11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 1 The Wind Lidar Mission ADM-Aeolus Data...
-
Upload
kristin-spencer -
Category
Documents
-
view
215 -
download
0
Transcript of 11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 1 The Wind Lidar Mission ADM-Aeolus Data...
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 1
The Wind Lidar Mission ADM-Aeolus
Data Processing
David Tan
Research Department
ECMWF
Acknowledgements:
ESA (Mission Science & Aeolus project team)
Aeolus Mission Advisory Group
Level-1B/2A/2B Development Teams
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 2
Summary on 2 SlidesBackgroundData Processing
Assimilation of Level-2B hlos wind Simulations of Level-2B hlos wind data Assimilation impact study
Level-2B processor development How to make operational Level-2B hlos Algorithms & rationale Validation
Conclusions
Contents
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 3
Prepared for assimilating L2B hlos wind
2002-04, example for other centres
Developing Level-2B processor ECMWF is lead institute, 5 sub-contractors
2004-present
Other ongoing work/operational phase MAG, GSOV, Cal/Val, In-orbit commissioning
ECMWF to generate operational L2B/L2C products, monitor & assimilate Aeolus data, assess impact on NWP
Maintain, develop & distribute L2B processor On behalf of ESA, using NWP-SAF approach
Summary of ECMWF activities for ADM-Aeolus
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 4
Status summary: Day-1 system on track1. Level-2B hlos winds – primary product for assimilation
a. Account for more effects than L1B products
b. Will be generated in several environments
c. Motivated strategy to distribute source code
2. Main algorithm components developed & validated
a. Release 1.33 available – development/beta-testing
b. Documentation and Installation Tests
c. Portable – tested on several Linux platforms
3. Ongoing scientific and technical development
a. Sensitivity to inputs, QC/screening, weighting options
4. Contact points – ESA and/or ECMWF
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 5
Summary on 2 SlidesBackgroundData ProcessingConclusions
Contents
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 6
Background for ADM-Aeolus Measurement Concept
• Backscatter signal
• Aeolus winds are derived from Doppler shift of aerosols and molecules along lidar line-of-sight
• Error estimates, cloud & aerosol properties derived from signal strength
Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations
Orbit: 705 km, 98° inclination,in formation with Aqua,CloudSat and Parasol
Launch beginningof 2005
Mission duration: 3years
Threeco-alignedinstruments:
• 3-channel lidar– 532 nm||– 532 nm– 1064 nm
• Imaging IR radiometer• Wide-field camera
Mission Concept
Complementary Instruments
• CloudSat radar (cloud profiles)• Aqua CERES (top-of-the-atmosphere radiation)• Aqua AIRS /AMSU-A/HSB (atmospheric state)• Aqua MODIS (aerosol/cloud properties)• PARASOL (aerosol/cloud properties)• Aura OMI (aerosol absorption)
AquaCALIPSOCloudSat
PARASOL
Aura
Vertical distribution ofaerosols andclouds
Aerosol / cloud properties
CALIPSO lidar – vertical cross sections of backscatter
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 7
ADM-Aeolus with single payload Atmospheric LAser Doppler INstrument
ALADIN Observations of Line-of-Sight LOS wind profiles in
troposphere to lower stratosphere up to 30 km with vertical resolution from 250 m - 2 km
horizontal averages over 50 km every 200 km (measurements downlinked at 1km scale)
Vertical sampling with 25 range gates can be varied up to 8 times during one orbit
