Coastal Ocean Impact on Hurricane Irene Intensity

In 2011, Hurricane Irene's intensity (using maximum sustained 10m winds) was over-predicted by several hurricane models and over-forecast by the National Hurricane Center (NHC). The NHC final report on Irene lists possible factors leading to the consistent high bias in the official intensity forecasts (an incomplete eyewall replacement cycle while in an environment of relatively light wind shear and over the warm waters of the South Atlantic Bight) and even a high bias in the operational analysis of its intensity (deep central pressure and strong flight-level winds but comparatively low surface winds).

This paper investigates one potentially major control on Irene's intensity in its latest stages: the upper coastal ocean's impact in the hours before NJ landfall. Using buoy, satellite, and underwater glider observations, we determine the magnitude, timing, and spatial structure of the surface ocean cooling in the Mid-Atlantic Bight. These observations indicate 4-6°C to as much as 11°C surface cooling occurring primarily ahead of the eye's passage, presumably under the first outer wind bands of Irene. Further, we combine temperature and depth-averaged current data from the glider with HF radar surface current measurements to attain physical understanding of the turbulent mixing occurring at the thermocline due to shear between strong onshore surface currents and opposing offshore bottom currents. These physical conclusions are confirmed with the Regional Ocean Modeling System (ROMS).

With this knowledge, we conduct a series of over 100 simulations with the Weather Research and Forecasting (WRF) model to compare the sensitivity of Irene's intensity to 1) model setup [e.g. horizontal, vertical resolution; boundary conditions], 2) atmospheric physics [e.g. microphysics, planetary boundary layer scheme, radiation scheme], 3) Advanced Hurricane WRF options [e.g. parameterizations of air-sea fluxes (momentum, latent heat, and sensible heat fluxes), 1D ocean mixed layer model], and 4) sea surface temperature (SST). We find one of the largest sensitivities using satellite-observed cold post-storm SST (a new coldest dark pixel composite developed at Rutgers University) vs. fixed warm pre-storm SST (Real-Time Global, RTG SST). This sensitivity totaled over 35 m/s in maximum sustained 10m winds across 13 hours of simulation (nearly 2.7 m/s each hour) while the storm was over the MAB and New York Harbor. With the most correct SST and advanced air-sea flux parameterizations available, we achieve the lowest RMSE (0.83 m/s vs. up to 3.7 m/s elsewhere) and bias (-1.1 m/s vs. up to +16.4 m/s elsewhere) across the same 13 hours of simulation.