Observing System Simulation Experiments in the Joint Center for Satellite Data

1
Observing System Simulation Experiments in the
Joint Center for Satellite Data

• Lars Peter Riishojgaard and Michiko Masutani
• JCSDA

2
NCEP Michiko Masutani, John S. Woollen, Yucheng
Song, Stephen J. Lord, Zoltan Toth ECMWF Erik
Andersson KNMI Ad Stoffelen, Gert-Jan
Marseille JCSDA Lars Peter Riishojgaard
NESDIS Fuzhong Weng, Tong Zhu, Haibing Sun,
SWA G. David Emmitt, Sidney A. Wood, Steven
Greco NASA/GFSC Ron Errico, Oreste Reale, Runhua
Yang, Emily Liu, Joanna Joiner, Harper Pryor,
Alindo Da Silva, Matt McGill, NOAA/ESRLTom
Schlatter, Yuanfu Xie, Nikki Prive, Dezso
Devenyi, Steve Weygandt MSU/GRI Valentine
Anantharaj, Chris Hill, Pat Fitzpatrick, JMA
Takemasa Miyoshi , Munehiko Yamaguchi JAMSTEC
Takeshi Enomoto So far most of the work is done
by volunteers.
3
OSSEs

• Observing System Simulation Experiment
• Typically aimed at assessing the impact of a
  hypothetical data type on a forecast system
• Not straightforward EVERYTHING must be simulated
• Simulated atmosphere (nature run)
• Simulated reference observations (corresponding
  to existing observations)
• Simulate perturbation observations
• (object of study)
• gt Costly in terms of computing and manpower

4
Data assimilation
Nature (atmospheric state)
Assessment
End products
Sensors
Observations (RAOB, TOVS, GEO, surface, aircraft,
etc.)
Short range product
Analysis
Forecast model
Initial conditions
5
OSE, conceptual model
Nature (atmospheric state)
Assessment
End products
Sensors
Reference observations (RAOB, TOVS, GEO,
surface, aircraft, etc.)
Short range product
Analysis
Forecast model
Candidate observations (e.g. AIRS)
Initial conditions
6
OSSE, conceptual model
Nature run (output from high resolution, high
quality climate model)
Assessment
End products
Simulator
Reference observations (RAOB, TOVS, GEO,
surface, aircraft, etc.)
Forecast products
Analysis
Forecast model
Candidate observations (e.g. GEO MW)
Initial conditions
7
Role of a National OSSE Capability

• Impact assessment of future missions
• Decadal Survey and other science and/or
  technology demonstration missions (NASA)
• Future operational systems (NOAA)
• Objective way of establishing scientifically
  sound and technically feasible user requirements
  for observing systems
• Tool for assessing performance impact of
  engineering decisions made throughout the
  development phases of a space program or system
• Preparation/early learning pre-launch tool for
  assimilation users of data from new sensors

8
Why a Joint OSSE capability?

• OSSEs are expensive
• Nature run, entire reference observing system,
  additional observations must be simulated
• Calibration experiments, perturbation experiments
  must be assessed according to standard
  operational practice and using operational
  metrics and tools
• OSSE-based decisions have many stakeholders
• Decisions on major space systems have important
  scientific, technical, financial and political
  ramifications
• Community ownership and oversight of OSSE
  capability is important for maintaining
  credibility
• Independent but related data assimilation systems
  allows us to test robustness of answers

9
Main OSSE components

• Data assimilation system(s)
• NCEP/EMC GFS
• NASA/GMAO GEOS-5
• NCAR WRF-VAR
• Nature run
• ECMWF
• Plans for embedded WRF Regional NR
• Simulated observations
• Reference observations
• Perturbation (candidate) observations
• Diagnostics capability
• Classical OSE skill metrics
• Adjoint sensitivity studies

10
ECMWF Nature Run (Erik Andersson)

• Based on recommendations/requirements from JCSDA,
  NCEP, GMAO, GLA, SIVO, SWA, NESDIS, ESRL
• Low Resolution Nature Run
• Free-running T511 L91 w. 3-hourly dumps
• May 12 2005 through June 1 2006
• Two High Resolution periods of 35 days each
• Hurricane season Starting at 12z September
  27,2005,
• Convective precipitation over CONUS starting at
  12Z April 10, 2006
• T799 L91 levels, one-hourly dump
• Initial condition from T511 NR

11
Nature Run validation

• Purpose is to ensure that pertinent aspects of
  meteorology are represented adequately in NR
• Contributions from Emmitt, Errico, Masutani,
  Prive, Reale, Terry, Tompkins and many others
• Clouds
• Precipitation
• Extratropical cyclones (tracks, cyclogenesis,
  cyclolosis)
• Tropical cyclones (tracks, intensity)
• Mean wind fields
• .