High requirement on random error of HLOS <1 m/s (z=0-2 km, for Δz=0.5 km) <2 m/s (z=2-16 km, for Δz= 1 km), unknown bias <0.4 m/s and linearity error <0.7 % of actual wind speed; HLOS: projection on horizontal of LOS => LOS accuracy = 0.6*HLOS
Operating @ 355 nm with spectrometers for molecular Rayleigh and aerosol/cloud Mie backscatter
First wind lidar and first High Spectral Resolution Lidar HSRL in space to obtain aerosol/cloud optical properties (backscatter and extinction coefficients)
Atmospheric Dynamics Mission ADM-Aeolus
[H]LOS
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 8
3200 wind profiles per day: about factor 3 more than radiosondes
3 hour data availability after observation (NRT-Service) => 1 data-downlink per orbit;30 minutes data availability for parts of orbit (QRT-Service with late start of downlink)
launch date May 2010 (consolidated launch date prediction in some months expected)
mission lifetime 39 months: observations from 2010-2012
ADM-Aeolus Science Report(ESA publication SP-1311, 2008)
TELLUS 60A(2), Mar 2008 special issue on ADM-Aeolus workshop 2006
50 km observations during 6 hour period
ADM-Aeolus Coverage and Data Availability
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 9
Mass and Power Budgetsmass: 1100 kg dry +116-266 kg fuelpower: 1.4 kW avg. (solar array 2.4 kW peak)mass instrument: 470 kgpower instrument: avg. 840 W (laser 510 W)Volume: 4.3 m x 2.0 m x 1.9 m
Doppler Lidar Instrument ALADIN Nd:YAG laser in burst mode operation(120 mJ @ 355 nm, 100 Hz)1.5 m Cassegrain telescopeDual-Channel-Receiver with ACCD detector (Accumulation Charge Coupled Device)
Orbitpolar, sun-synchronous, dawn-dusk (6 pm LTAN), 97° inclination; height 410 km (395-425 km), 7 days orbit repeat cycle (109 orbits); 92.5 min orbit duration
Pointing and Orbit ControlGPS, Star-Tracker, Inertial Measurement Unit, Yaw steering to compensate for earth rotation
Launcher tbd 2008 Rockot (Russia), Dnepr (Russia) or Vega (ESA)
ALADIN
Satellite and Instrument ALADIN
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 10
LIDAR Instruments for Earth Observation Missions
ADM-Aeolus/ALADINESA, launch 2010
wind profiles, aerosol and clouds EarthCARE/ATLID
ESA, launch 2013
aerosol and clouds
Future Lidar Instruments, e.g.
A-SCOPE for CO2
Calipso/CALIOPNASA, launch 2006
aerosol and clouds
IceSAT/GLASNASA, launch 2003
elevation, aerosol and clouds
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 12
ADM-Aeolus Ground Segment
[DLR]
[ESA-ESRIN]
[ESA-ESOC]
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 14
ADM-Aeolus Data ProductsProduct Contents Processor developer
and locationSize in
MByte/orbit
Level 0 Time ordered source packets with ALADIN measurement & housekeeping data
MDA (Canada)
Tromsø (Norway)
47
Level 1b
Geo-located, calibrated observational data
preliminary HLOS wind profiles (standard atmosphere used in Rayleigh processing) – not suitable for assimilation
spectrometer readouts at “measurement” scale ( 1-5 km ) – input for Level 2a/b processing
viewing geometry & scene geo-location data
MDA (Canada)
Tromsø (Norway)
10-15 (BUFR)
+
22 (EE XML Format)
Level 2a
Supplementary product
Cloud profiles, coverage, cloud top heights
Aerosol extinction and backscatter profiles, ground reflectance, optical depth
DLR-IMF (Germany)
Tromsø (Norway)
12
Level 2b
Meteorologically representative HLOS wind observations
HLOS wind profiles at “observation” scale ( ~ 50 km ) suitable for assimilation - temperature T and pressure p (Rayleigh-Brillouin) correction applied with ECMWF (or other) model T and p
ECMWF Reading (UK)
(and other NWP/research centres)
18
Level 2c
Aeolus assisted wind vector productVertical wind profiles (u and v component);
NWP model output after assimilation of Aeolus HLOS wind
ECMWF Reading (UK)
22
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 15
