Towards Fully Coupled OceanAtmosphere Mesoscale Prediction

1
Towards Fully Coupled Ocean-Atmosphere Mesoscale
Prediction
James D. Doyle, John Cook, Richard Hodur
Developers and Contributors C. Amerault1, S.
Chen, J. Cummings, S. Gabersek2, T. Haack, T.
Holt, X. Hong, Q. Jiang2,Y. Jin, H. Jin3, M. Liu,
P. May4, J. Nachamkin, J. Pullen, J. Schmidt, W.
Thompson, S. Wang Naval Research Laboratory,
Monterey, CA 1NRC, 2UCAR, 3SAIC, 4CSC,

• Outline
• Background
• COAMPS System Current Status
• Physical Process Representation
• Air-Sea Interaction Along California Coast
• Future Plans

2
Background Coupled Ocean/Atmosphere Mesoscale
Prediction System
COAMPS is a registered trademark of the Naval
Research Laboratory.
3
COAMPS Coupled Ocean/Atmosphere Mesoscale
Prediction System
COAMPS is a registered trademark of the Naval
Research Laboratory.
4
COAMPS-OS Coupled Model Initiative COAMPS-On
Scene Background
COAMPS-OS is a turnkey, globally relocatable
environmental data assimilation production server

• Highly automated quick response time globally
  relocatable
• Web-based configuration forward-deployable or
  reach-back modes of operation
• Ocean and atmosphere analyses, including
  dedicated cloud and Doppler radar wind analyses
• COAMPS mesoscale forecasts with full physics
• Automatically transforms output into dynamic web
  graphics and web-accessible data
• Custom interfaces to provide data for HPAC (JEM),
  AREPS and other applications
• Web-based Remote Monitor capability
• DOD, Federal agencies, NATO, university and lab
  customers
• Operational since 1998 with continuous product
  improvements through NRL RD programs
• Sponsored by ONR, N84 (PMW 180 CNMOC) and DTRA

Environmental battlespace characterization
reach-back strategy directly addresses CNMOC
Battlespace on Demand (BOND) requirements and
supports AMOP roadmap
COAMPS and COAMPS-OS are registered trademarks
of the Naval Research Laboratory.
5
COAMPS Modeling Infrastructure Coupled
Ocean-Atmosphere Prediction
ESMF Modeling Superstructure
Atmospheric Model
Ocean Model
Wave Model
Dynamics
WRF Physics Interface to Physics Suites
Flux Coupler
Share Community Models/Share Community Physics
Relatively easy to add new modeling components
6
High-Resolution Tropical Cyclone Modeling Moving
Nested Grid Capability
Coarse Mesh 27-km Fine Mesh 9-km
TRACK ERROR
Improved TC Track Forecast

• Highest resolution can be focused where it is
  needed
• Improves efficiency
• Can be used to track a tropical storm, Navy
  battle fleet, plume

7
COAMPS Parameterization Research Improved Surface
Flux Parameterization
Reformulate Louis scheme using TOGA COARE, CBLAST
field data
Forecast Hour
Moana Wave N m-2
Moana Wave N m-2
8
COAMPS Parameterization Research Improved Surface
Flux Parameterization
New surface flux scheme improves COAMPS surface
wind speed and moisture prediction
New moisture roughness length is smaller than the
heat roughness because the heat transfer is more
efficient than the moisture.
S. Wang
9
COAMPS Atmospheric Reanalyses High-Resolution
Forcing for Ocean Models

• Cold start at first analysis time
• 12 hour incremental data assimilation cycle
• Hourly output from forecast model
• SST analysis every 12 hours for each grid
• Four areas
• Mediterranean (81/27 km)
• Eastern Pacific (81/27/9 km) (1998-2001)
• Eastern Pacific (81/27/9/3 km) (2002-03)
• Adriatic (36/12/4 km)
• Baltic Sea (81/27/9 km)

