Overview of Data Assimilation at NCMRWF

1
Overview of Data Assimilation at NCMRWF M.
Das Gupta, V. S. Prasad, John P. George and E. N.
Rajagopal NCMRWF
2

• Outline
• Observation Pre-processing
• Data coverage
• Data monitoring
• Data Assimilation
• Global
• Meso-scale
• Observing system experiments few results
• Future Plan

3
Observations Pre-processing Reception of
Global observations - GTS via RTH
New Delhi , half-hourly
files
dedicated link
90MB/day -
internet download through ftp
satellite observations 550
MB/day
Volume of data reception through GTS (MB/day)
4
Decoders operational at NCMRWF (i)
SYNOP/SHIP/SYNOP-MOBIL (ii)
TEMP/TEMP-SHIP/TEMP-DROP/TEMP-MOBIL (iii)
PILOT (iv) BUOY (v) AIREP/AMDAR/ACARS
(vi) SATOB (vii) BUFR RDB -
data base structure ECMWF since 1990
BUFR-tank - data base structure NCEP from
2007 - modules developed for bypassing
Unidata-LDM
5
Global coverage of observations on a typical day
Surface
Upper-air Sonde
AIRCRAFT
Satellite Winds
6
Satellite Sounding
Scatterometer winds (QSCAT)
comparison of real time data reception
at NCMRWF UKMO -
August 2006 conventional observations -
comparable
SYNOP aircraft less at
NCMRWF satellite observations - less at
NCMRWF
7
Comparison of observation received at NCMWF
UKMO (Average for 1st -14th August 2006 )
8
(No Transcript)
9
Data monitoring report - monthly report (
CBS recommended format ) - weekly report
available at NCMRWF web page special
emphasis on monitoring of Indian observations
10
Global Data Assimilation
11
Global Data Assimilation system six hourly
intermittent scheme 4 times a day - 0000,
0600, 1200,1800 UTC 3 hours window SSI
analysis (NCEP) T80L18 - June 1994
- OSEs conducted for various
insitu and satellite
observations T254L64 - Jan 2007
12
Implementation of high-resolution assimilation
scheme T254L64 (50km x 50 km , 64 levels)
since January 2007
Internet
13
Comparison of observations used in T254L64 and
T80L18 GDAS system
14
Radiance AMSU-A HIRS (NOAA 16 18 Metop),
AMSU-B(NOAA 16 18 )
Ozone (SBUV Total ozone) from NOAA 16 18

• downloaded from NESDIS site (gp12)
• level-1B data - channel separated raw
  radiance with ancillary information like Earth
  Location Points (ELPs), solar zenith angle and
  calibration etc.
• processed for time-window, geo-location,
  calibrated radiances
• Radiative transfer model - CRTM (Community
  Radiative Transfer Model developed
  maintained by JCSDA )

NOAA 16 data coverage
15
AMSU-A Brightness Temperature (ch-1) for a
typical day 15 Jul 2007 a) Satellite observed
(NOAA-16 and NOAA18) b) Difference between
Model simulated and Observed BTs.
16
168 hr. FCST WITHOUT RADIANCE VT 0000UTC
29th April 2007
VERIFYING ANALYSIS IC 0000UTC 29th April
2007
168 hr. FCST RADIANCE
VT 0000UTC 29th April 2007
17
averaged over 1st 30th APRIL 2007, 0000 UTC run
18
OSE with NOAA radiance at T62L64 resolution
for July 2007
850 HPa wind spd. over Indian region
July 07
Forecast verification against observations
GPCP observed pptn./day July 2007
with RAD
without RAD
19

• Tropical Cyclone Relocation ( GFS based - Liu
  et al. 2000)
• relocation of vortex in guess field (modified
  guess)
• insertion of synthetic vortex
• Nargis (27th April 3rd May 2008) - Very
  Severe Cyclonic Storm crossed southwest coast of
  Myanmar

First guess (T254L64) valid for 0000 UTC 29th
April
original modified guess height
Original Guess
Modified Guess
20
Analysis (T254L64) for 0000 UTC 29th April
after TC relocation
before TC relocation
observed position 13N/85.5E
21
FCST Track IC 0000UTC 29.04.08
22
Improvement in analyses T254L64 vs. T80L18
RMSE (Analyses against RS/RW observations)
00UTC July 2007
Global
23
RMSE (Analyses against RS/RW observations)
00UTC July 2007
Tropics
24
Meso-scale data assimilation
25