Ongoing ADM-Aeolus Scientific Studies
Title Team
Consolidation of ADM-Aeolus Ground Processing including L2A Products
DLR Germany
Météo-France, KNMI, IPSL, PSol
Development and Production of Aeolus Wind Data Products
ECMWF UK
Météo-France, KNMI, IPSL, DLR, DoRIT
ADM-Aeolus Campaigns DLR GermanyMétéo-France, KNMI, IPSL, DWD, MIM
Optimisation of spatial and temporal sampling KNMI Netherlands
Tropical dynamics and equatorial waves MISU Sweden
Rayleigh-Brillouin Scattering Experiment tbd
ESA plans an Announcement of Opportunity AO for ADM-Aeolus scientific use of data for late 2008 – distinct from the AO for Cal/Val
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 16
Atmospheric LAser Doppler INstrument ALADIN
Direct-Detection Doppler Lidar at 355 nm with 2 spectrometers to analyse backscatter signal from molecules (Rayleigh) and aerosol/clouds (Mie)
Double edge technique for spectrally broad molecular return, e.g. NASA GLOW instrument (Gentry et al. 2000), but sequential implementation
Fizeau spectrometer for spectrally small aerosol/cloud return
Uses Accumulation CCD as detector => high quantum efficiency >0.8 and quasi-photon counting mode
ALADIN is a High-Spectral Resolution Lidar HSRL with 3 channels: 2 for molecular signal, 1 for aerosol/cloud signal => retrieval of profiles of aerosol/cloud optical properties possible
Fig. U. Paffrath
Principle of spectrometer for molecular signal
principle of spectrometer for aerosol signal
Principle of wind measurement with ALADIN
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 18
Summary on 2 SlidesBackgroundData Processing
Assimilation of Level-2B hlos wind Simulations of Level-2B hlos wind data Assimilation impact study
Level-2B processor development How to make operational Level-2B hlos Algorithms & rationale Validation
Conclusions
Contents
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 20
ADM-Aeolus Ground Segment
[DLR]
[ESA-ESRIN]
[ESA-ESOC]
L2B HLOS
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 21
90% of Rayleigh
data have
accuracy better
than 2 m/s
In priority areas
(filling data gaps
in tropics & over
oceans)
Complemented by
good Mie data
from
cloud-tops/cirrus
(5 to 10%)
Tan & Andersson
QJRMS 2005
60°S60°S
30°S 30°S
0°0°
30°N 30°N
60°N60°N
150°W
150°W 120°W
120°W 90°W
90°W 60°W
60°W 30°W
30°W 0°
0° 30°E
30°E 60°E
60°E 90°E
90°E 120°E
120°E 150°E
150°E
Rayleigh channel Level 8ECMWF Analysis VT:Friday 7 February 2003 00UTC Surface: **high cloud cover
0.5
0.6
0.7
0.8
0.9
1.0
1%
25%
50%
75%
100%
60°S60°S
30°S 30°S
0°0°
30°N 30°N
60°N60°N
150°W
150°W 120°W
120°W 90°W
90°W 60°W
60°W 30°W
30°W 0°
0° 30°E
30°E 60°E
60°E 90°E
90°E 120°E
120°E 150°E
150°E
Mie channel Level 8ECMWF Analysis VT:Friday 7 February 2003 00UTC Surface: **high cloud cover
0.5
0.6
0.7
0.8
0.9
1.0
1%
25%
50%
75%
100%
LIPAS-simulated HLOS data – operational processors later
L2B data simulated using ECMWF clouds …
Yield (data meeting mission requirements in % terms) at 10 km
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 22
S.Hem
0.0 0.5 1.0 1.51000
100
… & impact studied via assimilation ensembles
Spread in zonal wind (U, m/s)
Scaling factor ~ 2 for wind error
Tropics, N. & S. Hem all similar
Simulated DWL adds value at all altitudes and in longer-range forecasts (T+48,T+120)
Differences significant (T-test)
Supported by information content diagnostics
Cheaper than OSSEs
ADM-Aeolus
NoSondes
Pre
ssur
e (h
Pa)
Zonal wind (m/s)
p<0.001
12-hr fc impact (Tan et al QJRMS 2007)
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 23
Global information content - consistent Mike Fisher for
Entropy Reduction & DFS
S ~ log( det( PA ) )
~ tr ( log ( J’’ -1 ) )
J’’ = 4d-var Hessian
PA = analysis error covar.