10
One-Way vs.Two-Way Coupling Air-Sea Interaction
in the Adriatic Sea 30 day means 23 Sept 23
Oct 2002
Lowest SST and near-surface wind RMS and Bias
Errors found using 2-way Coupling
11
Summer Mean Atmospheric Boundary
Layer Large-Scale Forcing for U.S. West Coast
Warm season climatology (April-September)
850-mb Geopotential Heights
850-mb winds
H
L
Mass and Bond (1996)
12
COAMPS Reanalysis and Real Time Forecasts 4-Year
Summer Mean (June-Sept 2002-2005)
13
COAMPS Real-Time Forecasts for Central
Coast Products for Atmospheric and Oceanic
Forecasting

• Real-time forecasts (48h) for July 2003-present
  run by NRL at FNMOC
• Nested grids (81, 27, 9, 3 km)
• Surface and BL fields saved every 1 h for ocean
  model coupling
• Forcing for Real-Time Ocean Forecasts (JPL ROMS,
  NRL-SSC NCOM)
• Real Time to NWS-Monterey (AWIPS ingest)
  Forecaster Feedback
• Support for Numerous Field Studies (AOSNII, ASAP,
  AESOP, LOCO)
• Forecasts on the Web
• http//www.nrlmry.navy.mil/coamps-web/web/mbay
• http//cimt.ucsc.edu/
• http//cimt.jpl.nasa.gov/

Overall, it was found that these forcing
(COAMPS) were of quality superior to anything we
had utilized before. Haley et al. (2007)
HOPS AOSN Team
Future Status Ported to New LNXI Cluster at FNMOC
Planned Upgrade (BL, clouds, Sep. 2007) Support
for FY08 from CeNCOOS
14
COAMPS Real-Time Forecasts for Central
Coast Products for Atmospheric and Oceanic
Forecasting
Graphs on right show observed (upper) and COAMPS
(lower) Surface Stress
15
COAMPS Real Time Forecasts AOSN II Aircraft
Observations and COAMPS (Aug. 2003)
Courtesy of J. Paduan
16
What Does High Resolution Buy? COAMPS (Dx333 m)
24 June 1996)
Monterey Bay
Point Sur
High-Resolution (lt 1 km) is needed to
resolve fine-scale boundary layer interactions
with topography
Burk and Haack (2000)
17
COAMPS Sea Surface Temperature Analysis NRL
Coupled Ocean Data Assimilation (NCODA)

• Sea surface temperature analyzed directly on
  COAMPS grid
• SST has profound impact on boundary layer winds
  (Important for
  atmosphere and ocean)

COAMPS is the state-of-the-science model for
coastal wind prediction forcing for ocean models
D. Chelton, T. Haack, J. Doyle, J. Pullen
18
COAMPS Reanalysis and Real Time Forecasts 4-Year
Summer Mean (June-Sept 2002, 2004, 2005)
QuikSCAT COAMPS Wind stress Statistics
? Bias
? RMSE
Courtesy of D. Chelton
19
COAMPS Reanalysis and Real Time
Forecasts Summertime Ekman Upwelling Velocities
(2002,2004,2005)
QSCAT
COAMPS
Courtesy of D. Chelton
20
COAMPS Reanalysis and Real Time Forecasts 4-Year
Summer Mean (June-Sept 2002-2005)
Coupling Coefficients
Quikscat slope2.96
Quikscat slope2.28
Anomaly
Anomaly
Offshore curl and divergence are more closely
linked with SST features in agreement with
satellite observations.
21
COAMPS-OS Coupled Model Initiative Joint Project
with NRL-SSC (Code 7300)
Local Data
User configurable 1, 6, 12, or 24 hr atm update
cycle
NRLSAT
SWAN
Cloud Analysis
NAVDAS
COAMPS
Radar Wind Analysis
SST Analysis
1-way coupling every hr
gNCOM HYCOM
coupler
BC/IC
2-way coupling freq TBD
NCODA
QC FNMOC
NCOM
DA-IU t, s, u, v
Wave Watch 3
Data Altimeter SST Ship Profile
T, S, U, V
Global Wave Watch 3
Climate Bathymetry GDEM, MODAS, DBDBV
OSU Tides
Currently MVOI, will be replaced by NAVDAS in
FY07
NOGAPS
COAMPS and COAMPS-OS are registered trademarks
of the Naval Research Laboratory.
22
COAMPS-OS One-Way Coupling Cascade Coupled
Ocean-Atmosphere Prediction
Run within COAMPS-OS
SWAN 0.01 deg
SWAN 0.008 deg
15 km
Oahu
Oahu
COAMPS 5 km
COAMPS 45 km
Delft3D Kaena Pt. Wave Height and Direction
External Run Delft3D prediction of modulated wave
field important for NSW exercise (Tech Eval / Op
Eval Feb 2007)
Significant Wave Height (ft) and Direction
40 m grid spacing
Wave Height (ft)
COAMPS and COAMPS-OS are registered trademarks
of the Naval Research Laboratory
23
Future Plans and Challenges

