1. Richard M. Hodur Marine Meteorology Division Naval Research Laboratory Monterey, CA NRL Monterey Plans for Tropical Cyclone Modeling using the Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS) 61 st Interdepartmental Hurricane Conference, New Orleans, LA 6 March 2007 COAMPS® is a registered trademark of the Naval Research Laboratory

2. NRL Monterey Plans for Tropical Cyclone Modeling using COAMPS Outline Introduction/Background Modeling Infrastructure TC Related Research Analysis/Initialization Boundary Layer Moving, Nested Grids Air/Ocean Coupling Ensemble Prediction Summary

3. NRL Monterey Plans for Tropical Cyclone Modeling using COAMPS Introduction/Background COAMPS operational at FNMOC since 1998 Operational use primarily for mid-latitudes Atmosphere only 0-3 day forecasts Not used operationally as a ″TC model″ Non-optimal grid configurations High-resolution nest not centered on TC NRL research now focused on TC analysis and prediction Infrastructure Analysis Physical parameterizations Air-Ocean-Wave Coupling ″COAMPS-TC″ scheduled for operational use in FY08

4. NRL Monterey Plans for Tropical Cyclone Modeling using COAMPS Outline Introduction/Background Modeling Infrastructure TC Related Research Analysis/Initialization Boundary Layer Moving, Nested Grids Air/Ocean Coupling Ensemble Prediction Summary

5. COAMPS Modeling Infrastructure Atmospheric Model Dynamics Physics

6. COAMPS Modeling Infrastructure Dynamics Physics Atmospheric ModelOcean Model Dynamics Physics

7. COAMPS Modeling Infrastructure Dynamics Physics Atmospheric ModelOcean Model ESMF Modeling Superstructure Dynamics Physics ESMF superstructure used to couple models

8. COAMPS Modeling Infrastructure Flux Coupler Dynamics Physics Atmospheric ModelOcean Model ESMF Modeling Superstructure Dynamics Physics ESMF superstructure used to couple models Flux coupler transfers fields between models

9. COAMPS Modeling Infrastructure ESMF superstructure used to couple models Flux coupler transfers fields between models WRF interface allows for different physics suites Dynamics WRF Physics Interface to Physics Suites NRL NCARNCEP Flux Coupler Dynamics Physics Atmospheric ModelOcean Model ESMF Modeling Superstructure

10. COAMPS Modeling Infrastructure ESMF superstructure used to couple models Flux coupler transfers fields between models WRF interface allows for different physics suites Dynamics WRF Physics Interface to Physics Suites NRL NCARNCEP Flux Coupler Dynamics Physics Atmospheric ModelOcean Model ESMF Modeling Superstructure Bottom Line: Share Community Models/Share Community Physics

11. COAMPS Modeling Infrastructure ESMF superstructure used to couple models Flux coupler transfers fields between models WRF interface allows for different physics suites Bottom Line: Share Community Models/Share Community Physics ESMF Modeling Superstructure Atmospheric ModelOcean ModelWave Model Dynamics WRF Physics Interface to Physics Suites NRL NCARNCEP Flux Coupler Dynamics Physics Dynamics Physics Relatively easy to add new modeling components

12. Using ESMF to couple models Using WRF to include physics suites Advantages: Easy to couple models Can test many physics options Lessons Learned Large overhead in running “plug-and-play” physics Learn from past/present: Set community standards Grid orientation (i,j,k) Input variables (i.e., namelist control) Standard way of building models/systems Collaborate w/NCEP, NCAR, others on ESMF development/use Treat computer science as science COAMPS Modeling Infrastructure