DWL data are accurate and fill data gaps
subject to usual caveats about simulated data
TEMP/PILOT Simulated
DWL
Data considered u,v to 55
hPa
HLOS
Entropy_Reduction
(“Info bits”)
4830 3123
Deg_Free_Sig 3707 2743
N_Obs 90688 50278
Info bits per obs 0.053 0.062
N_Obs/Deg_Free_Sig 24.5 18.3
Redundancy 2 3 %
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 24
Observation Processing
Data Flow at ECMWF
Non-IFS processing
Observation Screening
Assimilation Algorithm
Diagnostic post-processing
“Bufr2ODB”Convert BUFR to ODB format
Recognize HLOS as new known observable
IFS “Screening Job”Check completeness of report, blacklisting
Background Quality Control
IFS “4D-VAR”Implement HLOS in FWD, TL & ADJ Codes
Variational Quality Control
“Obstat” etc (Lars Isaksen)Recognize HLOS for statistics
Rms, bias, histograms
Prototype Level-2B (LIPAS simulation, includes representativeness error)
Analysis
Assimilation of prototype ADM-Aeolus data2003/4: introduced L2B hlos as new observed quantity in 4d-Var
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 25
Observation Processing
Data Flow at ECMWF
Non-IFS processing
Observation Screening
Assimilation Algorithm
Diagnostic post-processing
“Bufr2ODB”Convert BUFR to ODB format
Recognize HLOS as new known observable
IFS “Screening Job”Check completeness of report, blacklisting
Background Quality Control
IFS “4D-VAR”Implement HLOS in FWD, TL & ADJ Codes
Variational Quality Control
“Obstat” etc (Lars Isaksen)Recognize HLOS for statistics
Rms, bias, histograms
Level-1B data
(67 1-km measurements)
Analysis
Assimilation of prototype ADM-Aeolus data2004-: Receive L1B data & L2B processing at NWP centres
L2BP (1 50-km observation)
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 26
Aeolus Ground Segment & Data Flows - schematic view
Level-2B processor will run in different environmentsECMWF will supply source code - use as standalone or callable subroutine
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 27
Retrievals account for receiver properties …
Tan et al Tellus 60A(2) 2008
Dabas et al same issue Mie light reflected into Rayleigh channel
Rayleigh wind algorithm includes correction term involving scattering ratio (s)
ADM-Aeolus Optical Receiver - Astrium Satellites
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 28
… and for atmospheric scattering properties
Line shapeDepends on
P, T and s
Rayleigh(P,T)
Mie
Transmissionthrough the Double FP
Light transmitted through TA and TB Response (function of fD)
Response curveFunction of P, T and s
ILIAD – Impact of P & T and backscatter ratio on Rayleigh Responses - Dabas Meteo-France, Flamant IPSL
1km-scale spectra are selectively averaged Account for atmospheric variability - improve SNR
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 29
Retrievals validated for idealized broken multi-layer clouds – E2S simulator +
operational processing chainSpecified cloud layers
Retrieved clouds and aerosol
Retrieved Rayleigh winds are accurate in non-cloudy air
50 km
Retrieved Mie winds are accurate in cloud and aerosol layers
Specified wind=50 m/s
Classify scene (threshold) then average cloudy/non-cloudy regions separately
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 30
Real scattering measurements obtained from the LITE and Calipso missions
ESA’s software (E2S) is used to simulate what ADM-Aeolus would ‘see’
The L1B software retrieves scattering ratio at the 1 km measurement resolution
Our input not perfect
Realistic scenes simulated
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 31
Wind retrieval validated in the presence of heterogeneous clouds and wind – E2S
simulation
Outliers being examined
Level-1B
Rayle
igh
mole
cula
rM
ie p
art
icle
s
Retrievals fairly accurate
Backscatter from Calipso
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 33
-100 -80 -60 -40 -20 0 20 40 60 80 100
0
5
10
15
20
25
HLOS wind [m/s]Al
titude
[km]
BRC #2
E2S input (238)2b Rayleigh Cloudy (11)2b Rayleigh Clear (14)1b Rayleigh obs (24)2b Mie Cloudy (12)2b Mie Clear (14)1b Mie obs (8)
… but only after bugs were fixed in earlier versions of the L1B processor
Retrieved Mie winds revealed systematic error in L1B input
Level-1B
Rayle
igh
mole
cula
rM
ie p
art
icle
s
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 34
Wind retrieval error from ACCD digitization- theory confirmed by E2S simulation
Photon noise will dominate
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 35
Level-2B hlos error estimates – reqts met
Poli/Dabas
Meteo-France
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 36
Summary on 2 SlidesBackgroundData ProcessingConclusions
Contents