• Fully Coupled Two-Way Interactive System
  (air-ocean-waves-spray) COAMPS
• Coupled prototype ready for testing
• Long-term development path
• Test system along California Coast (leverage
  existing datasets, expertise)
• Continued Improvements to COAMPS
• Data assimilation, physical parameterizations
  (clouds, PBL, aerosols)
• Physics interoperability with other models (e.g.,
  WRF)
• Development of new dynamical cores (DG, Spectral
  Element, WENO)
• New Capabilities for COAMPS
• Ensemble system
• Adjoint system
• Very high resolution configuration (LES)
• ESMF compatibility with COAMPS-OS (Turnkey
  Automated System)
• Challenges / Opportunities
• Fully Coupled Unified Systems (Across Disciplines
  and Scales)
• Integrative Methods for Physical Processes
  Representation in Coupled Systems
• Predictability of the Mesoscale Coupled
  Environment
• Coupled Data Assimilation (Ensembles, 4D-Var,
  Hybrid Approaches)
• Community Modeling and Collaboration Across
  Disciplines

24
COAMPS Cloud Prediction Real-Time Forecasts June
2006 00 UTC
ICLW (g / m2)
Cloud mask ()
500
100
COAMPS 9-km
250
50
0
0
500
100
GOES Satellite
250
50
0
0
25
COAMPS Reanalysis Monthly Mean Winds (1999)
10
Apr
Jan
Jul
Oct
8
6
4
2
Buoy 26 (San Francisco) Time Series
140
(a)
140
(c)
Dir (deg)
Dir (deg)
300
300
100
Direction
100
Direction
RMS 1.05 Bias 3.04
20
1.35 2.37
20
V (m s-1)
V (m s-1)
10
10
0
0
Speed
Speed
0
100
200
300
400
500
600
0
100
200
300
400
500
600
700
700
January Time (hrs)
July Time (hrs)
140
140
(b)
(d)
Dir (deg)
Dir (deg)
300
300
Direction
Direction
100
100
0.26 2.09
20
20
V (m s-1)
V (m s-1)
10
10
1.45 2.68
0
0
Speed
Speed
0
100
200
300
400
500
600
700
0
100
200
300
400
500
600
700
April Time (hrs)
October Time (hrs)
26
Fr from COAMPS
Coastal Hydraulic Flows 12 June 1996 (CW96)
292 K Potential Temperature Surface
Cape Blanco
Expansion fan
Compression jump
Cape Mendocino
C-130 flight data (Rogers et al., 98)
Expansion fan
Marine boundary layer depth features
Fr V/(gh)1/2 g g/? ??
27
COAMPS High-Resolution Prediction Coupled
Dispersion Modeling Using COAMPS-HPAC
Total HPAC Dosage, 8 Nov. 1996
27 km input plume missed the network
Increased Horizontal Resolution Improves Dosage
Forecasts
1 km input correct shape, dosages overestimated
Measure of Effectiveness Diagram
Under-prediction

• All dosage forecasts 30 PPT-hr
• The 1 km forecasts have greater overlap with a
  positive coverage bias

Better
HITS/ OBS
Obs input approx. ground truth
Over-prediction
J. Nachamkin
HITS/ FCSTS
28
NRL Marine Meteorology Division Telescoping
Modeling Strategy