• MM5 -3DVAR - 2004 (experimental)
• - carried out several experiments
• impact of satellite observations (ATOVS, QSCAT
  ..)
• assimilation of horizontal winds from Doppler
  Radar at Calcutta
• WRF-3DVAR - since 2006
• carried out thorough comparison of 3DVAR vs.
  NOVAR for monsoon 2006

WIND OBS
TRMM RAIN OBS
RAIN without DWR
RAIN without DWR
RAIN with DWR
26
WRF Domain
Verification of daily Analyses against
observations July-August-Sept 2006 RMSE ( u,
v, t)
3DVAR NOVAR
u
v
t
27
Observing system experiments
Impact of various types of observations on
Monsoon and Tropical cyclone
28
x
Impact of METEOSAT-5 AMVs on Monsoon Simulation
(1999)
With METEOSAT-5
Without METEOSAT-5
Longitude-pr. cross-section of v-comp.of wind at
equator (Mean Monthly Analysis 00UTC June 1999)
Conclusion Depth of south westerly flow
associated with monsoon over Arabian sea
increased after utilization of data producing
better rainfall prediction over west coast of
India
29
Impact of METEOSAT winds on TC Track Prediction
Best Track (JTWC) of Tropical Cyclone Pyarr ( 16
20th Sept. 2005)
30
With METEOSAT
Control
Assimilation experiment with high-density
METEOSAT winds
00UTC16th Sept ANA
D2 FCST VT 00UTC18th
D3 FCST VT 00UTC19th
850 hPa Analysis on 00UTC 16th and subsequent
predictions Stream lines Wind speed (m/s)
D4 FCST VT 00UTC20th
Verifying Analysis 00UTC 20th
31
Experiments with Ocean Surface Winds
32
exp
Impact of ERS-2 scatterometer surface wind data
on analysis of Kandla Cyclone (0000UTC 08June
1998) ( MAUSAM, 2002 ,53)
925
850
500
700
Analysis with ERS2 data
Analysis without ERS2 data
925
850
Conclusion Impact on analysis is seen to be
confined only in the lower levels
700
500
33

Impact of satellite derived surface wind speed
data on analysis of Orissa Super cyclone
(Oct1999) (The Global Atmosphere and Ocean
System, 2003 )
Most intense circulation in analysis with SSM/I
data
34
Impact of QSCAT(1ºx1º) on TC (MUKDA 22-24SEP
2006) Track
Analysis 00z 20SEP06 850 hPa wind Without
QSCAT With QSCAT
QSCAT OBSN 00z 22 SEP 2006
48hr. FCST IC 00z200906 VT00z220906 Without
QSCAT With QSCAT
35

• Impact of Buoy observations on monsoon simulation

36
July 2002
July 2003
37
Root Mean square error (RMSE) of analyses
computed against RS/RW observations
July 2002 Geop Ht (z)
Conclusion With the present distribution of
buoys over Indian Ocean and adjoining Seas,
impact of buoy observation on subsequent
prediction is seen to be mainly restricted over
southern hemispheric Indian Ocean Buoy
observation, though basically surface
observations, has impact in the upper level
analyses of various meteorological parameters,
which is due to assimilation of surface pressure
observations from buoy Buoy data has shown
positive impact on the analysis of lower level
westerly jet near the west coast of India
38
Synthetic Vortex (bogussing) for tropical
systems over Indian seas with T No. gt 2.5
Impact of Synthetic vortex on the track
prediction of a Tropical storm using a Global
Spectral Model
Synthetic wind observations (Hollands Empirical
model)
Bay of Bengal very severe cyclonic storm
26th -30th,
November 2000
Observed track and predicted tracks of
tropical cyclone based on 00UTC of 26-Nov-2000
initial condition
39
limitation
without Syn vortex
with Syn vortex
Limitations (dipole)
Impact of Synthetic Vortex on T80/L18 Data
Assimilation Forecast System Track Error (BOB
10 cases )
40
Arabian Sea Monsoon Experiment ARMEX-I (15th
June-15th Aug 2002)
Better rainfall prediction over Gujarat
ARMEX
CTRL
Observed Rainfall
41
Future plan in data assimilation Grid Point
Statistical Analysis (GSI NCEP) Assimilation
of INSAT-3D ( sounder imager channels)
Oceansat (scatterometer RO )
Megha-tropiques ( SAPHIR - humidity profile ..,
MADRAS ocean surface winds,
integrated water vapour .) 4D-VAR (UKMO)
42
Thanks