Analysis Systems

1
Analysis Systems
What is the purpose of analysis systems? -
To combine available observations, climatological
data, and other analyses into one consistent
picture, and to derive non-observed fields such
as sound velocity from observed fields such as
temperature and salinity. Analysis systems do not
make predictions or forecasts.
2
Analysis Systems

• What are the Navys Ocean Analysis Systems?
• MODAS Modular Ocean Data Assimilation System
• OTIS Optimum Thermal Interpolation System
• MVOI Multivariate Optimum Interpolation

3
MODAS Modular Ocean Data Assimilation System

• Primary contacts Dan Fox (NRLSSC), Martin Booda
  (NAVO)
• MODAS was developed at the Naval Research Lab at
  Stennis Space Center in the 1990s (Fox et al.,
  2002). It is currently used in a stand-alone
  mode and also to initialize a relocatable version
  of the Princeton Ocean Model. A subset of the
  full capabilities is available for use on
  submarines.

4
MODAShttp//www7320.nrlssc.navy.mil/modas/
Analysis system that uses optimal interpolation
to incorporate MC-SSTs, SSH from satellite
altimetry, T and S data from profilers (XBTs and
CTDs) and fixed or drifting buoys, with
climatological data to produce 3D T and S fields.
3D sound velocity fields, and associated acoustic
parameters, and geostrophic velocity fields, are
derived from the temperature and salinity fields.

5
Physics

• Geostrophic.
• In water depths greater than the reference level,
  geostrophic velocities are referenced to the
  reference level. In water depths less than the
  reference level, geostrophic velocities are
  referenced to the bottom, i.e. it is assumed that
  there is no horizontal pressure gradient at the
  bottom, which can produce a velocity field that
  is locally divergent. Strictly speaking, the
  geostrophic velocity field should be
  non-divergent.
• The reference level is user-selectable. The
  default is 1000 m.

6
Domain

• Determined by user
• The EOF scheme used to compress the 3D
  temperature and salinity output has a limit of
  360 grid points in the east-west direction and
  181 grid points in the north-south direction. The
  spatial resolution, in combination with the
  maximum number of grid points, sets the maximum
  allowable domain, if the EOF-compressed output is
  needed.

7
As of 28 Feb. 2002
8
Spatial Resolution

• Determined by user as a result of specifying the
  domain size, and the number of grid points or the
  spatial resolution
• xmin longitude at the left edge of the grid
  (E, -W)
• xmax longitude at the right edge of the grid
• ymin latitude at the bottom edge of the grid
  (N, -S)
• ymax latitude at the top edge of the grid
• nx number of grid points in the east-west
• direction (max 360)
• ny number of grid points in the north-south
• direction (max 181)

9
E-W grid spacing dx (xmax - xmin) / (nx -
1) N-S grid spacing dy (ymax - ymin) / (ny -
1) grid pts E-W dir. nx 1
(xmax-xmin)/dx grid pts N-S dir. ny 1
(ymax-ymin)/dy If specifying, dx and dy,
program will calculate nx and ny, and GUI will
alert user if nx, ny arent integers.
From the MODAS 2.1 User's Manual, applies to the
GUI-based version.
Try out the Interactive MODAS Grid Calculator
10
Temporal resolution

• None
• This is not a time-stepping predictive model.
• The analysis can be updated at whatever interval
  the user chooses, but should be based on the
  availability of new data to assimilate. NAVOs
  update cycle is generally once per day.

11
Initialization

• The user has the option of using a previous MODAS
  run as the first guess or using climatology.
  MODAS2.1 has its own climatology. The MODAS
  climatology is based on a blend of MOODS T and S
  data in the upper 1500 m of the ocean with the
  Levitus climatology in the deeper ocean. The
  MODAS climatology is stored as bimonthly T and S
  at 37 depth levels from 0 to 6500 m, on a grid
  with horizontal resolution ranging from 1/2? in
  the open ocean to 1/4? in coastal seas and 1/8?
  near the coasts. The MODAS climatology extends
  into water depths as shallow as 5 m.

12
Data Assimilation

• Assimilates SSH, SST, XBTs, fixed buoys, and
  PALACE floats at this time (May 2002).
• There is a time window over which data is
  assimilated in other words how far back in time
  should data be included and with what weighting.
  This time window is not easily changed, or
  viewable, by the user, but can be accessed and
  modified.
• 1/8 resolution global field of SSH is supplied
  by the altimetry data fusion center (ADFC).
  Does not improve solution in all regions (due to
  seiching, among other things), so is not used
  everywhere.

