GMAO Satellite Data Assimilation

Document Sample
GMAO Satellite Data Assimilation Powered By Docstoc
					GMAO Satellite Data Assimilation
                        Michele Rienecker
       Max Suarez, Ron Gelaro, Ricardo Todling, Emily Liu
   Yanqiu Zhu, Ivanka Stajner, Meta Sienkiewicz, Rolf Reichle
                Christian Keppenne, Robin Kovach




            Global Modeling and Assimilation Office (GMAO)
                 NASA/Goddard Space Flight Center




                      JCSDA SSC Meeting
                       May 30-31, 2007
Global Modeling & Assimilation Office
            http://gmao.gsfc.nasa.gov



   • Atmospheric Assimilation:
   • Atmospheric Assimilation:
       • NCEP’s GSI
        • NCEP’s GSI
       • AIRS
        • AIRS
       • Data impacts - Adjoint tools
        • Data impacts - Adjoint tools
       • MLS Ozone
        • MLS Ozone
   • Land Surface: EnKF
   • Land Surface: EnKF
   • Ocean: EnKF
   • Ocean: EnKF
   • Ocean Color: SEIK
   • Ocean Color: SEIK




                                         2
      GEOS-5 Atmospheric Data Assimilation System
                                                       Ricardo Todling, Max Suarez, Larry Takacs, Emily Liu


AGCM                                Analysis
    Finite-volume dynamic core            Grid Point Statistical Interpolation (GSI)
    Bacmeister moist physics              Direct assimilation of satellite radiance data
    Physics integrated under the          JCSDA Community Radiative Transfer Model
    Earth System Modeling
    Framework (ESMF)                      (CRTM) for most current instruments in space
    Catchment land surface model          GLATOVS for TOVS (HIRS2, MSU, SSU) on
    Prescribed aerosols                    board of TIROS-N, NOAA-06,…, NOAA-12
    Interactive ozone                     Variational bias correction for radiances

                                    ∂q n 
Assimilation                             
                                      ∂t  total
                                                 = dynamics (adiabatic ) + physics (diabatic ) + ∆q
                                   
    Apply Incremental Analysis
                                   Total “observed change”    Model predicted change     Correction from DAS
   Increments (IAU) to reduce
                                   00z   03z   06z    09z     12z    15z   18z   21z       00z     03z     06z
   shock of data insertion                                  Analysis
    IAU gradually forces the                                                           Raw analysis (from GSI)
                                                                                       Background (model forecast)
   model integration throughout                                                        Assimilated analysis
                                                              IAU                      (Application of IAU)
   the 6 hour period
                                                                                                              3
The next System - 4D-VAR
                     Background state




                                        Cost Function




    Analyzed state



                                                   4
Progress in 4D-VAR Development (Tremolet & Todling)


        1. Trajectory Model: GEOS-5 with full physics

        2. Model Adjoint: FV core with simple physics

        3. Extension of GSI components for 4D-VAR

           • Observation windowing flexibility
           • Observation handling (higher temporal-resolution bins)
           • Computation of time-dependent departures (OmF’s)
           • Preliminary version of model-analysis interface
           • Options for minimization algorithm

         4. Fine ⇔ Coarse mappings: ESMF



                                                                      5
                                                    MERRA
                                http://gmao.gsfc.nasa.gov/merra/

MERRA System                                                     EMPHASIS ON WATER CYCLE
                                                                       Global Precipitation,
                                                                     Evaporation, Land Hydrology,
1/2° × 2/3° × 72L to .01 mb
                                                                     Cloud parameters and TPW
1979-present
GSI Analysis with IAU                                            GLOBAL HEAT AND WATER BUDGETS
Parallel AMIP run                                                FOR ALL PROCESSES

                                                                 DIURNAL CYCLE FROM HOURLY 2-D
                                                                 FIELDS



     MERRA      77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07
     Stream 1
     Stream 2
     Stream 3
     ROSB
     G5-AMIP

                      Spinup years
                      Reduced Observing System Baseline (ROSB)




                                                                                                               6
                                                                       Date & Instrument
In-house Radiance Data                                                                               Bufr Library
                                                    Level-1B
      Processing                                Orbit Binary Files
                                                                                                      Bufr Table
                                                                                                      Calibration
Emily Liu