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 37
Conclusions – Day-1 system on track1. Level-2B hlos winds – primary product for assimilation
a. Account for more effects than L1B products
b. Will be generated in several environments
c. Motivated strategy to distribute source code
2. Main algorithm components developed & validated
a. Release 1.33 available – development/beta-testing
b. Documentation and Installation Tests
c. Portable – tested on several Linux platforms
3. Ongoing scientific and technical development
a. Sensitivity to inputs, QC/screening, weighting options
4. Contact points – ESA and/or ECMWF
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 38
Baker et al 1995, BAMS
ESA 1999 Report for Assessment (Stoffelen et al 2005,
BAMS) and 2008 Science Report
Weissman and Cardinali 2006, QJRMS
N. Zagar & co-authors, QJRMS & Tellus A
Tan & Andersson 2005, QJRMS
Tan et al 2007, QJRMS
Tan et al 2008, Tellus A (Special Issue on ADM-Aeolus)
Key references
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 39
5.2 Key assimilation operators Tan 2008 ECMWF Seminar Proceedings
HLOS, TL and AD
H = - u sin φ - v cos φ
dH = - du sin φ - dv cos φ
dH* = ( - dy sin φ, - dy cos φ )T
Generalize to layer averages later
Background error
Same as for u and v (assuming isotropy)
Persistence and/or representativeness error
Prototype quality control
Adapt local practice for u and v
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 53
5.1 Prototype Level-2C Processing Ingestion of L1B.bufr
into the assimilation system
L1B obs locations within ODB (internal Observation DataBase)
Assimilation of HLOS observations (L1B/L2B)
Corresponding analysis increments (Z100)
1480
1640
60°S60°S
30°S 30°S
0°0°
30°N 30°N
60°N60°N
120°W
120°W 60°W
60°W 0°
0° 60°E
60°E 120°E
120°E
NH=0.93 SH= 2.92 Trop= 1.35 Eur=-0.64 NAmer= 0.12 NAtl= 1.58 NPac= -0.29Lev=100, Par=Z, anDate=20041002-20041002 0Z, Ana Step=0, Fc Step=6Diff in RMS of an-Incr: RMS(an_erhg - bg_erhg) - RMS(an_ercp - bg_ercp)
-2.5
-2
-1.5
-1
-0.5
-0.25
-0.10.1
0.25
0.5
1
1.5
2
2.5
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 54
2a-4. Other NWP configurations
L2C Product
(optional)
Assimilation
(L2C Processor)
Other observations Forecast ModelL2B Product
L2B Processor
L1B ProductProcessing
ParametersAuxiliary Data
Meteorological
products
Integrated or Standalone, other options possible.
Possibility to modify algorithm source code.
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 58
2a-1. ECMWF “operational” configuration
L2C Product
Assimilation
(L2C Processor)
Other observations Forecast ModelL2B Product
L2B Processor
L1B ProductProcessing
ParametersAuxiliary Data
Meteorological
products
Integrated with assimilation system
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 59
2a-2. ESA-LTA late- and re-processing
L2B Product
L2B Processor
L1B ProductProcessing
ParametersAuxiliary Data
Standalone
configuration
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 60
2a-3. Research/general scientific use
Other observations Forecast ModelL2B Product
L2B Processor
L1B ProductProcessing
ParametersAuxiliary Data
Standalone configuration.
Possibility to modify algorithm source code.
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 61
1.3 Integration of Aeolus L2BP at ECMWF
L2C
AuxMet/L2B
in Screening
L1B preprocessing (if required)
L1B/2B/2C
IFS(ODB) allocation
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 62
1.3 ECMWF operational scheduleProcessing of L1B 09-21Z starts at 02Z (D+1) “dcda-12utc”
AP/L2BP
within
Screening
Aux
L1B L2B L2C
Orbit from 20:00--21:40Z is split over two assimilation cycles
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 63
2b. Why distribute L2BP Source Code? Distribution of executable binaries only permits
─ limited number of computing platforms
─ different settings in processing parameters input file
─ thresholds for QC, cloud detection
─ different auxiliary inputs
─ option to use own meteorological data (T & p) in place of ECMWF aux met data (available from LTA)
Provide maximum flexibility for other centres/institutes to generate their own products
─ different operational schedule/assimilation strategy
─ scope to improve algorithms
─ feed into new releases of the operational processor
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 64
3a. How it works – Tan et al Tellus A 2008 Rayleigh channel HLOS retrieval – Dabas et al, Tellus A
─ R = (A-B) / (A+B) and HLOS = F-1 (R;T,p,s)
─ T and p are auxiliary inputs
─ correction for Mie contamination, using estimate of scattering ratio s
Mie channel HLOS retrieval
─ peak-finding algorithm (4-parameter fit as per L1B)
Retrieval inputs are scene-weighted
