John Kindle Sergio derada Igor Shulman Ole Martin Smedstad Stephanie Anderson Data Assimilation in Coastal Modeing April 3 2007
MODELS Motivation: Global->Coastal Real-Time Regional Coastal Models Global (NCOM) Models California Monterey Bay Current (MB) System (CCS) 1/12-degree 30-50N, 116-134W GLOBAL HYCOM - Mercator Grid(+bi-polar) (~7Km 35.5-37.5N, -32 Layer ρ/σ/z 121.5-123.5W (σ2) -1/12-degree ~1-4Km Expt 60.5 Resolution (O-M CCS Smedstad) HYCOM - 32 29 Layer σ levels ρ/σ/z (σ2) 1/8-degree - Mercator Orthogonal GLOBAL grid curvilinear NCOM (~7km) -Curvilinear 239x312 grid (~12Km) -41 Level σ/z (19/21) -- GLB8_2f 81x58(operational) 1/12-degree - NCODA data CCS assimilation NCOM - 41 Level σ/z (19/21) - Lat/Lon grid (~9Km) - 163x229 Igor Shulman
The Global Models: HYCOM 1/12º Global HYCOM Configuration Horizontal grid: 1/12 equatorial resolution 4500 x 3298 grid points, ~6.5 km spacing on average, ~3.5 km at pole Mercator 79 S to 47 N, then Arctic dipole patch Vertical coordinate surfaces: 32 hybrid layers with z-levels near surface, sigma in shallow water and isopycnal interior coordinates KPP mixed layer model & thermodynamic sea-ice model Surface forcing: 3 hourly wind stress, wind speed, thermal forcing, precipitation, relaxation to climatological SSS from.5 deg. NOGAPS Monthly river runoff (986 rivers) Initialize from January climatology (GDEM3) T and S, then SSS relaxation from PHC 3.0 NCODA used for data assimilation. Multi-variate OI scheme ( see Smedstad et al poster) Assimilates altimeter SSH ( 3 altimteters), MCSST, observed profiles of T & S Cooper-Haines vertical projection
The Global Models: NCOM 1/8º Global NCOM Configuration Horizontal grid: 1/8 equatorial resolution Curvilinear grid from 78S to 90N Vertical coordinate surfaces: 40 hybrid layers with 19 sigma levels in upper 150m over 21 z-levels. Mellor-Yamada mixed layer model Surface forcing: 3 hourly wind stress, wind speed, thermal forcing, precipitation, relaxation to climatological SSS from.5 deg. NOGAPS Monthly river runoff (1003 rivers) Operationally available sea-surface temperature (MCSST) and altimetry (SSH) data are incorporated into the NAVO Modular Ocean Data Assimilation System (MODAS) and Navy Layered Ocean Model (NLOM) analyses with forecasts of SSH and SST. These surface fields are combined with the MODAS synthetic database to yield threedimensional fields of temperature and salinity for assimilation into global NCOM.
Global Models: Availability 1/8º Global NCOM: NRL LAS-DODS Server for Fields older than 30 days Http://supply.nrlssc.navy.mil:8000/las/servlets/dataset Real-time available through NRL PI s s for research collaborations 1999-present Regional NCOM CCS ( 9 km and 4km) NRL LAS-DODS Server 1999-present in near real time Includes biological fields 1/12 Degree Real-time Global HYCOM Available in future from FSU( COAPS) 100 TB Server 24 hour delay
MODULAR OCEAN DATA ASSIMILATION SYSTEM Synthetic CTD Profiles from Satellite Measurements Climatological Temp MODAS results Actual AXBT Temp SSH + SST + Clim Satellite measurements and MODAS synthetic profile algorithms reveal ocean feature details. MODAS Temperature at 200m This will now be used as a first guess field to blend in any local data.