    In house data processing to support              HIRS2      MSU             SSU
    Modern Era Retrospective-analysis for
    Research and Applications (MERRA)
                                                     HIRS3      HIRS4          AMSUA            AMSUB        MHS
    Level-1b TOVS/ATOVS radiance data
    were converted to calibrated radiance in
    BUFR format with appropriate quality
    controls                                         Quality
    Data available from 1979 to present              Control
    Data blacklists from ECMWF ERA40,                                                                    Collect Data
    JMA25 reanalysis, and GMAO GEOS-4                                  Orbit BUFR Files                 6 hour window
    reanalysis (CERES) for further data
    screening
                                                                      Synoptic BUFR Files
    Can reprocess the radiance data if                               Four output files per day for         Combine
    calibration coefficients can be estimated                           each instrument type              BUFR Files
    from a better technique such as SNO
    (simultaneous nadir overpass)
                                                      Thinned                                              Warmest

    Receiving full spatial resolution AIRS
    and AMSU-A data from NESDIS
    Processing full resolution data set into
    thinned and warmest data sets in
    BUFR format
                                                                               Full Resolution                          7
The GEOS-5 ADAS Validation: Precipitation

     GEOS-5




                                            8
Adjoint tools for Observation Impact Studies
                                                                                                                        Ron Gelaro



      Efficient estimation of sources of forecast error and observation
    sensitivity (observation impact)

        • determined with respect to observational data, background fields
        or assimilation parameters, all computed simultaneously

        • useful for designing intelligent data selection strategies and
        guiding future observing system design
                                                               Forecast

                                                                                                               eb
                                                        observations
                                                                                                      t
                                                         assimilated                                as
                                                                                               ec              ea
                                                                                           for                      Forecast
                                                                                  u   nd
                                           Error                              gro
                                                                xb     ba
                                                                         ck               ast                       Error
                  DAS                                                                  rec
                                                                                   s fo
                                                                            l   ysi
                                                               xa        ana                              xv
                                     Sensitivity to
                                                      t − 6h     00Z                                      t +24h
                                      initial state
             Sensitivity to                                                            t=24h
             observations
                                       GEOS-5 Model Adjoint
          GSI Analysis Adjoint
                                                                                                                               9
  GEOS-5 used to Evaluate Impact of AIRS in NWP
                                                                            Emily Liu, Ron Gelaro, Yanqiu Zhu


Forecast Skill vs. Time

                     NH                                Data from most AIRS
                                                       channels improve




                                       Channel Index
                                                       numerical weather
         Control                                       forecasts
         Control + AIRS


                                                                                              Some AIRS
                      SH                                                                      channels
                                                                                              degrade the
                                                                                              forecast


                                                           Forecast Error Reduction (J/kg)

                      NH

        Control                  AIRS brings slightly positive impact on forecast skill in
        Control + AIRS without   Northern Hemisphere; clear positive impact in Southern
        moisture channels        Hemisphere. But forecast skills are increased when
                                 moisture channels from AIRS are not included

                                                                                                         10
          Diagnosing impact of hyper-spectral observing systems
                               GEOS-5 July 2005 00z Totals


          AMSU-A (15 ch)                                AIRS (153 ch)                    Negative
                                                                                          Impact

                                                          H2O Channels




                                            Channel
Channel




   -7.0          δe   (J/Kg)         0           -0.6      Forecast Error Reduction (J/Kg)0
                                                                δ e (J/Kg)

            …several AIRS water vapor channels currently degrade the 24h forecast
            in GEOS-5…
                                                                                              11
                       Comparison with OSEs
GEOS-5 Observation Impact: Comparison with OSEs                                                                  Ron Gelaro and Yanqiu Zhu

                         16


                         15
                                                                                       multiple OSEs

                         14
       24h Forecast
       Error Energy      13                                                                                           control
                                                                                                                      no airs
          (J/Kg)         12
                                                                                                                      no raobs
                         11                                                                                           no amsua16
                         10


                          9


                          8
                               1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
                                                                                                                     July 2005 00z
                         25

                         20
                                                                    control observation impact
                         15
       Observation
                         10
      Count (millions)
                          5

                          0




                                                                                  ft
                                                                                 bs




                                                                                 nd
                                                                                  s
                                                               a




                                                                                 ce
                                                                                 mi
                              a




                                            s
                             b




                                                   s

                                                              es




                          -5
                                                                     u



                                                                               ind
                                                             su
                           su




                                         air




                                                                              cra
                                                hir
                           su




                                                                   ms




             δe

                                                                             rfa
                                                                            rao
                                                            go




                                                                             ss




                                                                             sw
                                                           am
                         am

