Date post: | 20-Jan-2016 |
Category: |
Documents |
Upload: | augustine-rogers |
View: | 218 times |
Download: | 0 times |
UMAC data call page 1 of 28Operational wave modeling suite
EMC Operational Models
NCEP WAVE MODELING SUITES
Arun Chawla and Jose-Henrique AlvesNCEP Wave Modelling Suite,Environmental Modeling CenterNOAA / NWS / [email protected], [email protected]
UMAC data call page 2 of 28Operational wave modeling suite
Content
• Description of current modeling system and data assimilation components,– Global wave model Multi_1,– Global wave model Multi_2,– Global Wave Ensemble System GWES,– Great Lakes Wave System GLW.– Nearshore Wave Prediction System (NWPS) in separate slide deck.
• Future plans for the next 5 years (including new systems)• Existing model verification statistics documentation and
associated web sites.• Sources, selection of requirements and development path.
UMAC data call page 3 of 28Operational wave modeling suite
MULTI – 1
• A global wave model consisting of a mosaic of two way nested grids ranging in resolution from ½ degree (global domain) to 1/12 degree (US coastal waters)
• Model is driven off of GFS winds and runs side by side with the GFS model
• Short term forecast : 4 cycles a day. Each cycle consists of 9 hours of hindcasts and 180 hours of forecast
• System uses the WAVEWATCH III model • No data assimilation in the system • website : http://polar.ncep.noaa.gov/waves• Skill scores developed offline (continuous validation page
under development)
UMAC data call page 4 of 28Operational wave modeling suite
Multi-1 domain
All resolutions in arc-minutesYellow covers full globe
UMAC data call page 5 of 28Operational wave modeling suite
Multi -1 skill scores across available NDBC buoys for
72 hour forecast
Monthly skill scores Seasonal skill scoresTime in MM/YY
UMAC data call page 6 of 28Operational wave modeling suite
Multi – 1 Future Plans
• Extend the domain to the North Pole using a curvilinear grid (current domains stop at 83 N due to CFL limitations)
• Add data assimilation module– Development of wave data assimilation along two tracks
• LETKF• GSI
• Couple multi-1 with GFS using NEMS • Enhanced wave-ice interaction processes (damping , scattering etc.)• Replace nearshore regular coastal domains with unstructured grids
and significantly increasing the resolution from the current 7.5 km to < 1 km
• Add wave - surge coupling (couple with ESTOFS model)
UMAC data call page 7 of 28Operational wave modeling suite
Multi – 1 Reference
A. Chawla, H. L. Tolman, V. Gerald, D. Spindler, T. Spindler, J-H G. M. Alves, D. Cao, J. L. Hanson, E-M Devaliere (2013) “ A Multigrid Wave Forecasting Model: A New Paradigm in Operational Wave Forecasting”, Wea. & Forecasting, 28, pp 1057 - 1078
UMAC data call page 8 of 28Operational wave modeling suite
Multi -2
• Global model with a mosaic of grids like multi-1. • Regional domains are driven with a blend of GFS and
HWRF winds• Short term Forecast : 4 cycles a day. Each cycle consists of
6 hours of hindcast and 126 hours of forecast• Development cycle keeps pace with hurricane wind
developments• No data assimilation in wave model system yet.
UMAC data call page 9 of 28Operational wave modeling suite
Multi-2 domain
All resolutions in arc-minutesYellow covers full globe
Additional regional ¼ degree grids for blending HWRF/GFS winds
UMAC data call page 10 of 28Operational wave modeling suite
Multi-2 validation
Buoy validation during hurricane Sandy (black line – latest upgrade)
UMAC data call page 11 of 28Operational wave modeling suite
Multi – 2 Future Plans
• Extend regional hurricane grids to Indian Ocean, Western Pacific and Southern Ocean (to account for new hurricane domains)
• Coupling of hurricane wind – wave – ocean models – Initially on an NCEP coupler and later transitioning to NEMS
• Improved physics packages under high wind conditions• Development of a moving grid formulation to allow for wave
model to move on similar resolution grids as the hurricane model
UMAC data call page 12 of 28Operational wave modeling suite
Global Wave Ensemble Systems (GWES/NFCENS)
• A global wave model consisting of a single grid with ½ degree spatial resolution.
• Model is driven off of GEFS winds, with 21 members (one control forced with GFS data).
• Models target medium-term forecasts, with 4 daily cycles, each consisting of a 24 h hindcast and a 240 h forecast.
• System uses the WAVEWATCH III model.• No data assimilation in the system.• A combined product blends GWES data with US Navy
FNMOC global wave ensemble data, onto an the NFCENS operational wave-height product.
UMAC data call page 13 of 28Operational wave modeling suite
Global Wave Ensemble System (GWES/NFCENS)
• Domain
UMAC data call page 14 of 28Operational wave modeling suite
Global Wave Ensemble System (GWES/NFCENS)
• Future plans for the next 5 years– Switch hindcast to GFS hybrid EnKF forcing,– Expand suite of probabilistic wave products attending customer
request (NHC, OPC etc),– Two-way coupling with GEFS components,– Extend to high-resolution coastal ensembles using neural-networks,– Expand members to include multi-physics components,– Expand suite of combined products via data exchange with FNMOC
and the Canadian Meteorological Center (North American Wave Ensemble System, NAWES),
– Inclusion of LETKF-based data assimilation, coupling wave and atmospheric models DA steps.
