270
280
290
300
200
230
240
250
260
Temperature K
270
280
290
3
5
T2m , loge q2m , Ps , u2m, v2m , Tskin
PCT
7
9
11 13
15 17 19 21 23 25 27 29 31 33
Temperature resolution (model level)
35 37
39 41
43
200
300
300
400
400
500
600
800
800
900
900
1000
•Emissivity:
600
700
−5
−4
−3
−2
−1
Loge specific humidity g kg−1
0
1
2
3
−7
IR
MW
ISEM-6
FASTEM
−6
−5
−4
−3
−2
−1
0
Loge specific humidity g kg−1
1
2
3
4
5
7
9
11 13 15 17 19 21 23 25 27 29 31 33
Loge specific humidity resolution (model level)
35 37
39 41
43
Loge specific humidity
N
200
P → 100%
300
300
As a function of ATOVS instrument
35 channels
HIR/3 only
AMSU−A only
AMSU−B only
100
200
400
Pressure hPa
500
500
600
600
700
AMSU-B
HIRS/3
800
900
900
1000
0
5
10
15
20
25
30
35
40
45
Percent improvement in profile temperature (EPAC (clear) Profile 1A background)
50
0
10
20
30
40
50
60
70
Percent improvement in profile loge specific humidity (EPAC (clear) Profile 1A background)
80
Loge specific
humidity
N
0
0
35 ch
AMSU−A ch 1 only
AMSU−A ch 2 only
AMSU−A ch 3 only
AMSU−A ch 4 only
AMSU−A/B ch 15/1 only
AMSU−B ch 2 only
100
35 ch
AMSU−A ch 1 only
AMSU−A ch 2 only
AMSU−A ch 3 only
AMSU−A ch 4 only
AMSU−A/B ch 15/1 only
AMSU−B ch 2 only
100
200
300
300
400
400
500
600
500
600
AMSU-B Ch 2
700
700
As a function of ATOVS “window channel”
AMSU-B Ch2
800
800
900
900
1000
1000
0
1000
−6
3
0
35 channels
HIR/3 only
AMSU−A only
AMSU−B only
200
500
700
1
Temperature
100
2A1 957.44 hPa
2B2 922.46 hPa
1.00
CFC
8
600
0
2A1 90 g
2B2 285 g m-2
•Cloud elements:
7.5
NOGAPS BN
Theoretical Retrieval Performance (P)
m-2
•Surface elements:
300
U.S. Standard Atmosphere
EPAC (cloudy) Profile 2A
EPAC (cloudy) Profile 2B
100
200
Pressure hPa
Pressure hPa
100
−7
220
7
500
N
210
6.5
900
0
U.S. Standard Atmosphere
EPAC (clear) Profile 1A
EPAC (clear) Profile 1B
6
U.S. Standard Atmosphere
EPAC (clear) Profile 1A
EPAC (clear) Profile 1B
EPAC (cloudy) Profile 2A1
EPAC (cloudy) Profile 2B2
Loge specific humidity
1000
210
0
5.5
800
900
1000
200
•Profile LWP:
3
3.5
4
4.5
5
Standard deviation of error K
1000
1
1000
600
2.5
400
700
RTTOV-6 reference profile
2
700
800
800
•Profile O3:
1.5
Loge specific
humidity
N
1000
As Sii / Bii → 0
1
100
200
500
700
0.5
0
U.S. Standard Atmosphere
EPAC (clear) Profile 1A
EPAC (clear) Profile 1B
EPAC (cloudy) Profile 2A1
EPAC (cloudy) Profile 2B2
Pressure hPa
Pressure hPa
200
0
200
900
Summertime conditions
Clear and cloudy sky
U.S. Standard Atmosphere
EPAC (cloudy) Profile 2A
EPAC (cloudy) Profile 2B
100
observation error
effective background error (Profile 3)
effective background error (Profile 1A)
effective background error (Profile 1B)
effective background error (Profile 2A1)
effective background error (Profile 2B2)
Temperature
N
0
Number of RTTOV-6 model levels per
degree of freedom for signal
Pressure hPa
U.S. Standard Atmosphere
EPAC (clear) Profile 1A
EPAC (clear) Profile 1B
100
6
Total degrees of freedom for signal
EPAC (clear) Profile 1B
400
Cloudy sky
0
4
Full B
N
Representative EPAC Background State Vectors (xb)
Clear sky
−2
0
2
Loge specific humidity Jacobian K
= tr ( A)
P = 100[1 – (Sii / Bii)]
0
−4
Singular Value Decomposition (H’ = R-½ H BN½)
Pressure hPa
Pressure hPa
Background error (Bii)
NOGAPS
EPAC COAMPS
200
200
−0.02
0
0.02
Temperature Jacobian K
Projects B into radiance space and allows for direct
comparison with R.
For channels with large effective background error, the
retrieval is nearly insensitive to the background and the
solution is determined primarily by the satellite
observation.
- COAMPS™ EPAC 6-hr forecast errors
- Correlated surface-atmosphere temperature errors
0
100
−0.04
Vertical Resolution
BC
log specific humidity
air temperature
−0.06
Larger effective background error for AMSU window
channels
- Globally averaged NOGAPS errors
- First-order approximation
- No surface-atmosphere error correlations
NOGAPS
EPAC COAMPS
1000
Effective Background Error (H BN HT)
Background Error Covariance Matrix (B)
0
700
900
i
After http://amsu.cira.colorado.edu
BN
600
800
dsi =
1.96
1.29
1.15
0.78
0.98
500
700
800
Pressure hPa
Transmittance
N2 CO2
11
600
Pressure hPa
12
500
ATOVS channel
HIRS/3 channel
15-13
700
Pressure hPa
200
EPAC (cloudy) Profile 2A1
EPAC (cloudy) Profile 2A2
EPAC (cloudy) Profile 2B1
EPAC (cloudy) Profile 2B2
AMSU-B CH 2
150.0 GHz
100
200
600
ATOVS Channels
0
U.S. Standard Atmosphere
EPAC (clear) Profile 1A
EPAC (clear) Profile 1B
AMSU-B CH 2
150.0 GHz
100
Pressure hPa
• Establish the a priori elements critical for successful 1DVAR retrievals within the clear and cloudy sky EPAC environment
Pressure hPa
• Quantify total profile temperature and humidity information
AMSU−B channel 2 (150.0 GHz)
0
U.S. Standard Atmosphere
EPAC (clear) Profile 1A
EPAC (clear) Profile 1B
HIRS/3 CH 13
4.57 µm
100
Information content and retrieval error sensitivity analyses based upon representative EPAC background state vectors
10
20
30
40
50
60
70
80
Percent improvement in profile loge specific humidity (U.S. Standard Atmosphere background)
0
10
20
30
40
50
60
70
Percent improvement in profile loge specific humidity (EPAC (clear) Profile 1A background)
80