NWP Perspectives

1
NWP Perspectives

• Nancy L. Baker
• Marine Meteorology Division
• Naval Research Laboratory
• Monterey, CA
• baker_at_nrlmry.navy.mil

2
Expanded View of NESDIS Products and Customers

• Retrieved products and radiances.
• User notification of system changes
• Common and consistent data formats
• Communications timeliness and availability of
  data
• Software and scientific knowledge/support
• Web pages for system information and
  documentation
• Validation products, radiances, formatted data

3
Expanded View of NESDIS Products and Customers

• Quality indicators for each stage of processing
• Distribution center for non-NOAA polar orbiter
  (e.g., AIRS)
• Community activities support for ITWG support
  for frequency protection and associated
  sensitivity studies
• Where does NESDIS responsibility end? Once the
  product is out the door, or once the product has
  been successfully monitored operationally?

4
User notification of system changes

• Require notification for changes affecting any of
  transmitted products
• Notification required well in advance
• Emergency situations (navigation errors HIRS
  failures)
• NESDIS has 24-hr systems operators that can
  provide immediate notification
• Quality indicators would be valuable
• System is working well for POES

5
Common and Consistent Data Formats

• Consistency between different NOAA satellites
• Shared processing, BUFR, 1B
• Consistency between NESDIS and EUMETSAT for
  NPOESS and METOP
• BUFR or other commonly accepted format
• BUFR preferred for operational users in Europe
• Time sink for both user and NESDIS
• Outgoing products need to be validated

6
Communications(timeliness)

• Data is required within 20-60 minutes of analysis
  time for mesoscale within 245 hours for global
• Consistency between direct readout and NESDIS
  provided data

7
Communications(availability)

• Availability of data (radiances) prior to product
  (retrieval) being declared operational
• Vital for rapid transition to operations
• NWP users are able to provide useful feedback to
  data producers
• Good for NOAA-15/16
• Not as good for DMSP instruments

8
Software and Scientific Knowledge / Support

• Examples are OPTRAN, surface emissivity models,
  algorithms for computing CLW and scattering
  indices for AMSU-A/B
• Software for encoding/decoding and displaying
  observations/products
• Knowledge of errors. Sources, biases,
  observation errors, RT and Jacobian errors,
  correlated errors, representativeness errors

9
Web Pages

• Information must be current and complete
• Data processing techniques, retrieval methods,
  validation, sensor characteristics, sensor status
  
• Historical record of changes for reanalysis
• List of reported problems status of problem
  resolution
• Stability of web page addresses

10
Quality Assurance and Validation

• Data providers do quality assurance of all data,
  including level 1b and level 1d. The quality of
  the data (including, e.g. navigation) should be
  monitored at all stages including the final
  stage, where data may have been reformatted. The
  provider should attempt to identify and flag
  questionable or poor quality data (ITWG 2001).

11
Quality Assurance and Validation

• Products, radiances and formatted observations
• Ability to read and display final transmitted
  product
• Consistency between products from different
  satellites
• SSM/I TPW and AMSU-B TPW
• NOAA-15/16 retrievals for upper atmosphere
• NESDIS should be responsive to user feedback on
  quality as it affects NWP

12
Quality Indicators

• Quality indicators for instrument health
• NESDIS has 24-hr operators
• HIRS failure during holiday period
• Navigation errors
• For retrieved products, need information on
• Retrieval algorithm
• Observation pre-processing
• Bias removal, limb correction, emissivity
  correction
• Quality and influence of the a priori

13
Looking to Europe as a Role Model

• EUMETSAT funds 2 visiting scientist positions at
  ECMWF and 2 positions at other NWP centers on
  rotating basis.
• Satellite Application Facility (SAF) for NWP has
  an additional 7 visiting scientist positions
• Visiting scientist positions in the U.S. are less
  common. What limits this?
• Location
• Financial
• Prestige

14
Joint Center for Satellite Data Assimilation
(JCSDA)

• Can the JCSDA serve as the NWP SAF for the U.S.?
• The NWP SAF is a EUMETSAT-funded project that
  exists to coordinate research and development
  efforts among the SAF partners to improve the
  interface between satellite data and NWP for the
  benefit of EUMETSAT member states.
• Use NWP SAF as a role model
• JCSDA will require a secure, long-term funding
  commitment.
• Should include opportunities for international
  collaboration

15
Funding, Computer and Staffing Issues

• Millions of dollars are spent on the development
  of new satellites.
• By comparison, little is spent on how to use
  those observations in NWP models.
• Data processing may not be adequately funded
  either
• Costs for assimilation AIRS
• 1 person/ 3 years at EACH NWP center
• Assumes that highly experienced staff and
  necessary tools (e.g., RT models) are available

16
Funding, Computer and Staffing Issues

• Need research computers to run assimilation
  tests, sufficient data storage and CPUs for data
  processing
• U.S. and Canadian Universities have few programs
  in data assimilation limited remote sensing
  training in Canada
• University of Reading data assimilation research
  center
• New hires
• Usually requires large amounts of in-house
  training