UMAC data call page 15 of 28Operational wave modeling suite
• Hs Bias
• Typical conditions (bulk, global)
• 2011-2012altimeterdata (J1, J2, ENVSat)
• Combination:• Reduces bias
GWES & NAWESVerification Statistics
UMAC data call page 16 of 28Operational wave modeling suite
• Hs RMSE and Spread
• Typical conditions (bulk, global)
• 2011-2012altimeterdata (J1, J2, ENVSat)
• Combination:
• Reduces RMSE
GWES & NAWESVerification Statistics
UMAC data call page 17 of 28Operational wave modeling suite
• CRPS
• Typical conditions (bulk, global)
• 2011-2012altimeterdata (J1, J2, ENVSat)
• Combination:• Reduces CRPS
(MAE equivalent)
GWES & NAWESVerification Statistics
UMAC data call page 18 of 28Operational wave modeling suite 18
• Results of evaluation: BAMS Dec 2013
GWES & NFCENSVerification: BAMS Dec 2013
UMAC data call page 19 of 28Operational wave modeling suite
Great Lakes Wave System (GLW)
• A regional wave model consisting of a single grid including all 5 major lakes, with a 2.5km spatial resolution.
• First operational implementation of a wave system using a curvilinear grid.
• Model has 4 daily cycles run using NAM surface fields, and 4 additional cycles using NDFD surface winds..
• Models target short-term forecasts. NAM cycles run out to 84 h, whereas NDFD cycles run out to 147 h.
• System uses the WAVEWATCH III model.• No data assimilation in the system.
UMAC data call page 20 of 28Operational wave modeling suite
• Domain
Great Lakes Wave System (GLW)
UMAC data call page 21 of 28Operational wave modeling suite
• Future plans for the next 5 years– Include hourly short-range (36 h) forecast sycles using NDFD winds,– Include wind downscaling to improve nearshore forecasts,– Improve ice concentrations intake and, later, include ice model,– Improve nearshore physics via new wave model packages,– Inclusion of a hindcast phase using RTMA-derived surface winds and
analysis wave heights.– Detach NAM cycle components from system, which will become a new
regional wave ensemble system, including GSI-based DA built within NCEP’s RTMA.
– Create a new combined wave ensemble product for the Great Lakes with data from the Canadian Meteorological Service.
– Development of an unstructured grid high resolution Great Lakes model (deterministic model only)
• Coupling with atmospheric and circulation (FVCOM) models
Great Lakes Wave System (GLW)
UMAC data call page 22 of 28Operational wave modeling suite
• Improved physics (2014)– Comparison with old model
(blue) and GLERL reference (green).
– Breakthrough-level improvement to Hs
– Improved bulk statistics (bias, RMS error, correlation)
– Matched/improved GLERL skill
– Higher precision in tracking observations
– Much improved wave periods relative to GLERL
2009
Great Lakes Wave (GLW) Verification
UMAC data call page 23 of 28Operational wave modeling suite
2009• Improved physics (2014)
– Prediction of extreme/storm waves, using high-quality GLERL wind analyses,
– NCEP GLW system provides breakthrough-level improvement in predicting 99% wave heights.
Great Lakes Wave (GLW) Verification
UMAC data call page 24 of 28Operational wave modeling suite
Buoys 45007 & 45149
• Improved physics (2014)
– Prediction of wave heights during post-tropical storm Sandy (Oct 2012),
– Improved physics provided excellent predictions of peak Hs at 2 buoy locations,
– Solved low Hs bias of previous implementation, in extreme storm cases.
Great Lakes Wave (GLW) Verification
UMAC data call page 25 of 28Operational wave modeling suite
• 2015 Upgrades• First operational implementation of curvilinear wave model grid,• Higher spatial resolution (4km -> 2.5 km),• Improved wave fields (red) vs old model (blue).
Great Lakes Wave (GLW) Verification
UMAC data call page 26 of 28Operational wave modeling suite
GLW Reference
Alves, Chawla, Tolman, Schwab, Lang, and Mann, 2014: The Operational Implementation of a Great Lakes Wave Forecasting System at NOAA/NCEP. Wea. Forecasting, 29, 1473–1497.
UMAC data call page 27 of 28Operational wave modeling suite
Wave coupling (new systems)
• Both CFS and RTOFS modeling systems work on different schedules than the current wave operational modeling suites
• New single grid wave domains for coupling with both systems
• Coupling using the NEMS approach• New physics packages for wave – atmosphere and wave –
ocean processes (many already in the wave model)
UMAC data call page 28 of 28Operational wave modeling suite
Requirements
• Model development requirements driven by– SOLAS 1974 is base requirement for wave modeling at NWS.– NWS customer base
• NHC, OPC, WFOS, Regional Headquarters– Partnerships with other NCEP centers
• EMC, MDL, NESDIS, JCSDA– Other government agencies
• GFDL, FNMOC, NRL, USACE, BOEM, NASA, Environment Canada– Scientific Community
• Universities, International Research Centers (e.g. IFREMER)– External users
• International operational agencies (BOM, UKMO, Meteo-France, INCOIS etc.)
• Private enterprises