13. COAMPS Coupled Ocean/Atmosphere Mesoscale Prediction System Complex Data Quality Control 3DVAR analysis: u, v, T, q v, radiances; observation-space based; supports nested grids; includes synthetic TC observations Initialization: Hydrostatic Constraint on Analysis Increments, and/or Digital Filter Navy Coupled Ocean Data Assimilation (NCODA) System 2D OI: SST 3D MVOI: T, S, SSH, Sea Ice, Currents Complex Data Quality Control Initialization: Stability check Numerics: Nonhydrostatic, Scheme C, Nested Grids, Sigma-z, Flexible Lateral BCs Physics: PBL, Convection, Explicit Moist Physics, Radiation, Surface Layer Aerosols: Surface databases, High-order Transport, Dry Deposition, Wet Removal NRL Coastal Ocean Model (NCOM) Numerics: Hydrostatic, Scheme C, Nested Grids, Hybrid Sigma/z Physics: Mellor-Yamada 2.5 Flux Coupler Ocean Analysis Ocean ModelAtmospheric Model Atmospheric Analysis Globally Relocatable: 5 Map Projections User-Defined Grid Resolutions, Dimensions, and Number of Nested/Parent Grids Applicable for Idealized or Real-Time Applications Single Configuration Managed System for All Applications FNMOC Operations: 8 Areas, 4 runs/day, grid spacing: 6 km, 30 levels; forecasts to 72 hours Features

14. NRL Monterey Plans for Tropical Cyclone Modeling using COAMPS Outline Introduction/Background Modeling Infrastructure TC Related Research Analysis/Initialization Boundary Layer Moving, Nested Grids Air/Ocean Coupling Ensemble Prediction Summary

15. High-resolution Synthetic Observations for TC Initialization Case: 0000 UTC 9 September 2000 Synthetic Observations Built From: Modified Rankine Vortex JTWC Warning Message w/Satellite Data NOGAPS T20/L15 truncated fields Blend Synthetics w/all other observations in MVOI NOGAPS Synthetics COAMPS Synthetics Improved TC representation with new synthetics Improved TC Track Forecasts with new synthetics Higher resolution with new synthetics Forecast Track Error (km) Forecast Time (Hours) NOGAPS Synthetics COAMPS Synthetics (74 cases)

16. Satellite Photo at Analysis Time TC Analysis: NAVDAS/Relocation 850 mb Wind, Analysis Time: 0000 UTC 29 September 2005

17. Satellite Photo at Analysis Time TC Analysis: NAVDAS/Relocation 850 mb Wind, Analysis Time: 0000 UTC 29 September 2005 First Guess First-guess field has TC circulation in wrong location

18. Satellite Photo at Analysis Time TC Analysis: NAVDAS/Relocation 850 mb Wind, Analysis Time: 0000 UTC 29 September 2005 First Guess First-guess field has TC circulation in wrong location Analysis w/o Relocation NAVDAS w/synthetic obs creates elongated TC circulation

19. Satellite Photo at Analysis Time TC Analysis: NAVDAS/Relocation 850 mb Wind, Analysis Time: 0000 UTC 29 September 2005 First Guess First-guess field has TC circulation in wrong location Analysis w/o Relocation NAVDAS w/synthetic obs creates elongated TC circulation Analysis w/Relocation Relocation improves analysis position of TC circulation

20. Satellite Photo at Analysis Time TC Analysis: NAVDAS/Relocation 850 mb Wind, Analysis Time: 0000 UTC 29 September 2005 First Guess First-guess field has TC circulation in wrong location Analysis w/o Relocation NAVDAS w/synthetic obs creates elongated TC circulation Analysis w/Relocation Relocation improves analysis position of TC circulation Also studying impact of using “spin-up” storm and NAVDAS options

21. Options: 2D OI SST analysis 3D MVOI analysis (T, S, SSH, Currents, Ice Concentration, Significant Wave Height) Features Integrated with COAMPS Assimilates MCSST; GOES, LAC, Ship, and Buoy SST; XBT; CTD; PALACE Float; Fixed and Drifting Buoy; Altimeter SSHA; SSM/I Sea Ice; Altimeter and Buoy SWH Nested grids 5 Map projections Integrated w/Modular Ocean Data Assimilation System (MODAS) databases Transitions to FNMOC: FY99: 2D SST analysis for COAMPS FY02: New QC code FY03: 3D ocean MVOI; 2D SST for NOGAPS Supports: Atmosphere: NOGAPS, COAMPS Ocean: NCOM, HYCOM, SWAFS, WW3 NCODA NRL Coupled Ocean Data Assimilation Focus: Optimal fit of observational and model data Increased resolution yields improved depiction of details in the SST field 81 km 27 km 9 km On-going Research: New data types: Currents: HF Radar, Drifters SST: Microwave, Skin SST from AATSR Detection and Correction of Aerosol Contamination in Infrared Satellite SST Retrievals Convert to 3d variational analysis