13
Altimetry Processing Upgrade CLAM II
Implements moving covariance function in MODAS
OI (more continuity of features between tracks)
Original Results
Updated Results
Courtesy of John Harding, NRL-SSC
Note particularly significant improvement in
these areas
Note sampling impact Red would be green but for
this area
14
Clam Shallow Water Evaluation
Most Accurate Modas First-Guess Field (Bottom
Depth lt 200 m)
MODAS Seasonal Climatology MODAS Climatology -
MCSSTs
Latitude
Longitude
15
NAVO MODAS areas not using SSH(as of 7/3/02)

• Kamchatka
• Mediterranean regional
• Bay of Biscay/NE Atlantic
• Northeast Pacific
• Gulf of Oman/Arabian Gulf
• SOCAL
• Straits of Sicily
• Western Med

• Adriatic
• Arabian Sea
• Gulf of Alaska/Eastern Aleutians
• Gulf of Cadiz
• Central Med
• Central North Atlantic
• Eastern Med
• Greenland/Iceland/ Norwegian Sea

16
Data Assimilation (cont.)

• 8.8 km resolution MC-SST, derived from 2x2 blocks
  of 4.4 km Global Area Coverage (GAC) data
• Relationships between SSH and subsurface T, and
  SST and subsurface T, based on historical data,
  are used together with climatology to produce
  synthetic vertical profiles of T, down to a depth
  of 1500 m.
• After SSH and SST have been used to create the
  dynamic climatology, in-situ data from XBTs,
  CTDs, buoys and floats is assimilated through an
  optimal interpolation (OI) scheme. The 3D grid
  of T is then modified near the surface by using
  an analysis of mixed layer depths. T/S
  relationships are then used to produce a 3D S
  field as well. If S observations are available,
  they are then used to modify the first guess
  salinity field.

17
Satellite Measured SSH and SST
Green line represents profile derived only using
satellite measured height and temperature
Decades of edited MOODS profiles are used to
derive statistical relationships between surface
height and temperature and subsurface temperature
and salinity
Relationships are stored on an irregular mesh,
varying from 1 to 1/8 degree in resolution to
permit high resolution analyses in shallow water
regions
Climatology MODAS Synthetic Final Analysis In
Situ BT
Courtesy of Dan Fox, NRL-SSC
18
MODAS Validation Example AXBT Survey
MODAS results
Cold core eddy
MODAS Temperature at 200m
Courtesy of Dan Fox, NRL-SSC
19
T/S plot from central California1997-2002
Equatorial Intermediate Water
Temperature (C)
El Nino affect
North Pacific Intermediate Water
35
31
33
31
34
Salinity
20
Implementation

• Set of Fortran programs and UNIX scripts run
  under the UNIX operating system. Different
  versions of MODAS have different subsets of the
  full suite of modules installed.
• MODAS modules have been "wired together" to
  produce several analysis and forecast systems
  presently running at various Navy facilities.
  Versions designed for shipboard and
  submarine-based users are under development now.
• PC-IMAT now includes MODAS-Lite, basically the
  same version as is run at the METOC centers, but
  with different GUI and Perl scripts.

21
MODAS2.1 Heavy Run at NAVOCEANO on SGI
Origin, Power Challenge Array, and ONYX
systems Satellite altimetry and MCSSTs plus in
situ data Includes reloctable Princeton Ocean
Model Approx 6 GB of disk space required
MODAS2.1 NITES-I ASHORE METOC Center
version All capabilities of above except would
normally receive first guess field from
NAVO via TEDS METCAST. Variable disk space
(max 6 GB) depending on desired area of
coverage UNIX systems (HP TAC4, Sun, SGI, PC
LINUX, )
MODAS2.1 NITES-I AFLOAT (MODAS/Lite) No
direct altimetry use. Requires first guess field
from MODAS run at Center. Input/output via TEDS
database Minimal disk space required ( less than
1 GB for global coverage ) UNIX systems (as
above)
22
MODAS 2.1 OUTPUT PRODUCTS
Text/Binary/Message Files

Byte-Encoded EOF-Compacted
Temperature/SV (also pushed to
centers, facilities ships at sea)