─ ACCD = Σ ACCDm Wm, Wm between 0 and 1
Error estimate provided for every Rayleigh & Mie hlos
─ dominant contributions are SNR in each channel
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 66
3b. Level-2B input screening & feature finding
3
Figure1: Scattering ratio input to theE2S (top) and scattering ratio calcu-lated by the level-1b processor (bottom)
5
Figure3: Level-2b classi cation map for each measurement bin (top) and re-sult of theL1B Input Screeningprocess for each measurement bin (bottom),for Rayleigh
Poli/Dabas
Meteo-France
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 67
3b. Level-2B hlos wind retrievals5
Figure3: Level-2b classi cation map for each measurement bin (top) and re-sult of theL1B Input Screeningprocess for each measurement bin (bottom),for Rayleigh
Poli/Dabas
Meteo-France
4
Figure2: Level-2bscatteringratioused in theRayleigh inversions(calculatedby weighting over the level-1b scattering ratios) for the clear and cloudyobservations
9
- 100 - 80 - 60 - 40 - 20 0 20 40 60 80 100
0
2
4
6
8
10
12
14
16
18
20
HLOS wind [m/s]
Alti
tude
[km
]
BRC #2
E2S input (206 excld:0)2b Rayleigh Cloudy (4 excld:0)2b Rayleigh Clear (16 excld:0)1b Rayleigh obs (24 excld:0)2b Mie Cloudy (3 excld:3)2b Mie Clear (2 excld:2)1b Mie obs (2 excld:0)
Figure7: Retrievals for the2nd BRC
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 69
3c. Future work Quality Indicators
─ Highlighting doubtful L2B retrievals
─ More complicated atmospheric scenes from simulations + Airborne Demonstrator
Advanced feature-finding/optical retrievals
─ Methods based on NWP T & p introduce error correlations
Modified measurement weights
─ More weight to measurements with high SNR?
Height assignment
─ In situations with aerosol and vertical shear
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 70
4. Distribution of L2BP software Software releases issued by ECMWF/ESA
─ Details & timings to be determined
─ Probably via registration with ECMWF and/or ESA
─ Source code and scripts for installation
─ Fortran90, some C support─ Developed/tested under several compilers
─ Suite of unit tests with expected test output
─ Documentation
─ Software Release Note─ Software Users’ Manual─ Definitions of file formats (IODD), ATBD, etc.
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 71
Conclusions Expectations for ADM-Aeolus are high
─ On track for producing major benefits in NWP
─ Meeting the mission requirements for vertical resolution & accuracy
─ Extending to stratosphere, re-analysis─ Our software available to NWP/science
community
─ Combine with other observations
─ Height assignment for AMVs─ Complement other cloud/aerosol missions
─ Related research
─ Background error specification
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 75
1 Baseline L2BP Algorithm Purpose of L2BP
Produce L2B data from L1B data and aux met data50 km observations from ~ 1 km
measurementsError estimates and quality indicators
Temperature and pressure corrections via met data Scene classification and selective averaging
Design a portable source code for three processing modes
Integration at many met centres Reprocessing @ ESA (ECMWF-supplied met data) Testing in a range of environments Simple to use, yet flexible to permit extensions
Auxiliary processing – prepares met data as L2BP input
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 76
Scene classification influences L2B output
L2bP
24
Mie
or
Rayle
igh h
eig
ht
bin
s
P L1 measurements L2B Profiles
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 77
1.4 Baseline architecture – L2BP Auxiliary L2B processing (centre-dependent)
Profiles of temperature and pressure vs height At requested locations, full model vertical resolution L2BP will perform conversion to WGS84 coords
Extract from “first-guess fields” during “screening” Nearest time (within 15 mins at ECMWF) At ECMWF, vertical profiles and not slanted Currently one profile per observation
Pre-processing step Standardize input for primary L2B processing Align met data with L1B measurements in horizontal Could be achieved via extrapolation or interpolation
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 78
Obtain from geolocation information in real L1B data
Offset from the sub-satellite track
Example shows 30 mins x 50 km spacing along-track
1.5 Locations for computing aux met data
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 79
1.4 L2BP - auxiliary pre-processing Collocation implemented, suitable for 1 met locn per BRC
Sensitivity study to guide extensions, eg interpolation code
PreProc
Aux m
et
heig
ht
bin
s, 1
per
model le
vel
P L1 measurement locationsN_met raw aux met profiles