12
Initial Nature Run validation
Study of drift in NR Michiko Masutani (NCEP)
Area averaged precipitation
Tropics
Zonal wind June 2006 By Juan Carlos Jusem
(NASA/GSFC)
NCEP reanalysis
Nature Run
Convective precipitation Large Scale
precipitation Total precipitation
Two to three weeks spinup in tropical
precipitation. - Michiko Masutani (NCEP/EMC)
13
Tropical cyclone NR validation Preliminary
findings suggest good degree of realism of
Atlantic tropical cyclones in ECMWF NR.
HL vortices vertical structure
Vertical structure of a HL vortex shows distinct
eye-like feature and prominent warm core
low-level wind speeds exceed 55 m/s
Reale O., J. Terry, M. Masutani, E. Andersson,
L. P. Riishojgaard, J. C. Jusem (2007),
Preliminary evaluation of the European Centre for
Medium-Range Weather Forecasts' (ECMWF) Nature
Run over the tropical Atlantic and African
monsoon region, Geophys. Res. Lett., 34, L22810,
doi10.1029/2007GL031640.
14
Extratropical Cyclone StatisticsJoe Terry
NASA/GSFC
1) Extract cyclone information using Goddards
objective cyclone tracker

• Nature Run
• One degree operational NCEP analyses (from
  several surrounding years)
• NCEP reanalysis for specific years (La Nina, El
  Nino, FGGE)

2) Produce diagnostics using the cyclone track
information
(comparisons between Nature Run and NCEP analyses
for same month)

• Distribution of cyclone strength across pressure
  spectrum
• Cyclone lifespan
• Cyclone deepening
• Regions of cyclogenesis and cyclolysis
• Distributions of cyclone speed and direction

15
(No Transcript)
16
Evaluation of Cloud Simpson weather associates
17
Case Events Identified from ECMWF HRNR(Plotted
from 1x1 data)

• May 2-4 squall line affecting all points along
  US Gulf coast

MSLP (hPa) 3-h convective precipitation (mm)
.
May 7-8 decaying squall line over TX Oct
10-11 squall line / tropical wave
Christopher M. Hill, Patrick J. Fitzpatrick,
Valentine G. Anantharaj Mississippi State
University
18
Simulation of observations

• Conventional observations (non-radiances)
• Resample NR at OBS locations and add error
• Problem areas
• Atmospheric state affects sampling for RAOBS,
  Aircraft observations, satellite AMVs, wind
  lidars, etc.
• Correlated observations errors
• J. Woollen (NCEP), R. Errico (GMAO)
• Radiance observations
• Forward radiative transfer on NR input profiles
• Problem areas
• Treatment of clouds has substantial impact on
  availability and quality of observations
• Desire to avoid identical twin RTMs
• H. Sun (NESDIS), R. Errico (GMAO)

