Retrieval Error Sensitivity
1DVAR Retrievals
• P calculated with respect to surface and cloud errors in BC
• ATOVS observations and DYCOMS II COAMPS
C
C
0
0 hPa
5 hPa
10 hPa
20 hPa
50 hPa
100
300
400
400
400
500
600
700
700
800
800
900
900
1000
5
10
15
20
25
30
35
40
45
Percent improvement in profile temperature (EPAC (clear) Profile 1A background)
50
600
800
CLW error = 0.03 g kg-1
900
besdo cloud errors = 60 hPa, 0.50, 0.03 g / kg
1000
0
10
20
30
40
50
60
70
80
90
Percent improvement in profile temperature (EPAC (cloudy) Profile 2A1 background)
εm background errors of
0,
1, 2, 5%
BC background
errors
100
0
10
20
30
40
50
60
70
80
90
Percent improvement in profile temperature (EPAC (cloudy) Profile 2A1 background)
100
CLW background errors of
0.00, 0.05, C0.25, 0.50 g kg-1
0%
1%
2%
5%
0.00 g/kg
0.05 g/kg
0.25 g/kg
0.50 g/kg
100
200
300
300
400
400
Pressure hPa
200
500
600
700
700
800
800
900
900
1000
10
20
30
40
50
60
70
Percent improvement in profile loge specific humidity (EPAC (clear) 1A background)
80
840
860
860
880
880
900
920
• ATOVS observations and COAMPS
0
™
900
920
940
960
960
980
980
1000
1000
1020
280
1000
0
Dropsonde observation
COAMPS background profiles
Retrieved profiles
820
840
940
600
AMSU only
800
Dropsonde observation
COAMPS background profiles
Retrieved profiles
820
Quality of q retrieval is highly
dependent upon prior knowledge
of CLW.
500
besdo cloud errors = 60 hPa, 0.50, 0.03 g / kg
AMSU only
800
Key Results
Satellite-derived information
adjusted profile T and loge q
retrieval toward observed in-situ
values.
0
0
100
Pressure hPa
500
700
1000
0
short-term forecast fields
Pressure hPa
600
Pressure hPa
200
300
Pressure hPa
200
300
500
0%
10 %
50 %
100 %
100
200
Pressure hPa
0.00 K
1.57 K
3.14 K
100
™
• MW channels only
AMSU-A and B observations from NOAA-16
Averaged over inner model domain
No bias corrections applied
• VT 0900Z 11 July 2001
™
• COAMPS 9-hr forecast fields as xb (N = 660)
• Retrievals compared with collocated dropsonde observation
C
0
0
282
284
286
288
290
292
Temperature K
294
296
298
1020
300
0
0.5
1
1.5
2
Loge specific humidity g kg−1
2.5
3
/ NAVDAS forecast system
10
20
30
40
50
60
70
80
Percent improvement in profile loge specific humidity (EPAC (cloudy) Profile 2B2 background)
• MW channels only
AMSU-A and B observations from NOAA-16
Bias corrected (AMSU-A)
•18 km, single nested domain corresponding to the DYCOMS II outer area
• Real data model simulation
• 6-hr forecast field VT 12Z 11 July 01
• 4, 550 “successful” retrievals
• Comparison with collocated dropsonde and NESDIS retrieval
Key Results
Near-surface profile T retrieval can be improved by prior knowledge of Tskin
HIR/3 surface channels (e.g., Ch 8) provides TCT information
Profile T retrieval is very sensitive to PCT and CFC error
AMSU-B Ch 2 (150 GHz) provides majority of low-level q information
q retrieval is very sensitive to εm and CLW error
20010711 1423Z 31.7 N 121.1 W
20010711 1423Z 31.7 N 121.1 W
800
Key Results
NRL 1DVAR Non-Linear Minimization Equation
Smaller adjustment:
• model run closer to truth
• bias corrected observations
xi+1 = xi + BHiT [Hi BHiT + R]-1 [yo – H(xi) + Hi xi – Hi xb]
800
Dropsonde observation
COAMPS background profile
Retrieved profile
820
840
860
860
880
880
900
920
900
920
940
940
960
960
Simulated 1DVAR Retrievals
980
980
1000
1000
1020
280
Dropsonde observation
COAMPS background profile
Retrieved profile
820
840
Pressure hPa
Satellite-derived information
adjusted profile T and loge q
retrieval toward observed in-situ
values.
