Assimilation of satellite data for mesoscale modeling

1
 Assimilation of satellite data for mesoscale
modeling John. P. George and Munmun Das
Gupta   National Centre for Meduim Range Weather
Forecasting (NCMRWF) (Department of Science and
Technology, Govt. of India) Noida (U.P) ,
INDIA-201307
2

• OUTLINE
• 1. NCMRWF models and data assimilation system
• 2. Regional Data Assimilation System at NCMRWF
• 3. Assimilation-forecast experiments with
  different satellite data

3
National Centre for Medium Range Weather
Forecasting (NCMRWF) is the premier institution
in India to provide Medium Range Weather
Forecasts through deterministic methods and to
render Agro Advisory Services (AAS) to the
farmers. Application Researh on Numerical
Weather Prediction Diagnostic studies Crop
Weather Modeling Computer Science.
4
NWP Models at NCMRWF

• Global Model

• T80 at 150x150 km resolution Operational
• T170 at 75x75 km resolution Experimental

• Mesoscale Models

• MM5 (Nested 90, 30, 10 km)Operational
• ETA at 48 km resolution Operational
• Regional Spectral Model at 50 km resolutio
  
• Experimental

5
  
Aircraft
RS/RW

• Global Data assimilation System (GDAS)
  operational at NCMWRF
• 6-hrly intermittent
• 3D-VAR analysis (SSI)
• conventional and satellite obs.

PilotBalloon
Surfaceobservations
GTS DATA
Ships Buoys
Satellite data
RTHNEW DELHI
FTP Satellite Data MSMR, SSMI etc.
DATA RECEPTION DECODING at NCMRWF(1/2 hrly)
DATA PROCESSING AND QUALITY CONTROL

• Satellite data assimilated at NCMRWF
• CMVs (AMVs) from GOES,
  METEOSAT,GMS and Kalpana/INSAT
• High resolution winds from METEOSAT-5(63ºE)
• ATOVS (120km) temperature and humidity
  profiles
• SSM/I wind speed
• QSCAT winds

PREVIOUS (6 HOURS)ANALYSIS
Repeated four times a day 00,06,12 18 UTC
PREVIOUS (6HR.)ANALYSIS
DATA PROCESSING QUALITY CONTROL
SSI ANALYSIS
SURFACEBOUNDARYCONDITIONS
ANALYSIS
SURFACE BOUNDARYCONDITIONS
T80 GLOBAL SPECTRAL FORECAST MODEL
MEDIUM RANGE WEATHER FORECAST BASED ON 00 UTC
ANALYSIS
6
Global observations received at NCMRWF
(through GTS and ftp) in January-2005
7
Comparison of global observations received at
NCMRWF (through GTS and ftp) ECMWF in
January-2005
8
Regional Assimilation- Forecast system at
NCMRWF - MM5 Model (since 2002- with
interpolated global model analysis) /
WRF(2005) - NCAR 3DVAR (2005)
9

• Regional Assimilation- Forecast system at NCMRWF
• NCMRWF is running MM5 Model (NCAR) in real time
  basis since 2002
•  Domain
•      Horizontal (Triple Nested)
• Vertical 23 Levels (Sigma-Hybrid)
• Time Steps Domain-1 270 S,
• Domain-2 90 S,
• Domain-3 4 30 S
• Topography USGS (Interpolated depending on
  resolution)
• Vegetation/ Land use 25 Categories (USGS)
• Initial and lateral boundary conditions are from
  NCMRWFs global model
• Boundary conditions are updated every 12 hours.

10
(No Transcript)
11

• NCAR MM5/WRF-3DVAR at NCMRWF
• MM5-3DVAR system mainly consists of the following
  four components
• (a)      Background Pre-processing
• Observation Pre-processing and quality
  control
•     Variational Analysis
• Updation of Boundary Conditions

3DVAR has been implemented as 6-hrly
intermittent scheme with 3UTC window
(one time)
12
(a) Background Preprocessing -Terrain (Defines
domain, orography, land use etc.) -Pregrid
(reads background forecast) -Regridder
(Horizontal interpolation of background
forecast) -Interpf (Vertical interpolation of
background forecast) (b) Observation
Preprocessing Observation Preprocessor prepares
the observation in a form which can be ingest
into 3DVAR - Preparation of Background error
covarience statistics (Once) (c ) 3DVAR (d)
Update the boundary condition -Using new analysis
13
Overview of MM5/WRF 3DVAR
Namelist File
Xb
BE
Yo
Setup Observations
Setup Background Errors
Read Namelist
Setup Background
3DVAR START
Compute Analysis
Calculate (O B)
Minimise Cost Function
Outer Loop
Output Analysis
Calculate Diagnostics
3DVAR END
Diagnostic File
Xa
14
Basic aim of MM5/WRF - 3DVAR is to produce an
optimal analysis through iterative solution of
where x analysis state xb ,
background yo observation
y H(x ) B, E and F are the background,
observation (instrumental) and representivity
error covariance matrices respectively
15

• Practical implementation of 3DVAR requires
  simplifications
• Simplified error covariances.
• Linearized observation operators, balance
  equation.
• Thinning of observations.
• Suitable choice of analysis control variables
• etc.