COMPARISONS (Global Models) Gridded Global NCOM Altimeter with SSH SSH-June SeaWIFS Anomaly15 Chlorophyll: NAVO Data Fusion June 15 Center 2004 June 15 2004 Gridded Global HYCOM Altimeter with SSH SSH-June15 SeaWIFS Anomaly NAVO Chlorophyll: Data Fusion June 15 Center 2004 June 15 2004
Model Evaluation: Regional NCOM Gridded NCOM Regional Altimeter SSH (CCS) June15 with SSH Anomalies SeaWifs NAVO Data Chlorophyll Fusion Center NCOM CCS 9 Km
Regional NCOM US West Coast: 4km Resolution May 16-20 2006 Model Currents & SSH with MODIS Chlorophyll For NW Coast Model Currents & SSH with Model Chlorophyll For NW Coast Model Currents & SSH with MODIS Chlorophyll
/ http://www7320.nrlssc.navy.mil/ccsnrt
NCOMCCS RTM NOGAPS COAMPS Kraken 64PEs. HPQ under reservation 28-minutes NLOM MODAS GNCOM -2d hindcasts NOWCAST +2d forecasts +7d LT forecasts 7GB RAM 350MB DATA
Model Evaluation: Anomaly Correlations Global NCOM and HYCOM and NCOM CCS Global NCOM Regional NCOM Global HYCOM
Model Evaluation: Anomaly Correlations Regional NCOM-CCS CCS With MODAS Assimilation: Open Ocean With MODAS Assimilation: Coastal (200Km) With MODAS Assimilation No Assimilation
Model Evaluation: Tide Gauges Observed Sea level Regional NCOM Gridded Altimeter product
Model Evaluation: Tide Gauges
Evaluation of Global Models: Line 67 CalCOFI Line 67 CalCOFI Line 67 Global Global NCOM NCOM with SSH Anomaly SeaWIFS Chlorophyll: June 15 June 2004 15 2004 Global HYCOM with SSH SeaWIFS Anomaly Chlorophyll: June 15 June 2004 15 2004
Evaluation of Global Models: Line 67 Global NCOM Global HYCOM
Evaluation of Global Models: Cross Sections Global Global NCOM NCOM T with SSH Section@41 Section@36 Anomaly SeaWIFS +/-1 Chlorophyll: June June 15 June 15 2004 2004 15 2004 Global HYCOM T with SSH Section@41 Section@36 SeaWIFS Anomaly+/-1 Chlorophyll: June 15 June 15 2004 15 2004
Downscaling: Strategies & Evaluation NCOMGLBda HYCOMGLB NCOMCCSda HYCOMCCS NCOMCCS MB MBda MB MBda MB MBda MB MBda MB MBda August 2-September 2, 2003 (AOSN period), daily OB update 1-hourly 3-Km COAMPS Forcing, Net HF HYCOMGLB->CCS Regridding Issues MBda: 12hrly incremental updates via NCODA (aircraft and glider data) Tidal Forcing from OSU Flather and Orlansky BC scheme, MY2.5 Mixing Coupled to COSIN Ecosystem Model ( See Igor Shulman for more)
NESTING IMPLEMENTATION Generalized Methodology for mapping vertical/horizontal grids On the HYCOM (source grid) side: Outer Nest HYCOM ρ/σ/z HYCOM tools -layer to z HYCOM TOOLS Intermediate Z Grid NCOM TOOLS to generate and intermediate Z grid. ( Suggested to match NCOM Z-structure) *NCODA coupling on Z-Grid On the NCOM (target grid) side: Inner Nest NCOM σ/z Use REGRID_GEN to interpolate: -to final σ or σ/z grid - horizontally (optional) NCOM SETUP tools to create final: OINIT and OPNBC
Downscaling: Strategies & Evaluation Concentration in the AOSN Period (August 2003) AVHRR on August 15, 2003
Downscaling: Strategies & Evaluation Concentration in the AOSN Period (August 2003) AVHRR on August 15, 2003 10m winds at M1 and M2 BUOYS (hourly) NOGAPS COAMPS 81Km 9KM (hourly) 3KM (3-hourly) (3-hourly) UPWELLING Relaxation
Downscaling: Strategies & Evaluation Concentration in the AOSN Period (August 2003) AVHRR on August 15, 2003 Model Response NCOMGLB->MB
Downscaling: Strategies & Evaluation Concentration in the AOSN Period (August 2003) AVHRR on August 15, 2003 Model Response NCOMGLB->MBda
Downscaling: Strategies & Evaluation Global NCOM---MB ( No Assimil) Global NCOM---MB ( With Assimil) SST at M1 and M2 No Assim.
Downscaling: Strategies & Evaluation Concentration in the AOSN Period (August 2003) AVHRR on August 15, 2003 Model Response NCOMGLB->NCOMCCS->MB
Downscaling: Strategies & Evaluation Concentration in the AOSN Period (August 2003) AVHRR on August 15, 2003 Model Response NCOMGLB->NCOMCCS->MBda
Downscaling: Strategies & Evaluation Global NCOM---NCOMCCS---MB ( No Assimil) Global NCOM---NCOMCCS---MB ( With Assimil) SST at M1 and M2 No Assim.
Downscaling: Strategies & Evaluation Concentration in the AOSN Period (August 2003) AVHRR on August 15, 2003 Model Response NCOMGLB->NCOMCCS->MBda (best comparison)
Downscaling: Strategies & Evaluation August 15, 2003 Nested in Global NCOMGLB->MBda Nested in Regional NCOMGLB->NCOMCCS->MBda SST at M1 and M2
CONCLUSIONS Real-time global NCOM and HYCOM simulations are being conducted, the results of which will be made available to the community for coastal modeling efforts NCOM Global( 1/8 th deg): NRL LAS-DODS after 30 day delay HYCOM Global( 1/12 th deg.): FSU COAPS LAS-DODS Server NCOM CCS ( 9 and 4 km): NRL LAS-DODS server: 24 hr delay NCOM is the Navy operational model; HYCOM will be transitioned to NAVO for operational testing in Fall 2007 Both Global NCOM and HYCOM provide adequate representations of mesoscale variability in CCS domain Thermocline spread is a known issue for both in CCS; problem is being addressed Can provide adequate boundary values for higher resolution coastal models. Results are consistent with expectations. The closer the boundary values are to observations, the less burden is placed on the assimilation scheme for the hi-res model