17
Computational Resources

• Availability of computers, computer time
• Need access to research computers
• DoD HPC resources are structured for large,
  single jobs, not data assimilation
• Operational computers do not have sufficient free
  resources for RD
• Limited data storage CPUs for data processing
• Limitations of GTS for large amounts of data

18
Formal education and training

• U.S. and Canadian Universities have few programs
  in data assimilation limited remote sensing
  training in Canada
• University of Reading data assimilation research
  center
• New hires
• Usually requires large amounts of in-house
  training
• Little expertise being produced by universities
• Difficulty hiring experts in radiative transfer
• Interest in operational problems, capable of
  running the data assimilation system and
  evaluating the results, able to write good
  efficient code on parallel systems.
• Someone who understands the science issues,
  preferable data assimilation also, has a good
  approach to coding.
• Support staff for programming, running
  assimilation tests

19
Scientific Challenges

• Mesoscale data assimilation
• Moisture, clouds, winds are important
• Temperature information will be limited by broad
  vertical weighting functions, presence of clouds
  (IR).
• Resolution and timeliness of data different
  forecast goals (fronts/squalls/ducts)
• Assimilation of satellite information
• Complex terrain
• Coast zones
• Areas with sparse conventional observations
• Land, snow and ice-covered surfaces, polar
• Cloudy regions (IR)

20
Scientific Challenges

• Fast RT model improvements
• Improved surface emissivity models
• Cloud detection/cloud-clearing/assimilation of
  cloud-contaminated radiances
• Sensitive areas tend to be heavily clouded at low
  levels (where information is most needed).
• Improved moisture assimilation, assimilation of
  other parameters (e.g., soil moisture)
• Upper atmosphere, aerosols

21
NWP Requirements and Issues

• Science issues for radiances and retrievals are
  similar.
• Consistency between software processing for
  different satellites (e.g., AIRS and ATOVS)
• Availability of data from research satellites
  (AIRS/TRMM/GIFTS)
• May not have stability of algorithms,
  communications, orbits, measurements, and
  calibration software. Poses problem for
  operational user.
• User often accepts these issues in order to have
  early access to data
• Users look to NESDIS to provide access to
  non-NOAA satellites

22
NWP Requirements and Issues

• Fast and accurate RT direct calculations and
  Jacobians
• Common interface
• Consistency between RT for different instruments
• Increasing requirements for information on ozone,
  aerosols, upper atmosphere, vegetation indices.
• Difficulty working with DAS and forecast models
• NWP requirements give equal (or more) weight to
  winds.

23
Are the observation under-utilized?

• Dramatic increase in satellite use in past 10
  years.
• Science issues not dealt with prior to launch of
  satellite
• Availability of RT models/ RTM needs improvement
• Early sensor failure (DMSP)
• Observations over land/ice/snow are more
  difficult to use properly
• Surface contributions are not well known (skin
  temperature and surface emissivity)
• Precipitation and cloud detection
• Cloud detection/cloud-clearing

24
Are the observation under-utilized?

• Delays due to non-release of data inadequate QC
  changes in data formats
• Limited independent information in observations
• Cost of adding data vs. expected payoff
• Desire for a good mix of observations
• Model resolution
• Limited resources (funding and scientists)
• International cooperation/collaboration

25
NWP Forecast Impact

• ECMWF Combined TOVS, SATOB, SSM/I
• NH 12 hours at 7 days
• SH 30 hours at 6 days (25 forecast skill)
• NH impact from radiosondes and satellites are
  equivalent
• SH satellites clearly dominate
• TOVS CPU cost 5
• UKMO skill improvement over past year from
  assimilating ATOVS radiances greater than total
  improvement 1992-1999.
• NCEP Use of radiances nearly double forecast
  impact in both hemispheres. ATOVS adds
  additional information in cloudy areas from
  AMSU-A and useful moisture information from
  AMSU-B.

26
NESDIS Role in Data Assimilation

• NESIDS should become involved in assimilation
  issues in a collaborative sense.
• NESDIS should provide data and data expertise
  (but not be directly involved in data
  assimilation).

27
Satellite Products used by U.S. Navy NWP Models

• ATOVS radiances (research mode)
• ATOVS temperature retrievals
• SSM/I wind speeds and total precipitable water
• Ambiguous scatterometer winds
• Feature-tracked winds (vis, IR, water vapor)
• Sea-surface temperatures (MCSST from Naval
  Oceanographic Office)
• Altimeter data (experimental)
• Sea-ice concentration (National Ice Center)
• Snow fields (Air Force)
• RTNEPH (Air Force research mode)
• TC warning messages to produce TC bogus
• SH sea-level pressure bogus (PAOBS)
• Aerosol retrievals
• Other products used for verification