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 80
1.5 L2BP - primary processing Primary processing (HLOS retrieval)
L1B product validation (mainly in Consolidation Phase)
Signal classification (+ further code from L2A study)
Assign weights to signals (+ further development)
Apply weights to a general parameter lat & lon = L2B centre-of-gravity temperature & pressure = Tref & Pref
HLOS temperature & pressure corrections
Error estimates, quality indices
Output in EE format
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 81
2 Future work Key inputs from other activities
L1B test datasets
Cloud detection and scene classification algorithms/codes based on L2AP
Details of temperature & pressure correction scheme
ILIAD results & implementation (e.g. lookup table)
Algorithms for HLOS error estimates & Quality indicators
Check suitability of interfaces for many met centres
Basic concept ~ screening of radiosonde observations
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 82
Facts and figures for ADM-Aeolus ESA point of contact – Dr Paul Ingmann
Mission Experts Division, ESA/ESTEC, The Netherlands
Orbit Sun-synchronous Dawn-dusk
- inclination & altitude 97 408 km
Mass - total & “ALADIN” lidar component
1100 kg 450 kg
Transmitter
- laser type & pulse energy
Nd:YAG, frequency tripled
to 355 nm
150 mJ
- pulse repetition freq. & duty cycle
100 Hz 10 s every 28 s
Receiver - telescope diameter 1.5 m
- spectrometers Fizeau (Mie) Dual edge etalon (Rayleigh)
Average power demand 1400 W
Launch date & mission lifetime 2008 3 years
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 83
1 Baseline L2BP Algorithm Baseline architecture
HLOS retrieval - TN2.2, Fig 2
Generation of aux met data – TN2.2, Fig 1
L1B BRCs processed independently (& possibly in parallel)
No communication of intermediate L2BP results
L1B data arriving within met centre operational schedule
Met centre produces aux met data, L2B (and L2C)
L1B data missing the ECMWF schedule
ECMWF produces aux met data at locations inferred from predicted flight tracks
L2B possible via re-processing
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 84
1.1 L2BP – Portability considerations Common design accommodating three processing modes
Met Centres Operational (ECMWF) Re-proc (ESRIN)
L1B data (input) in EE format
(or predicted orbit locations)
Received in
Q/NRT (30m-3h)
Received in NRT (~5h) LTA/reprocessing
Auxiliary meteorological input
(T & p profiles, EE/BUFR)
Self-generated Self-generated & sent
to LTA
Oper available
(via LTA)
Primary L2BP code Oper available Oper Oper available
Auxiliary parameter input files Oper available Oper Oper available
L2B data output in EE format Yes Yes Yes
L1B/L2B data in BUFR format
(for assimilation purposes)
EE2BUFR EE2BUFR Not required
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 85
The ILIAD Study Why the ILIAD study ?
The L1 processing scheme proposed by the industry for
Rayleigh winds does not take into account the impact of the
pressure and the potential presence of Mie scattering.
Preliminary studies conducted by DLR (O. Reitebuch) and
ESA (M. Endemann) suggested the impact of both exceed
requirements on data quality.
Objectives
Find a correction scheme.
Study Team.
IPSL/LMD (P. Flamant, C. Loth), IPSL/SA (A. Garnier),
ONERA/DOTA (A. Dolfi-Bouteyre), HOVEMERE (D. Rees),
MF/CNRM (A. Dabas, M. L. Denneulin)
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 86
ILIAD - Impact of P, T and backscatter ratio on Rayleigh Responses
Line shapeDepends on
P, T and
Rayleigh(P,T)
Mie
Transmissionthrough the Double FP
Light transmitted through TA and TB Response (function of fD)
Response curveFunction of P, T and
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 87
ILIAD - Baseline Inversion Scheme
ModelP,T
L1Bp
Tenti +Gauss.
CalibrationTA(), TB()
Line shape
NA, NB
RR(fDP,TInversion
dP
vd,
dT
vd,v rr
r
RR
USR
DB,ADB,A df,T,P;ffIfT,T,P;fN
,T,P,fN,T,P,fN,T,P,fN,T,P,fN
,T,P,fRdBdA
dBdAdR
,T,P;RR
2,T,Pv R
1RR
Input
Output
Au
x.
data
dMdRd fI1T,P;fI,T,P;fI
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 88
ILIAD - Simplified correction schemeBased on a simplification of baseline inversion.
Two-step appraoch:
1. Inverse response RR as if there were no Mie.
Method: Look-up in the 3D matrix Fd (i,j,k) giving the inverse
frequency (or velocity) for pressures Pi=P0+iP, Tj=T0+jT and
Rk=R0+kR
Output parameters:
Vr(Pmod,Tmod,=1) where Pmod and Tmod are the pressure and temperature
inside the sensing volume as predicted by the NWP model.
dvr/dP, dvr/dT and dvr/dR, that is, the first order derivative of vr with
respect to P, T and the response RR.