                         am




                                                                           tw


                         -10

                                                                           air
                                                                          sp




                                                                          su

                                                                          qk
                                                                         sa
                                                         s_




           (J/Kg)        -15
                                                       eo




                         -20

                         -25

                         -30
                                                                                                                                   12
Assimilating AURA/MLS ozone
                                                                        Meta Sienkiewicz and Ivanka Stajner
                                                            Zonal mean ozone 9/30/2004 00UTC
SBUV daytime only – no data near South                      SBUV only
Pole due to high solar zenith angle
MLS orbital limit ±82º




                                                            MLS only




           NOAA 16 SBUV
           MLS                       Ozone hole develops
                                      in MLS assimilation      Ozone partial pressure (mPa)
                                                                                                  13
     Global assimilation of AMSR-E soil moisture retrievals
                                                                                           Rolf Reichle




Assimilate retrievals of
surface soil moisture
from AMSR-E (2002-06)
into NASA Catchment
                                                           Validate with USDA SCAN stations
model (GEOS-5)
                                                           (only 23 of 103 suitable for validation)
                               Soil moisture [m3/m3]
                                 Anomaly time series correlation           Confidence levels:
Reichle et al.                   coeff. with in situ data [-]              Improvement of
JGR, 2007                        (with 95% confidence interval)            assimilation over
                           N      Satellite      Model        Assim.        Satellite      Model

Surface soil moisture      23      .38±.02       .43±.02      .50±.02       >99.99% >99.99%

Root zone soil             22        n/a         .40±.02      .46±.02          n/a       >99.99%
moisture
Assimilation product agrees better with ground data than satellite or model alone.
                                                                           14
Modest increase may be close to maximum possible with imperfect in situ data.
                                     Volumetric soil moisture (m3m-3)

Kumar, Reichle, et al. (2007), Adv. Water
Resources, submitted.                                                   15
                Forecast skill (ACC) from CGCMv1
                 Forecast skill (ACC) from CGCMv1
               Heat content anomaly in upper 300m
                Heat content anomaly in upper 300m
                           1993-2006
                            1993-2006

                  EnKF                    OI-TS



1-month lead




3-month lead




6-month lead




                                                     16
The impact of Argo - preparing for Aquarius
                                        Christian Keppenne and Robin Kovach




                                                                    17
GMAO’s Collaborations with JCSDA Partners
  Atmosphere:-
  Atmosphere:-
    •• GSI - NCEP
       GSI - NCEP
    •• Adjoint tools - NRL
       Adjoint tools - NRL
    •• AIRS
       AIRS
    •• Ozone
       Ozone
    •• Aerosols
       Aerosols
    •• OSSEs (emerging) - NCEP, NESDIS, et al
       OSSEs (emerging) - NCEP, NESDIS, et al

  Land Surface:-
  Land Surface:-
    • EnKF development
    • EnKF development
    • LIS implementation for Catchment and Noah LSMs
    • LIS implementation for Catchment and Noah LSMs
  Ocean:-
  Ocean:-
    • EnKF and MvOI development for MOM4 - NCEP
    • EnKF and MvOI development for MOM4 - NCEP
    • Altimetry with online-bias-estimation
    • Altimetry with online-bias-estimation
    • Ocean color
    • Ocean color

                                                       18
          GMAO - Near-term Plans
Atmosphere:-
Atmosphere:-
  •• Development of 4Dvar
     Development of 4Dvar
  •• Contribute to OSSE capability
     Contribute to OSSE capability
  •• AIRS (QC) - IASI - CrIS
     AIRS (QC) - IASI - CrIS
  •• Ozone - GOME-2 - OMPS
     Ozone - GOME-2 - OMPS
  •• Real-time MLS
     Real-time MLS
  •• MODIS Winds - VIIRS
     MODIS Winds - VIIRS
  •• CO, CO2 (OCO)
     CO, CO2 (OCO)

Land Surface:-
Land Surface:-
  • EnKF: Surface Temperature and Snow
  • EnKF: Surface Temperature and Snow
  • LIS implementation for Catchment and Noah LSMs
  • LIS implementation for Catchment and Noah LSMs
Ocean:-
Ocean:-
  • MOM4: retrospective analysis for seasonal forecast
  • MOM4: retrospective analysis for seasonal forecast
  • Surface Salinity
  • Surface Salinity
  • Ocean color: removing instrument biases
  • Ocean color: removing instrument biases
                                                         19