Pressure hPa
Part II: 1DVAR RETRIEVAL STUDY
1020
282
284
286
288
290
292
Temperature K
294
296
298
300
0
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Loge specific humidity g kg−1
2.25
2.5
2.75
3
• Simulated ATOVS observations (yos) and simulated background state vectors (xb) (N = 1000)
• 1DVAR retrieval versus NESDIS Retrieval
• “True” profile xt is set of representative EPAC xb
y os = H ( xt ) +
λ i B*i
∑γ
λ j R*j
2
• P compared for error statistics of converged solutions (σr ) versus theoretically derived values (Si)
⎡
σ r = ⎢ x − xt
⎣
(
)
2
1/ 2
⎤
− µ2 ⎥
⎦
−1
C
1DVAR retrieval with
COAMPSTM background is
better choice for data
assimilation.
C
0
0
simulated retrieval
theoretical retrieval
Good agreement between
theoretical and iterated solution
simulated retrieval
theoretical retrieval
400
500
600
1000
Pressure hPa
Pressure hPa
500
600
700
800
900
1000
40
45
50
0
10
20
30
40
50
60
Percent improvement in profile loge specific humidity
70
80
™
900
220
230
240
250
260
Temperature K
270
280
290
300
−7
−6
−5
−4
−3
−2
−1
0
Loge specific humidity g kg−1
1
2
3
Analyses of information content and theoretical retrieval performance showed that, when treated optimally, significant
humidity and temperature information can be derived from ATOVS retrievals within the clear and cloudy sky summertime
EPAC environment.
The 1DVAR results are consistent with the theoretical information content study and indicate that these satellite
observations can provide information that, when used in concert with COAMPS™ short-term forecasts, reduce the
retrieval error and adjust the retrieval within the shallow boundary layer toward the designated “true” profile.
A study of theoretical retrieval error sensitivity to representative EPAC background state vector elements and associated
errors established the a priori elements critical for successful 1DVAR retrievals.
The generally good agreement between theoretical retrieval errors and the error statistics calculated using non-linear
iteration demonstrates the consistency and reliability of the NRL 1DVAR retrieval scheme.
Cooper, G.A., 2002: Variational retrieval of Eastern Pacific atmospheric boundary layer parameters using ATOVS with COAMPS mesoscale forecast system. Ph.D. Dissertation, Dept. of Meteorology, Naval Postgraduate School, Monterey, CA, 272pp.
besdo cloud errors = 20 hPa, 0.00, 0.10 g/kg bsv cfc = 1.00, BC, mod 3
800
"true" profile
COAMPS background profiles
retrieved profiles
820
Daley, R., and E. Barker, 2000: NRL Atmospheric Variational Data Assimilation System (NAVDAS) Source Book 2000. NRL/PU/7530—00-418, Naval Research Laboratory Marine Meteorology Division, Monterey, CA, 155pp.
"true" profile
COAMPS background profiles
retrieved profiles
820
840
English, S.J., 1999: Estimation of temperature and humidity profile information from microwave radiances over different surface types. J. Appl. Meteor., 38, 1526-1541.
840
860
+4K
Pressure hPa
880
900
920
860
_____, R.J. Renshaw, P.C. Dibben, A.J. Smith, P.J. Rayer, C. Poulsen, F.W. Saunders, and J.R. Eyre, 2000: A comparison of the impact of TOVS and ATOVS satellite sounding data on the accuracy of numerical weather forecasts. Quart. J. Roy.
Meteor. Soc., 126, 2911-2931.
880
Eyre, J.R., 1989: Inversion of cloudy satellite sounding radiances by nonlinear optimal estimation. I: Theory and simulation for TOVS. Quart. J. Roy. Meteor. Soc., 115, 1001-1026.
