High Resolution Ensemble Storm Surge Predictions for Superstorm Sandy Around the New York City Region

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 6 February 2014: 4:45 PM
Georgia Ballroom 2 (The Georgia World Congress Center )
Brian A. Colle, Stony Brook University/SUNY, Stony Brook, NY; and J. Kuang, H. Bowman, M. J. Bowman, C. Flagg, F. Zhang, Y. Weng, and E. B. Munsell

Superstorm Sandy resulted in catastrophic loss of life and property for Metro New York (NY), New Jersey (NJ) and Long Island (LI). The track and intensity of the storm was relatively well predicted a few days in advance, yet sea level surge predictions from the various research and operational surge models (e.g., Stony Brook's STORMY, Stevens Institute's NYHOPS, NOAA's ET system) significantly underestimated the peak surge by almost a meter. In addition, the mostly widely used NOAA/FEMA/USACE models for the characterization/mapping of coastal inundation (flooding) threats for Metro New York, northern New Jersey and coastal Long Island also suffered from limitations. For example, observed inundations during Sandy in some locations overran the FEMA estimated 500-year flood risk contour.

This presentation will first highlight the surge predictions made for Sandy using the Advanced Circulation (ADCIRC) model using a nested 3-km Weather Research and Forecasting (WRF) grid following Sandy. These runs will highlight the impact of lead time (2 day versus 4-day forecast), importance ADCIRC resolution in some of the narrow bays and channels (down to 30 m resolution), impact of the surge model domain given Sandy's relatively large wind field, and influence of wave forcing on the surge predictions. The atmospheric prediction uncertainty was quantified using an Ensemble Kalman Filter (EnKF), which uses an ensemble of forecasts to estimate flow-dependent background error covariance for the data assimilation cycling. The EnKF system assimilated NOAA hurricane hunter aircraft observations (radar and flight level) as well as other observational systems (satellite winds, surface data, and rawinsondes) to create a 60-member ensemble at 3-km grid spacing. This ensemble system allows for a better probabilistic estimate of the surface winds and pressures around Sandy. The sensitivity of storm surge predictions to relatively small changes in the track and intensity of Sandy in this ensemble will be highlighted as well as probabilistic benefits of using this ensemble as compared to a single deterministic surge prediction.