recent activities on amsr-e data utilization in nwp at jma masahiro kazumori, koichi yoshimoto,...
TRANSCRIPT
Recent activities on AMSR-E data utilization
in NWP at JMA
Masahiro Kazumori,Koichi Yoshimoto, Takumu
Egawa
Numerical Prediction DivisionJapan Meteorological Agency
2-3 June, 2010 AMSR-E Science Team Meeting, Huntsville, AL, U.S.A.
Outline
Status of JMA NWP models and Microwave imager data utilization
Verification of AMSR-E TPW retrieval algorithm with global GPS TPW data
Application to SSMIS TPW retrieval and the assimilation experiment in JMA NWP
Expectations for Microwave imager dataObservational local timeData latency
Summary
JMA NWP models
Global Model (GSM) Meso Scale Model (MSM)
PurposesShort- and medium-range forecast
Very-short-range forecast
Forecast domain GlobeJapan and its surrounding areas
Grid size and/or number of grids
0.1875 deg. (TL959) 5 km / 721 x 577
Vertical levels / Top
60 / 0.1 hPa 50 / 21,800 m
Forecast hours (Initial time)
84 hours (00, 06, 18 UTC)216 hours (12 UTC)
15 hours (00, 06, 12, 18 UTC)33 hours (03, 09, 15, 21 UTC)
Analysis 4D-Var 4D-Var
MW Imager data utilization in JMA
For Global Model:Radiance assimilation Brightness Temperature in clear sky condition
For Meso scale Model:Retrieval Assimilation Total Precipitable Water(TPW) and Rain Rate (RR)
Data thinning : 200km grid boxQC : cloud screening and bias correction Colored point data are actually assimilated.
Recent update in MSMGround based GPS TPW data in Japan
GPS TPW data in Japan was introduced in operational JMA MSM DA system in Oct. 2009. The GPS data provide accurate and periodic TPW information over land. Improvements of rain prediction were confirmed in heavy rain cases. Atmospheric moisture information is essential to produce better rain forecast. Also global GPW TPW data set are available in JMA for verifications of NWP model’s TPW and satellite TPW products.
GPS data are delivered from Geospatial Information Authority of Japan (GSI) and converted to TPW products in JMA.
Without GPSWith GPSAnalyzed precipitation
Three-hourly accumulated precipitation of 3-hour forecasts from 20 Jul. 2009 at an initial time of 21 UTC. From the left, analyzed precipitation, the forecast of Test (with GPS TPW) and that of Control (without GPS TPW) .
Verification of AMSR-E TPW products with global GPS TPW
data
AMSR-E and GPS collocation criteria:
GPS altitude <= 200m, Spatial diff. <= 20km,Time diff. <= 10 min.
Period: 20 Jun. – 20 Aug. 2009
ZWDZHDZTD
Hg
Pgm
RkZHD
m
sfcmd
00028.02cos00266.01
784.9
10 16
Φ
dzTP
dzTP
T
Tk
mm
kkR
PWVZWD
v
v
m
md
vv
2
312
5
1
,10
ZTD : Zenith Tropospheric Delay ZHD : Zenith Hydrostatic DelayZWD : Zenith Wet Delay
Locations of collocated GPS Data (35 sites)
GPS analysis・ GPS satellite ephemeris : final ephemeris of International Global Navigation Satellite System Service (IGS).・ GPS data (RINEX) : IGS station・ Software : GIPSY/OASIS-II
JAXA-L2
Verification of AMSR-E TPW products
with global GPS TPW dataScatter diagram of TPW GPS vs. AMSR-E
NEW The National Snow and Ice Data Center (NSIDC)
Verification of AMSR-E TPW products by global GPS TPW data
setTPW’s time sequences for NEW, JAXA-L2, and NSIDC products
CHICHIJIMA
Chatham Island
Time sequence of observed hourly rain fall in Yamaguchi prefecture
Hourly Rainfall (left axis)Total Rainfall (right axis)
A case study:Assimilation of SSMIS TPW & RR in MSMHeavy rain case in Japan
July 19 – 26, 2009
00UTC Jul. 21, 200924hr observed rainfall
MTSAT IR image
The average year value for July’s one month rainfall
Data coverage of Microwave Imager data in JMA MSM
MSM analyses were executed in every 3 hour (00,03,06,09,12,15,18 and 21UTC)SSMIS TPW and RR assimilation period : July 19 to 26, 2009
33 hours forecasts were produced from 03,09,15 and 21UTC initial.