JJXX/JJYY/KKXX Synthetic
BTs

OVLY2 of Physical/Acoustic
parameters

NetCDF of Temperature/SV/Salinity

ARCVIEW Format (for REACTs)
Physical/Acoustic Graphics

Temperature Contours
at Depth

Currents over Temperature
at Depth

DSCA, SSCA, MLD, SLD, ZX

Observations Chart (Secret)
Wavelet-compressed fields became available in May
2001
23
Note that velocity scale is different for every
picture
MODAS
24
The arrows are parallel to the streamlines
everywhere and their length indicates the speed,
as referenced to the velocity scale vector. The
length of the curved arrow is a function of the
velocity all along it's short path - not just at
the beginning.
25
Typical Satellite SSTs of Kuroshio Current
(2/19/1996)
Estimated surface current conditions in Kuroshio
area during time period of the MODAS data
analysis. Data are from weekly survey reports
produced by Japanese researchers. These reports
are independent of MODAS calculations, though the
same data may be used.
Adapted from Johnson and Broome 1999
26
References

• Fox, D. N., W. J. Teague, C. N. Barron, M. R.
  Carnes, and C. M. Lee, 2002 The Modular Ocean
  Data Assimilation System (MODAS). Journal of
  Atmospheric and Oceanic Technology, 19, 240-252.
• Fox, D.N., C.N. Barron, M.R. Carnes, M. Booda, G.
  Peggion, and J. Gurley, The Modular Ocean Data
  Assimilation System, Oceanography, 15 (1), 22-28,
  2002a.
• Johnson, A. and R. Broome, 1999 Validation Test
  Report for the Modular Ocean Data Assimilation
  System (MODAS 2.1), 42 pp.
• Naval Research Laboratory, P. S. I., 1999 User's
  Manual for the Modular Ocean Data Assimilation
  System (MODAS) Version 2.1. PSI Technical Report
  S-285.

27
OTIS Optimum Thermal Interpolation System

• Primary contact Webb DeWitt (FNMOC)
• OTIS predates MODAS. It was developed at FNMOC
  in the 1980s. While it is still being run by
  FNMOC, it is scheduled to be phased out.

28
OTIShttp//www.fnmoc.navy.mil/

• OTIS is an optimum interpolation (OI) based
  objective analysis scheme designed to produce
  analysis or "nowcasts" of temperatures in the
  upper 5000 m of the ocean.
• OTIS does not provide currents as output, but
  feeds into TOPS (see below), which does.

29
Domain

• OTIS has been implemented at FNMOC on a variety
  of regional (eddy resolving) and global (non-eddy
  resolving) grids.

30
(No Transcript)
31
Temporal and Spatial Resolution

• This is not a time-stepping predictive model.
• Presently, SST-only runs are done for the global
  domain twice a day at 1? by 1? horizontal
  resolution, and once a day at 0.25? by 0.25?.
  There is also a global 3D OTIS with 1? resolution
  run twice a day. Regional 3D runs are done once
  per day for the Atlantic and Pacific oceans using
  0.2? by 0.2? resolution.
• 34 levels are used in the vertical, with 5 m
  spacing near the surface, expanding to 100 m at
  400 m depth and 200 m at 2000 m depth.

32
Initialization

• The 3D implementation uses GDEM climatology as a
  starting point and then uses a real-time "ocean
  bogus" database. This is a significant
  difference from the way MODAS works. The "ocean
  bogus" is the fronts and eddies, or Oceanographic
  Features Analysis product, from NAVO. It gives
  positions of ocean features determined from
  satellite data, and each grid point in OTIS is
  assigned a water mass classification based on its
  position relative to these features. "The water
  mass classification determines the appropriate
  water mass climatology model to apply at the grid
  point" (Cummings et al. 1997). The water mass
  climatology model uses date, location, and remote
  and in-situ measurements, to determine the
  temperature and salinity versus depth for that
  grid point.
• The OTIS SST-only runs use the previous analysis
  as the first guess field.

33
Oceanographic Features Analysis(Fronts and
Eddies Bogus)
34
Data Assimilation

• OTIS makes full use of the global real-time
  observations received at FNMOC. This includes,
  ships, fixed and drifting buoys, and
  satellite-derived sea surface temperatures, and
  bathythermograph and buoy subsurface
  temperatures. All observations are quality-
  controlled prior to being assimilated by OTIS.
• The 2D and 3D versions may use different time
  windows over which they accept observations.