16

• Control variables
• In MM5/WRF there are three choices for the
  control variable
• cv_option 1
• U-component of wind
• V-component of wind
• Temperature
• Pressure
• Moisture variable as specific
  humidity or relative humidity
• cv_option 2
• Stream function (?)
• Velocity potential (?)
•    Unbalanced part of pressure (Pu)
• Moisture variable as specific humidity
  or relative humidity
• cv_option 3
• Stream function (?)
• Unbalanced part of velocity potential (?u)
• Unbalanced part of temperature (Tu)
• Log of surface pressure

17
About the Experiment MM5/WRF-3DVAR assimilation
cycle (6 hr intermittent) has been run a period
of 12 days (0006UTC 21st - 0000 UTC31st July
2004) Conventional data such as SYNOP, SHIP,
BUOY, AIREP, AMDAR, TEMP, PILOT and SATOB used
(CRTL)
ATOVS Temperature Humidity profile SSM/I
Sea surface wind speed
Total precipitable water vapor QSCAT Sea
surface wind direction and speed GPS-RO
Refractivity profile (During March-2004)
18
Coverage of conventional data used in the
assimilation cycle
SYNOP
BUOY
PILOT
TEMP
19
Variation of RMSE over the cyclic assimilation
period (21st 31st July 2004) of Back ground
Observation (OI) Analysis Observation (AO)
computed against RS/RW data
20
Variation of Bias over the cyclic assimilation
period (21st -31st July 2004) of Back ground
-Observation (OI) Analysis -Observation (AO)
computed against RS/RW data
21
METEOSAT -5 00 UTC image 27 July 2004(L) 28
July 2004(R) 29 July 2004(B)
22
(No Transcript)
23
Coverage of ATOVS data on a typical day
24
 
Analysed height and wind fields for CRTL ATOVS
run 850 hPa 00UTC 27th 28th 29th July 2004
25
24, 48 and 72 hr. forecasts of ht. wind fields
for CRTL ATOVS run 850 hPa based on 00UTC
26th July 2004
26
Results In ATOVS run, the centre of
circulation in wind field coincide the centre of
low in height field Utilisation of ATOVS-
Improves the track prediction
27
Coverage of SSSM/I data
28
Analysed height and wind fields for CRTL SSM/I
run 850 hPa 00UTC 27th 28th 29th July 2004
24, 48 and 72 hr. forecasts of ht. wind fields
for CRTL SSM/I run 850 hPa based on
00UTC 26th July 2004
29
Results Winds over Bay of Bengal are stronger
in SSMI analysis Though the position of
the system in forecasts are not very different
in CTRL and SSMI run, but the intensity
of the system is stronger in SSMI
30
Coverage of QSCAT data on a typical day
31
Analysed height and wind fields for CRTL QSCAT
run 850 hPa 00UTC 27th 28th 29th July 2004
24, 48 and 72 hr. forecasts of ht. wind fields
for CRTL QSCAT run 850 hPa based on
00UTC 26th July 2004
32

• Results
• Structure of the cyclonic system over Bay of
  Bengal region in QSCAT analysis is better defined
  and also stronger than that of CTRL analysis at
  850 hPa.
• The system is predicted much stronger in QSCAT
  run (forecast) compared to that of CTRL. This
  emphasize that the further tuning is required
  before utilising QSCAT data in the assimilation
  system

33
Assimilation of Global Positioning System (GPS)
data The GPS consists of a constellation of
satellites which transmit on two L-band
frequencies (1575.42 MHz for L1 and 1227.6 MHz
for L2). These two signals are delayed as
they propagate through the atmosphere due to the
presence of atmospheric water vapor. This "wet
delay" is detectable from the GPS phase
observations at the fixed ground receiver
stations and can be transformed into an estimate
of the perceptible water vapor (PW) present in
the troposphere above that location. GPS
radio occultation technique - When radio waves
from the GPS satellite (L1 L2) pass through the
atmosphere, either during a rise event or a set
event as seen from the receiver on the low earth
orbit (LEO) satellite, they are refracted through
an angle determined by the refractivity gradients
along the path. These, in turn, depend on the
gradients of air density (and hence temperature),
water vapor and electron density.
34
GPS Radio Occultation Measurements Processing
Phase Amplitude of the Signal ---gt Bending
Angle ---gt Refractivity ---gt T Q
35
GPS RO Measurement Processing
36