SSMIS data is available in these analysis time
Cntl (W/O SSMIS)
Test (With SSMIS)
00 03 06 09 12 15 18 21
00 03 06 09 12 15 18 21
Red : F13 SSMIBlue : TRMM TMI
Light Blue : Aqua AMSR-E Green : F16 SSMISPurple : F17 SSMIS
Impact on moisture analysis in July 20
TPW Analysis difference (Test-Cntl)
Generally, assimilation of SSMIS intensify moisture flow in the analysis.
Analyzed TPW field in Test (with SSMIS)
03UTC 09UTC 15UTC 21UTC
Jul. 20 15UTC INITIAL FT=0
12
TPW DIFF (TEST-CNTL)TEST TPW
FT=1 [hour]
13
TPW DIFF (TEST-CNTL)TEST TPW
FT=2
14
TPW DIFF (TEST-CNTL)TEST TPW
FT=3
15
TPW DIFF (TEST-CNTL)TEST TPW
FT=4
16
TPW DIFF (TEST-CNTL)TEST TPW
FT=5
1-4 March 2010 17
TPW DIFF (TEST-CNTL)TEST TPW
FT=6
18
TPW DIFF (TEST-CNTL)TEST TPW
FT=7
19
TPW DIFF (TEST-CNTL)TEST TPW
FT=8
20
TPW DIFF (TEST-CNTL)TEST TPW
FT=9
21
TPW DIFF (TEST-CNTL)TEST TPW
FT=10
22
TPW DIFF (TEST-CNTL)TEST TPW
FT=11
23
TPW DIFF (TEST-CNTL)TEST TPW
FT=12
24
TPW DIFF (TEST-CNTL)TEST TPW
Impact on Rain ForecastRadar observation CNTL(w/o SSMIS)TEST (with SSMIS)
TEST-CNTLTPW DIFF
TEST TPW
[mm][mm]
Valid Time: Jul. 21 12JST
FT=12 FT=12
3hr rain FT=12 FT=12
Strong rain band appeared,but, the crossing time of the rain band was not improved.
Increased TPW in moist area,decreased in dry area.
SSMIS intensified the moisture flow in the forecast
Observational Local Time For the purpose of operational use of satellite microwave imager data in NWP, observational local time is a key element. NWP centers use 6hrs assimilation time window. Continuity of MW measurements in A-train is indispensable.
12
1806
00
13:30
Light Blue : Aqua/AMSR-EPurple : DMSP F-16/SSMISGreen : DMSP F-17/SSMISOrange : Coriolis/WindSat
Dark black points indicate WindSat data in 6-hrs time
window
Data Latency
Data latency for AMSR-E (JAXA)
Data latency for ATOVS (MSC)
Data latency for AMSR-E (Global)
Data latency for MTSAT
Timely data delivery is also important for the use of satellite data in operational NWP.Especially, regional analysis demand strict cut off time for data receiving. MSM requires 50min cut off time after the analysis time for every analysis (8 time/day).Direct receiving in the frame work of WMO RARS and EARS are suitable for the regional data use for ATOVS.
SummaryTPW data from MW-Imager play important role for accurate rain forecasts in MSM.
TPW retrieval algorithm was verified with ground based GPS TPW data. Improvement was found compared with current JAXA L2 product, however, there is room for further improvement. NSIDC products showed better accuracy in GPS TPW verification.
The algorithm was applied for F-16 and F-17 SSMIS. The retrieved TPW and RR were assimilated in JMA MSM for a heavy rain case in Japan. Assimilation of new SSMIS TPW data produced strong rain band forecast, but the forecasted rain band location was not improved.
Data coverage is a key issue for satellite data utilization in operational NWP. Large coverage in each analysis is expected with timely data delivery. AMSR-E observation in afternoon orbit (A-train) is indispensable.