35
Output

• Global OTIS SST may be viewed on the FNMOC web
  site (www.fnmoc.navy.mil).
• Other OTIS fields are available as JMV
  thumbnails, or could be requested via Metcast,
  and viewed with JMV.

36
(No Transcript)
37
References

• Cummings, J. A., C. Szczechowski, and M. Carnes,
  1997 Global and regional ocean thermal analysis
  systems. Marine Technology Society Journal, 31,
  63-75.
• Documents on FNMOC web site.

38
3D-MVOI 3-Dimensional Multivariate Optimum
Interpolation

• Primary contact Jim Cummings (FNMOC)
• 3D-MVOI was developed at NRL Monterey in the late
  1990s.

39
3D-MVOI

• Currently being used in a 2D mode to provide SST
  to the operational COAMPSTM, the Coupled
  Ocean-Atmosphere Model Prediction System, and its
  regional versions DAMPS (Distributed Atmospheric
  Model Prediction System) or TAMS/RT.
• Ultimately, 3D-MVOI will serve as the ocean data
  assimilation scheme feeding the predictive ocean
  models (NCOM regionally and POP globally) that
  will be coupled to the atmospheric models
  (COAMPSTM regionally and NOGAPS globally).

40
Domain

• Must match domain of atmospheric or ocean model
  that its being used with.

41
Ocean Forecast Component Sequential
Incremental Update Cycle Analysis-Forecast-Analysi
s
Ocean Obs
Ocean QC
MCSST GOES SST Ship SST Buoy SST XBT, CTD PALACE
Float Fixed Buoy Drift Buoy Altim SSHA SSM/I Sea
Ice
Innovations
3D MVOI
Increments
Ocean Model
First Guess
Forecast Fields Prediction Errors
MVOI - simultaneous analysis 5 ocean variables
temperature, salinity, geopotential, velocity
(u,v)
Courtesy of Jim Cummings, NRL-Monterey
42
Initialization

• A nowcast of the observed state is obtained by
  combining new observations with a background
  field. The background field can be from a
  short-term model forecast or a previous analysis
  (warm start), or climatology (cold start). The
  use of a previous analysis as the background
  field implies a persistence forecast from the
  last time the analysis was executed.

43
Operational Observation Data SourcesNRL Coupled
Systems

• Satellite SST
  330,000 obs/day
• satellite SST retrievals (NOAA 16)
• In Situ SST/SSS
  15,000 obs/day
• surface ship, fixed and drifting buoys, CMAN,
  TRACKOB
• Subsurface Temperature and Salinity Profiles
  500 obs/day
• XBTs, CTDs (TESACS), PALACE floats
• fixed buoys (TAO, PIRATA), thermistor chain
  drifting buoys
• Sea Surface Height Anomaly (SSHA)
  100,000 obs/day
• altimeter (TOPEX, ERS2, GFO), in situ
  observations (PALACE floats)
• Sea Ice Concentration
  1,2000,000 obs/day
• SSM/I (DMSP F13, F14, F15)
• Synthetic Salinity-Temperature-Depth Profiles
  (STDs)
• temperature profiles computed from SST and SSHA,
  salinity computed from temperature (using MODAS
  databases)
• synthetic profiles are generated in a sampling
  pattern to capture analyzed changes in SSHA that
  exceed 2 cm

Note that the type and amount of data is subject
to change depending on what is available. (Slide
is courtesy of Jim Cummings, NRL-Monterey)
44
Data Assimilation

• In addition to numerous satellite and in-situ
  observations, MVOI uses synthetic temperature and
  salinity profiles calculated using the same
  databases and algorithms as MODAS, to project
  satellite sea surface height and temperature data
  down to the subsurface ocean.
• A sophisticated interpolation scheme, which
  varies the weighting of the observations as a
  function of variable type, time, depth and
  horizontal distance is used.

45
Horizontal Correlation Length Scales NRL
Coupled Systems
Rossby radius of deformation (from Chelton et al.
(1998), JPO 28 433-460). Used as default for
horizontal correlation length scales in the
3D-MVOI. Scales range from 10 km at the poles
to 240 km in the tropics.
Courtesy of Jim Cummings, NRL-Monterey
46
Data Assimilation (cont.)

• Another method that is used to incorporate the
  SSH into the analysis is to adjust the model T
  and S fields to improve the agreement between the
  observed and modeled SSH field. This has the
  advantage of not always trying to restore the
  modeled field towards climatology.

47
References

• Cummings, J., 2002. Powerpoint brief.