• Characteristics of GPS Radio Occultation (RO)
  Data
• Global 3-D coverage
• High accuracy
• High vertical resolution ( 100 m in lower
  troposphere)
• All weather-minimally affected by aerosols,
  clouds or prec.
• Independent height and pressure
• Requires no first guess sounding
• Independent of radiosonde calibration
• No instrument drift
• Compact, low-power, low-cost sensor
• No satellite-to-satellite bias

37
?? There are considerable uncertainties in
global analyses over data void regions (e.g.,
where there are few or no radiosondes),
despite the fact that most global analyses now
make use of satellite observations. GPS RO
missions (such as COSMIC) can be designed to
have globally uniform distribution (not limited
by oceans, or high topography). ?? The accuracy
of GPS RO is compatible or better than radiosonde
and can be used to calibrate other observing
systems. ??
38
3DVAR Assimilation Experiment - CHAMP RO Data
over Indian Region
(a)
(b)
500hPa (a) wind height and (b) Relative humidity
increments of NCAR-3DVAR assimilation (18UTC13
March 2004) with GPS refractivity data
39
3DVAR Assimilation Experiment - CHAMP RO Data
over Indian Region
Table.(a). OB (Observation-Background)
statistics of GPS refractivity (CHAMP Data)
assimilation - NCAR-3DVAR (Background - NCMRWF
T80 analysis)
Table.(b). AO (Analysis-Observation) statistics
of GPS refractivity (CHAMP Data) assimilation -
NCAR-3DVAR
Assimilation of GPS refractivity profile shows
improvement in AO
40

• Results
• Small scale wind features are more prominent in
  CTRL analysis compared to interpolated global
  analysis (IGLB)
• Mismatch between circulation center in wind and
  height filed, as seen in case of Bay of Bengal
  circulation, in CTRL analyses is reduced
  considerably with utilization of ATOVS data
• SSM/I and QSCAT data intensify the circulation in
  Bay of Bengal both in analysis as well in
  forecast (unrealistic). This emphasizes the need
  of proper tunings before assimilation of these
  data.
• Assimilation of GPS refractivity profile shows
  improvement in AO

41
Expected observations from Megha-tropique
satellite

• Megha-tropique satellite is proposed to carry
  three scientific instruments
• Multi-frequency Microwave Scanning Radiometer,
  MADRAS
• Surface winds
• Ocean rain
• Cloud liquid water content
• Deep convective areas
• Cloud top ice, anvil areas
• Multi-channel Microwave Instrument, SAPHIR
• Humidity profile
• Multi-channel instrument, SCARAB
• Radiation budget measurmentsTotal and SW
  radiation measurements

42
MADRAS is a microwave imager, with conical
scanning (incidence angle 56). The main aim of
the mission being the study of cloud systems, a
157 GHz channel is present in order to study the
high level ice clouds associated with the
convective systems.
43
SAPHIR is a sounding instrument with 6 channels
near the absorption band of water vapor at 183
Ghz. These channels provide relatively narrow
weighting functions from the surface to about 10
km
ScaRaB is a scanning radiative budget
instrument.The basic measurements of ScaRaB are
the radiances in two wide channels, a solar
channel (0.2 - 4 µm), and a total channel (0.2 -
200 µm)
44
Following parameters can be used as an input to
the NCMRWF assimilation- forecast system-
Ocean surface wind, integrated water vapor
and ocean rain (MADRAS) Water vapor profiles in
the cloud free troposphere (SAPHIR). Use of
these parameters in our assimilation system may
improve the distribution of the water vapor over
the tropical oceans in our analysis, which may
ultimately improve the convection and other
precipitation processes in the model. Cloud
liquid water and ice (MADRAS) can also be used an
input to the model, which can improve the
computation of cloud optical properties (input to
the radiation scheme) in the model and hence the
radiation fluxes and heating/cooling rates.
45
Shortwave and Longwave radiation (ScaRaB)
measurement can be used for the validation of the
radiation scheme over the tropical areas. The
radiative fluxes observation in this mission is a
valuable data to validate the model generated
cloud radiative forcing. Deep convection
areas, cloud liquid water, precipitation, cloud
top ice, anvil areas and humidity profiles
(SAPHIR, MADRAS) can be used for the validation
of the parameterization of convection and other
precipitation processes in the model.
46

• Future Scenario (NWP Global model)
• Increased horizontal and vertical resolution
• Horizontal resolution 8-15 km (2015), 3-5 km
  (2025)
• Vertical resolution
• Boundary layer 70m (2015), 40m (2025)
• Free atmosphere 300m (2015), 200m (2025)
• Stratosphere 500m (2015), 200m (2025)