Observing System Simulation Experiments

1
Observing System Simulation Experiments

• Lars Peter Riishojgaard
• Global Modeling and Assimilation Office
• And
• Joint Center for Satellite Data Assimilation

2
Overview

• OSEs and OSSEs
• OSSEs for future missions
• The Molniya Orbit Imager
• OSSEs for missions in development
• Orbiting Carbon Observatory (OCO)

3
OSEs and OSSEs

• Observing System Experiment
• Typically aimed at assessing the impact of a
  given existing data type on a system
• Straightforward data denial or data addition
• Observing System Simulation Experiment
• Typically aimed at assessing the impact of a
  hypothetical data type on a hypothetical system
• Not straightforward EVERYTHING (including nature
  itself) must be simulated
• gt Costly in terms of computing and manpower

4
Applications

• OSSEs are developed for studying the
  characteristics and impact of non-existing
  observations
• Often the best shot at looking into the future
  prior to making a 500M decision
• Useful at any phase in a space program
• Preliminary investigations (e.g.MW or IR?)
• Definition (e.g. which orbit?)
• Design (trades)
• Pre-flight (user trades)

5
Case studies two OSSEs

• The Molniya Orbit Imager
• Candidate future mission
• Typical OSSE question
• What would mission_X do for application_Y?
• OCO (Orbiting Carbon Observatory)
• Mission in development
• Different type of questions, e.g.
• What is the right strategy for assimilating the
  observations from mission_X for
  application_Y?
• What are the prospects for generating and using
  unplanned_data_product_Z from mission_X?

6
Molniya Orbit Imager

• High-latitude quasi-geostationary imager (GOES
  to the pole) built on MODIS and GOES heritage
• Full Earth disc image every 15 minutes at 1 km
  (VIS channel) and 2 km (5 IR channels) horizontal
  resolution
• High-latitude winds gt improved weather forecasts
  (fewer busts) also at low latitudes
• Improved hurricane landfall forecasting

7
(No Transcript)
8
(No Transcript)
9
(No Transcript)
10
Molniya Orbit Imager OSSE

• Central issue
• What would high-latitude satellite wind
  observations add to mid-latitude numerical
  weather prediction skill?
• Simple question
• Observations of a well-known (existing) type, but
  over a previously unobserved region
• No easy fix
• Everything needs to be simulated
• Atmosphere
• Reference observations
• Perturbation observations

11
Molniya Orbit Imager (III)

• Main OSSE ingredients
• Nature run simulated atmosphere (NCEP)
• Simulated reference observations
• RAOBS, SATOBS, HIRS, AMSU, SYNOPS,
  Scattterometer, ACARS, etc.
• Simulated Molniya Orbit Imager winds
• Data Assimilation System (GEOS-4)
• Control run (everything but MOI data)
• Perturbation run (adding MOI data)

12
How to simulate observations?

• By sampling the nature run and adding noise
• In some cases this is straightforward, in others
  less so
• Simulated MOI winds would require a
  2km-resolution model run with realistic clouds
  and mid-tropospheric humidity
• Next-best thing sample the nature run winds with
  realistic distribution characteristics
• Rotate geostationary winds distribution over the
  pole
• Simulate the MOI viewing geometry

13
Molniya OSSE (Observing system simulation
experiment) GEOS-4 Atlas et al.
6-hour winds coverage, 4 LEOs ?
Apogee winds coverage, Molniya ?
Forecast improvement over North America, 48 cases
?
14
Orbiting Carbon Observatory

• Mission already in development under ESSP slated
  for 2008 launch
• precise, time-dependent global measurements of
  atmospheric carbon dioxide (CO2) from an Earth
  orbiting satellite (JPL)
• Hyperspectral near-IR instrument primary science
  data product is CO2 column
• Two questions (based on discussions with Steve
  Pawson, GMAO)
• Can we do source estimation from OCO? what is
  the right strategy for assimilation?
• Can OCO do surface pressure? with what kind of
  accuracy impact?

15
(No Transcript)
16
OCO (I)

• Question
• What observations do we need and how do we need
  to assimilate them in order to estimate surface
  fluxes?
• Ingredients
• Transport given by GCM and assumed to be perfect
• Imposed best-estimate fluxes
• Simulated OCO observations of the scenario given
  by 1. and 2.
• Using 1., attempt to recover 2. from 3.
• This can iterated until the question is answered
• e.g. data selection, density, cloudiness,
  radiance vs. retrieval is

17
OCO (II)

• Satellite data often provide the unexpected in
  terms of benefits and data products
• MODIS
• Tropospheric winds
• TOMS
• Aerosol
• Stratospheric winds

18
OCO (III)

• Could OCO provide surface pressure observations?
• Intriguing from a numerical weather prediction
  perspective
• Relatively easy to define and execute OSSE
• Simulate generic surface pressure observations
  with pertinent coverage and error characteristics
• A positive answer would have substantial
  implications
• for OCO algorithm development, processing and
  dissemination (latency)
• for the users new data type, benefits

19
Summary

• OSSEs are costly and complex, but they are
  essential test beds through all phases of the
  definition and development of space-based
  observing systems
• Two examples
• Molniya Orbit Imager (candidate mission)
• OCO (approved ESSP mission)
• NASA should not be funding and flying missions
  that have not proven their worth in an OSSE