ASSIMILATION OF HIGH-FREQUENCY RADAR SURFACE CURRENTS INTO A COASTAL OCEAN MODEL OF THE MIDDLE ATLANTIC BIGHT Alan F. Blumberg George Meade Bond Professor Director Davidson Laboratory Stevens Institute of Technology Liang Kuang and Nickitas Georgas I EEE-MTS 12 Ocean Meeting October 17, 2012
New York Harbor Observing and Prediction System Integrated system of observing sensors and forecast models TO OBSERVE TO PREDICT TO COMMUNICATE Weather Currents Water Level Salinity Temperature Waves
How? Observe Ground-Truth Serve Automatically Forecast
A fully automated system of systems New York Harbor Observing and Prediction System 0.5 hrs hrs hrs
C:\Documents and Settings\hroarty\My Documents\COOL\01 CODAR\MARCOOS\Renewal HF radar System
6 SLDMB Drifter
Methodology—Data Assimilation Data Assimilation- Nudging Scheme 7
Non-tidal mean surface currents: HF radar vs. NYHOPS BeforeAfter From Jun 9 th, 2011 to Jul 21 st, Scale is in 10cm/s. 8
Tidal currents(M2 ellipses) after DA Before After From Jun 9 th, 2011 to Jul 21 st, Scale is in 10cm/s. 9
10 RMSE between NYHOPS Hindcast, Drifter currents before and after data assimilation (cm/s) UU_DAU_diffVV_DAV_diff A B C average Positive means improvement
11 43 (3X) Reseeding particle-tracking simulations
12 RMSE of NYHOPS Forecast, Drifter currents before and after data assimilation (cm/s) UU_DAU_diffVV_DAV_diff A B C Average
13 43 (3X) Reseeding particle-tracking simulations
Conclusions NYHOPS established as an urban ocean forecast system – large following with multiple constituencies Using currents derived from drifters for validation: Average RMS errors of hindcast and 1 day forecast shows 8% improvements Particle-tracking simulations showed improvements of 7% (hindcast) and 10% ( 1 day forecast) based on separation distances The future work - assimilation using more advanced schemes, such as Kalman Filter/LRTKF, 3D and 4D var 14