Backup slides
Comparison between RAOB and GPS(Spatial diff.<30km, altitude diff. < 200m)
GPS Remote Sensing
ReceiverReceiver
GPS satelliteGPS satelliteGPS satelliteGPS satellite GPS satelliteGPS satellite
VaporVaporVaporVapor
Pseudo Pseudo RangeRangePseudo Pseudo
RangeRange
Wet Delay
Zenith Tropospheric Delay
= Zenith Hydrostatic Delay
+ Zenith Wet Delay
GPS ephemeris
GPS software( GIPSY )
GPS observation data (RINEX)
ZTD
Surface Pressure, TemperatureTPW
Conversion
Procedure
Other data’s coverage in MSM
: Observed brightness temperature
: Atmospheric Transmittance
Theoretical basis of the algorithm
TrTrTTrTTrTT )1()1()1( sassab
2
s
sasa )1(1
)(
Tr
TrTTT
2sb )( TrT
Vertical mean temperature of atmosphere and ocean surface system
)1(a TrT )0(s TrTsa TT
s Determination of by pre-defined LUT as a function of frequency,
incidence angle, SST and SSWStep1
Step3
Step2
Step4
Step5
Step6
2.02 eTr Initial atmospheric transmittance is set as exp(-0.2)
)850,( 2aa TTrTT
HV , Calculation of mean emission temperature by using Eq. (1-4)
Calculation of Transmittance (V pol. & H pol.) by using Eq. (1-3)
s
b2
1
1
T
Tr
H2
V2 , TrTr
H2
V22 TrTrTr Calculation of new transmittance
s
Determination of by pre-defined LUT of and T850 based on RAOB2TraT
Iteration calculation of Step 3 – 6 to obtain optimized Transmittance
bT
Tr
aT
s: Mean emission temperature
: Ocean surface emissivity
Ta is defined as the average of upward Tu and downward TdWater vapor Ta is equal to cloud liquid water Ta
(1.1)
(1.2)
(1.3)
(1.4)
Microwave Brightness temperature Eq.
Retrieval of TPW and CLW
CWITrTrPWI )ln()ln( 2V22
2V19
)ln( 2Tr)cos(/)( sLV CLWkTPWk
22L19V19L22V
2219L1922Ls
)log()log()cos(5.0
kkkk
TrkTrkTPW
)log()log( 237
219 HVHV TrTrCWI
}/)/()log{(}/)/()log{( 37s37Hs37V37H37V19s19Hs19V19H19V TTTTTTCWI
)( sT a function of SST
Determined to be maximize the correlation between TPW index and RAOB match-up TPW
From Eq.(1.2)
TTrTT )( sHsV2
HV TPW can be derived by absorption coefficients of water vapor kv and cloud liquid water kl by using two different frequency. However, it is not able to calculate kv and kl because these depend on vertical profile of temperature, water vapor and liquid water.
TPW
CLW
TrTTTT /)( asa
Theoretical calculation
)*1/()1937/()*)(cos(2
1s KRLRklklKRQWICWICLW
)1922/()1937( kvkvkvkvKR )1937/()1922( klklklklLR
A function decreased with TPWA constant
)ln()ln( 222
219 TrTrQWI
)/()cos( 1937s klkl Theoretically estimated
Updated TPW algorithm for AMSR-E
LUT in the algorithm was updated by using 3-yr RAOB and AMSR-E collocated dataset (2006-2008).Updated LUTs :
T850, Transmittance and Mean atmospheric temperature tableWind speed correction table and extended to strong wind condition beyond 20m/sConversion table PWI (Precipitable water index) to TPWCorrection coefficients on SST , SSW dependency of emissivityNo use of internal Tb conversion from ver.2 to ver.1 (JAXA L1B Tb version)
***NEW Num: 1349 Min: -18.836 Max: 19.008 Ave: -0.135 Std: 3.355
*** Current Num: 1344 Min: -18.532 Max: 15.366 Ave: 0.817 Std: 4.071
RAOB TPW
AM
SR
-E T
PW
RAOB TPW
AM
SR
-E T
PW
TPW Verification against RAOB (2009.1-5) Collocation criteria: Within 60min. 150km
[mm] [mm]
[mm
]
[mm
]
V003 vs GPS_PWV ( 2009 年 6 月 20 日~ 8 月 20 日)
(mm)
Ver. 003
Optimized by 3years RAOB TPW data2007 - 2009
(mm)
Ver. 004
Optimized by 3months GPS TPW dataJun.20 – Aug. 20, 2009
(mm)
Ver. 005(preliminary)