22. NRL Monterey Plans for Tropical Cyclone Modeling using COAMPS Outline Introduction/Background Modeling Infrastructure TC Related Research Analysis/Initialization Boundary Layer Moving, Nested Grids Air/Ocean Coupling Ensemble Prediction Summary

23. CBLAST Coupled Boundary Layers/Air-Sea Transfer ONR DRI Objective Understand air-sea interaction in the coastal environment and improve boundary layer parameterization in COAMPS for low- and high-wind events over the ocean Approach Perform COAMPS simulations down to LES scales, and analyze CBLAST observational data Black and Chen (2006) CBLAST measurements (heavy black line with asterisks) suggest that the drag coefficient does not vary with wind speed as much as earlier estimates for winds over about 21 m/s. CBLAST Uncertainty also exists for C E /C D

24. COAMPS 4.2.4 COAMPS new Observed SLP COAMPS w/ARWSFC 1000 980 960 940 920 900 880 200406080 Forecast Time (h) Central Pressure (mb) CBLAST Coupled Boundary Layers/Air-Sea Transfer ONR DRI COAMPS 4.2.4 Based on Louis (1979) and COARE (Fairall 1996; 2003) COAMPS new Based on COAMPS 4.2.4, with constant drag coefficient over 22 m/s and modified z 0q and z 0h COAMPS w/ARWSFC COAMPS model implemented with the WRF ARW surface layer parameterization COAMPS TC intensity forecasts have strong sensitivity to surface flux parameters

25. NRL Monterey Plans for Tropical Cyclone Modeling using COAMPS Outline Introduction/Background Modeling Infrastructure TC Related Research Analysis/Initialization Boundary Layer Moving, Nested Grids Air/Ocean Coupling Ensemble Prediction Summary

26. COAMPS MPI Moving Nests Hurricane Gordon: 00Z September 17- 00Z September 19, 2000 Forecast position every 6 hours 0 h 48 h 81 km (61x61) 27 km (121x85) Fixed Nest Option

27. COAMPS MPI Moving Nests Hurricane Gordon: 00Z September 17- 00Z September 19, 2000 Forecast position every 6 hours 0 h 48 h 81 km (61x61) 27 km (121x85) Fixed Nest Option Nest 2 at tau = 0 h Nest 2 at tau = 48 h 81 km (61x61) 27 km (46x46) Moving Nest Option Forecast position every 6 hours

28. COAMPS MPI Moving Nests Hurricane Gordon: 00Z September 17- 00Z September 19, 2000 Forecast position every 6 hours 0 h 48 h 81 km (61x61) 27 km (121x85) Fixed Nest Option Nest 2 at tau = 0 h Nest 2 at tau = 48 h 81 km (61x61) 27 km (46x46) Moving Nest Option Forecast position every 6 hours Results Identical Track Predictions for Fixed and Moving Nests Moving Nest Option: 2.7x Faster

29. COAMPS MPI Moving Nests Hurricane Gordon: 00Z September 17- 00Z September 19, 2000 Forecast position every 6 hours 0 h 48 h 81 km (61x61) 27 km (121x85) Fixed Nest Option Nest 2 at tau = 0 h Nest 2 at tau = 48 h 81 km (61x61) 27 km (46x46) Moving Nest Option Results Identical Track Predictions for Fixed and Moving Nests Moving Nest Option: 2.7x Faster Recently Added Two-Way Interaction between all Nests Forecast position every 6 hours

30. NRL Monterey Plans for Tropical Cyclone Modeling using COAMPS Outline Introduction/Background Modeling Infrastructure TC Related Research Analysis/Initialization Boundary Layer Moving, Nested Grids Air/Ocean Coupling Ensemble Prediction Summary

31. Coastal Upwelling Adaptive Ocean Sampling Network II (AOSN-II) Horizontal Resolution Sensitivity/Wind Stress Validation COAMPS Forecast M1 Observations August 2003 Results: Improved representation of the wind stress curl using the 3 km grid at the coast leads to improved representation of upwelling COAMPS 3 km Forecast Surface Stress Compares Favorably to Observed Stress at M1 Buoy 3 km 9 km Representation of Coastal Jets, Wind Stress Curl, and Coastal Shear Zones Improved using Higher Resolution Grid 27 km  M1 Wind Speed (m/s) 121086420 The leftmost 3 boxes show COAMPS wind speed (color) and direction (arrows) for 27, 9, and 3 km grids Graphs on right show observed (upper) and COAMPS (lower) Surface Stress