_____, 1990: The information content of data from satellite sounding systems: A simulation study. Quart. J. Roy. Meteor. Soc., 116, 401-434.
900
_____, G.A. Kelly, A.P. McNally, E. Andersson, and A. Persson, 1993: Assimilation of TOVS radiance information through one-dimensional variational analysis. Quart. J. Roy. Meteor. Soc., 119, 1427-1463.
920
940
940
960
960
980
980
1000
1000
Franke, R., 1999: Vertical correlation functions for temperature and relative humidity errors. NRL/MR/7531—99-7240, 78pp. [Available from Naval Research Laboratory, Marine Meteorology Division, Monterey, CA, 93943-5502.]
Garand, L., 2000: Sensitivity of retrieved atmospheric profiles from infrared radiances to physical and statistical parameters of the data assimilation system. Atmos-Ocean, 38, 431-455.
Healy, S.B., and J.R. Eyre, 2000: Retrieving temperature, water vapor and surface pressure information from refractive-index profiles derived by radio occultation: A simulation study. Quart. J. Roy. Meteor. Soc. 126, 1661-1683.
Li, J., W.W. Wolf, W.P. Menzel, W. Zhang, H. Hunag, and T.H. Achtor, 2000: Global Soundings of the Atmosphere from ATOVS measurements: The algorithm and validation. J. Appl. Meteor., 39, 1248-1268.
Rodgers, C.D., 2000: Inverse Methods for Atmospheric Sounding: Theory and Practice. World Scientific, Singapore, 238pp.
1020
280
282
4
References
besdo cloud errors = 20 hPa, 0.00, 0.10 g/kg bsv cfc = 1.00, BC, mod 3
800
Key Results
800
1000
210
Summary
600
800
15
20
25
30
35
Percent improvement in profile temperature
700
500
700
10
600
800
200
400
500
700
300
IR and MW channels included
VT 0900Z 11 July 2001
Collocated ECMWF profile as xt
™
COAMPS 9-hr forecast fields as xb (N = 660)
Loge q profile adjusted toward
“truth”
300
400
300
• Simulated ATOVS observations and DYCOMS II COAMPS short-term forecast fields
4 K change in profile T above PCT
300
900
900
Satellite-derived information
adjusted profile T and loge q
retrieval toward designated true
profile.
200
200
1000
•
•
•
•
200
200
400
NESDIS retrieval
1DVAR retrieval
100
100
5
20010711 1423Z 31.7 N 121.1 W
0
100
100
0
NESDIS retrieval -.+-.+
NESDIS retrieval
1DVAR retrieval
Collocated NESDIS retrieval
lacks MABL structure.
Si = B − BHTi ⎡ Hi BHTi + R ⎤ H i B
⎣
⎦
Key Result
1DVAR retrieval -----VT 20010711 1200Z
0
Key Results
j
i
Pressure hPa
∑γ
Pressure hPa
x b = xt +
Pressure hPa
Pressure hPa
CFC background errors of
0, 10, 50, 100%
PCT background errors of
0, 5, 10, 20, 50 hPa
Tskin background errors of
0.00, 1.57, 3.14 K
284
286
288
290
292
Temperature K
294
296
298
300
1020
0
0.5
PCT error = 20 hPa CFC error = 0.00 CLW error = 0.10 g kg-1
1
1.5
2
Loge specific humidity g kg−1
2.5
3
Saunders, R.W., 2000: NWP-SAF 4 year plan for RT development and RTTOV-6 science and validation report. [Available online at http://www.met-office.gov.uk.]
Thepaut, J.-N, and P. Moll, 1990: Variational inversion of simulated TOVS radiances using the adjoint technique. Quart. J. Roy. Meteor. Soc., 116, 1425-1448.
University Corporation for Atmospheric Research / Joint Office for Science Support (UCAR / JOSS), cited 2002. Dynamics and Chemistry of Marine Stratocumulus (DYCOMS) Phase II data set. [Available online at: http://www.joss.ucar.edu.]