19
OBS91L Jack Woollen (NCEP/EMC)
For development purposes, 91-level NR variables
are processed at NCEP and interpolated to
observational locations with all the information
need to simulate data (OBS91L). OBS91L for all
footprints of HIRS, AMSU, GOES are produced for a
few weeks of the T799 period in October
2005. Thinned footprints for the entire period.
Thinning of the footprint is based on
operational use of radiance data. The OBS91L
are also available for development of a Radiative
Transfer Model (RTM) for development of other
forward model.
20
Radiance Simulation System for OSSEGMAO, NESDIS,
NCEP Ron Errico, Runhua Yang, Emily Liu, Lars
Peter Riishojgaard (NASA/GSFC/GMAO) Tong Zhu,
Haibing Sun, Fuzhong Weng (NOAA/NESDIS)Jack
Woollen(NOAA/NCEP)
Other resources and/or advisors David Groff ,
Paul Van Delst (NCEP) Yong Han, Fuzhong
Weng,Walter Wolf, Cris Barnet, Mark Liu
(NESDIS) Erik Andersson (ECMWF) Roger Saunders
(Met Office)
NASA/GMAO developing optimized strategies to
simulate complete set of footprints. This
includes development of cloud clearing algorithm.
NESDIS, NCEP working on thinned data. Full
resolution data for GOES-R. Initial data set
(OBS91L) produced by Jack Woollen at NCEP
Existing instruments experiments must be
simulated for control and calibration and
development of DAS and RTM Test GOESR,NPOESS, and
other future satellite data
21
OBS91L Jack Woollen (NCEP/EMC)
For development purposes, 91-level NR variables
are processed at NCEP and interpolated to
observational locations with all the information
need to simulate data (OBS91L). OBS91L for all
foot prints of HIRS, AMSU, GOES are produced for
a few weeks of the T799 period in October
2005. Thinned foot prints for the entire period.
Thinning of the foot print is based on
operational use of radiance data. The OBS91L
are also available for development of a Radiative
Transfer Model (RTM) for development of other
forward model.
22
Radiance Simulation System for OSSEGMAO, NESDIS,
NCEPTong Zhu, Haibing Sun, Fuzhong
Weng(NOAA/NESDIS)Jack Woollen(NOAA/NCEP)Ron
Errico, Runhua Yang, Emily Liu, Lars Peter
Riishojgaard (NASA/GSFC/GMAO)
Other resources and/or advisors David Groff ,
Paul Van Delst (NCEP) Yong Han, Walter Wolf, Cris
Bernet,, Mark Liu, M.-J. Kim, Tom Kleespies,
(NESDIS) Erik Andersson (ECMWF) Roger Saunders
(Met Office)
OBS91L is produced by Jack Woollen at
NCEP NASA/GMAO is developing best strategies to
simulate and work on complete foot prints. This
include development of cloud clearing
algorithm. NESDIS and NCEP are working on
thinned data. Full resolution data for GOESR.
Existing instruments experiments must be
simulated for control and calibration and
development of DAS and RTM Test GOESR,NPOESS, and
other future satellite data
23
Simulation of GOES-R ABI radiances for OSSE Tong
Zhu et al. 5GOESR P1.31 at AMS annual
meeting http//www.emc.ncep.noaa.gov/research/Join
tOSSEs/publications/AMS_Jan2008/Poster-88thAMS2008
-P1.31-OSSEABI.ppt
Simulated from T511 NR. GOES data will be
simulated to investigate its data impact
24
Current set of prototype simulated observations
at the GMAO derived from the ECMWF Nature
Run
HIRS3, HIRS2, AMSU-A/B, AIRS, Conventional
Obs.
The satellite data is thinned, but less so than
used operationally. Thinning is based on time of
report and defined effect of clouds. Clouds are
treated as black at their tops, when defined as
present. The presence of clouds affecting IR is
determined by a tunable stochastic function
using NR-provided cloud fractions. This
function is intended to account for holes in
grid-boxes and allows simple tuning for
possible unrealism in the NR cloud
distribution. The same CRTM is used as in GSI
(the only RTM available to us). Locations of
cloud track winds are independent of NR
clouds. The list of simulated obs. types will be
expanded along with the realism of the
simulations and their associated errors as
resources permit. We hope to have a suitably
tuned (validated) set of prototype simulated
obs. available by the end of Sept. 2009.
Slide from Errico
25
OSSEs planned
OSSEs to investigate data impact of GOES and
prepared for GOES-R Tong Zhu, Fuzhon Weng, J.
Woollen (NCEP) M.Masutani(NCEP) and more
OSSE to evaluate UAS N. Prive(ESRL), Y.
Xie(ESRL) possible at NCEP and others
OSSE to evaluate DWL M.Masutani(NCEP), L. P
Riishojgaard (JCDA), NOAA/ESRL, and others
OSSEs for THORPEX T-PARC Evaluation and
development of targeted observation Z. Toth,
Yucheng Song (NCEP) and other THORPEX team
Regional OSSEs to Evaluate ATMS and CrIS
Observations Cris M. Hill, Pat. J. Fitzpatrick,
Val. G. Anantharaj GRI- Mississippi State
University (MSS) Lars-Peter Riishojgaard
(NASA/GMAO, JCSDA)
26
Next steps

• Calibration impact of main classes of
  observation should mimic what is seen in
  operational OSEs
• GMAO will calibrate GEOS-5 using adjoint
  sensitivity tools
• EMC will calibrate GFS OSSE using OSEs
• Goal is to have calibrated systems available for
  actual OSSEs by late summer 2008
• Funding
• NASA ROSES proposal
• NOAA JCSDA-led budget initiative
• ESA/EUMETSAT encouraged by ADMAG to participate

27
Summary

• OSSEs are expensive, but can be a cost-effective
  way to optimize investment in future observing
  systems
• OSSE capability should be multi-agency, community
  owned to avoid conflict of interest
• Independent but related data assimilation systems
  allows us to test robustness of answers
• Joint OSSE collaboration remains only partially
  funded but appears to be headed in right direction