2. Correct from Mie contamination.
Method: First order, linear correction based on the estimation of dvr/d
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 89
ILIAD - Practical implementation
K1T
hPa50P
MHz25f
T,P;fI jimR
Global aux. FileProcessed once before launch
~8 Mb
Rayleigh-Brillouinspectra
TA(fm), TB(fm)f=25 MHz
ISR
01.0R
R,T,P kjid
F
Aux. FileProcessed every timean ISR is carried out
~2 Mb
Look-up table
T,P
R
modmod
R
INVERSION
Tv
Pv
ˆ,T,Pv
rr
modmodr
Part of L2BpApplied to all Rayleigh responses
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 90
Aeolus satellite layout
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 91
ALADIN Structure and Optical Structural Thermal Model
ALADIN structure has been completed for OSTM and tested.
Mass-dummies have been integrated for OSTM: Power Laser Heads (PLH), Reference Laser Heads (RLH), and Optical Bench Assembly (OBA)
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 92
ALADIN OSTM
Platform simulator
Laser Cooling Radiator
Transmit-Receive Telescope
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 93
ALADIN Laser Cooling System
Redundant PLH
Nominal PLH
OBA
Heatpipes
Laser Radiator
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 94
ALADIN OSTM
Before shipment to CSL (Liege) for installation in vacuum chamber and full thermal vacuum testing
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 95
5 km: 75% of
Rayleigh have
accuracy < 2 m/s
(also 15% Mie
not shown)
1 km: 66% of
Mie have
accuracy < 1 m/s
(aerosol & cloud
returns)
Adequate
transmission
through
overlying cloud
60°S60°S
30°S 30°S
0°0°
30°N 30°N
60°N60°N
150°W
150°W 120°W
120°W 90°W
90°W 60°W
60°W 30°W
30°W 0°
0° 30°E
30°E 60°E
60°E 90°E
90°E 120°E
120°E 150°E
150°E
Rayleigh channel Level 13ECMWF Analysis VT:Friday 7 February 2003 00UTC Surface: **high cloud cover
0.5
0.6
0.7
0.8
0.9
1.0
1%
25%
50%
75%
100%
60°S60°S
30°S 30°S
0°0°
30°N 30°N
60°N60°N
150°W
150°W 120°W
120°W 90°W
90°W 60°W
60°W 30°W
30°W 0°
0° 30°E
30°E 60°E
60°E 90°E
90°E 120°E
120°E 150°E
150°E
Mie channel Level 18ECMWF Analysis VT:Friday 7 February 2003 00UTC Surface: **low cloud cover
0.5
0.6
0.7
0.8
0.9
1.0
1%
25%
50%
75%
100%
Data simulations for ADM-AeolusYield (%age of data meeting mission requirements) at 5 & 1 km
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 96
ADM-Aeolus data simulations - comparison with radiosondes/mission spec
Aeolus median like obs error assigned operationally to radiosondes
Aeolus HLOS observations expected to receive appreciable weight
Rayleigh
Mie
With cross-track representativity(cf radiosonde – dashed)
Without representativity(cf mission spec – dash-dot)
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 97
ADM-Aeolus data simulations – Effects of model cloud cover (2)
Mid-latitude example
QC implications, Task 2
Tails of Rayleigh error distributions underestimated, median barely changed
Model
Observedcloudcover
(midlats)
11 July 2008 Data Processing for ADM-Aeolus – JCSDA Slide 98
Assimilation of prototype ADM-Aeolus data
Most QC parameters taken from conventional wind obs
Background errors & quality control thresholds (BgQC+VarQC)
Aeolus-specific Background Quality Control (recommended option)
Capping of observation error in bg departure classification
Testing with LITE period, LIPAS-simulated Level-2B data
Gaussian + non-Gaussian errors (instrument bias, input wind bias)
Operational model (Cy26r1) at full/reduced resolution, ERA40/NoSSMI
Quality Control for Aeolus data
Set B = (obs-bg) / ES(obs-bg), accept obs iff abs(B) < 4.
In standard BgQC for Aeolus, ES = (σo2 + σb
2)1/2.
Aeolus option: ES = (so2 + σb
2)1/2, where so = min(σo, 2.5 ms-1)