32. One-Way vs.Two-Way Coupling Air-Sea Interaction in the Adriatic Sea NCOM Initial Conditions (23 Sept 2002): T/S Analysis from CTD Casts (SACLANTCEN) 5-day diagnostic run using fixed hourly atmospheric forcing for 23 Sept 2. Two-Way Coupling 12 h incremental update cycle MVOI 6h Atmosphere Fcst NCOM FCST w/hourly forcing, SST to atm every 6 hrs NCOM SST 1. One-Way Coupling 12 h incremental update cycle MVOI 6h Atmosphere Fcst NCOM FCST w/hourly forcing Repeat SST Analysis Repeat Forecasts of Adriatic include a 4 km COAMPS nest and a 2 km NCOM grid Data Assimilation and Forecasts were run for the period of 23 Sept through 23 Oct 2002, which exhibited Bora conditions 28 24 20 16 12 8 4 0 Wind Speed (m/s) Bor a

33. Mean Potential Temperature for Two-way coupling (upper) and [Two-way] - [One-way] differences (lower), showing colder atmosphere mixed layer and warmer ocean mixed layer with two-way coupling. One-Way vs.Two-Way Coupling Air-Sea Interaction in the Adriatic Sea 30 day means: 23 Sept – 23 Oct 2002 One-way coupled Sensible Heat Flux Two-way coupled Sensible Heat Flux One-way coupled SST Two-way coupled SST Two-way coupling exhibits smaller sensible heat fluxes (left), presumably from mutual adjustment in atmosphere and ocean mixed layers. As a result, NCOM SSTs (right) from the two-way coupled system are slightly warmer than those from the one-way coupled system. Lowest SST and near-surface wind RMS and Bias Errors found using 2-way Coupling

34. Prototype Two-Way Air/Ocean Coupling in COAMPS Hurricane Frances

35. NRL Monterey Plans for Tropical Cyclone Modeling using COAMPS Outline Introduction/Background Modeling Infrastructure TC Related Research Analysis/Initialization Boundary Layer Moving, Nested Grids Air/Ocean Coupling Ensemble Prediction Summary

36. Mesoscale Ensemble Prediction System COAMPS: Coupled Ocean/Atmosphere Mesoscale Prediction System Mesoscale Ensemble Prediction System Development Collaborative w/Joint Ensemble Forecast System (JEFS) Navy/AF FNMOC/AFWA Computers 29-member COAMPS ensemble: Based on 29-member NOGAPS ensemble 6-hour update cycle Relocatable domain Initial condition perturbations Lateral boundary condition uncertainty Model uncertainty Mean of ensembles represents west coast stratus better than the control run or any ensemble member generated using different boundary layer parameters CTRL Mean l1l1 l2l2 S h1 S h2 Satellite photo at validation time CTRL is the control forecast Mean is the average of all the ensemble members I 1, I 2, S h1, and S h2 are members using different values of boundary layer parameters

37. NRL Monterey Plans for Tropical Cyclone Modeling using COAMPS Summary Current Research Infrastructure Analysis Boundary Layer Moving, Nested Grids Air/Ocean/Wave Coupling Ensemble Prediction Schedule FY07 Two-Way Air/Ocean Beta Test Surface Flux/Moist Physics Upgrades to FNMOC FY08 Initial “COAMPS for TCs” to FNMOC Improved TC analysis (e.g., relocation, spin-up) Initial installation will be uncoupled Air/Wave Coupling Development FY09+ Air/Ocean Coupling to operations Air/Wave Coupling to operations

38. Richard M. Hodur Marine Meteorology Division Naval Research Laboratory Monterey, CA NRL Monterey Plans for Tropical Cyclone Modeling using the Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS) 61 st Interdepartmental Hurricane Conference, New Orleans, LA 6 March 2007 COAMPS® is a registered trademark of the Naval Research Laboratory