Initial Tests of an Automated Operational Storm Surge Prediction System for the National Hurricane Center

Initial Tests of an Automated Operational Storm Surge Prediction System for the National Hurricane Center

CRISTINA FORBES AND JAMIE RHOME

NOAA/National Weather Service/National Centers for Environmental Prediction/National Hurricane Center

11691 SW 17th Street, Miami, Florida 33165

An automated real-time operational storm surge prediction system for the Sea Lake and Overland Surges from Hurricanes (SLOSH) model, which is run operationally during tropical cyclone events, has been developed for the National Hurricane Center (NHC).

AutoSurge automates and accelerates operational workflows, computes storm track input parameters with greater accuracy, eliminates labor-intensive tasks, and prevents human input error, giving storm surge forecasters additional time to conduct model analyses, generate forecast guidance products, calculate model output statistics, and assess model results.

Shortly after synoptic times, the input wind parameters for SLOSH are extracted from the NHC's best track, official track and intensity forecasts, and the NHC track and intensity models, enabling AutoSurge to run off any and all of the guidance available within the Automated Tropical Cyclone Forecast (ATCF) system.

Implementing the SLOSH wind model directly into the operational workflow ensures that the computed wind parameters are consistent with the SLOSH wind formulation. The grid basins for the ensuing SLOSH storm surge predictions are selected by the Storm Surge Specialist or automatically by AutoSurge. The basins can also be prioritized according to the distribution of the probabilistic P-Surge (Taylor and Glahn 2008) product's ensemble tracks, created from past performance statistics of the forecast advisories.

The closest point of approach is estimated from the selected basins' coastline locations and the tropical cyclone track. A cubic spline interpolation algorithm is applied to the storm track to provide the temporal resolution required to run SLOSH.

The system automatically generates a vast array of guidance products from SLOSH model output, including the maximum envelope of water and the temporal evolution of surge in areas that could be impacted by the storm.

In the present study, we build upon previous SLOSH verification analyses by developing an objective methodology for configuring, running and assessing the SLOSH simulations that can be replicated over multiple storms and for different forecast times and areas.

This is accomplished by employing real-time forecast simulations from Hurricane Irene (2011) and including a temporal analysis through the use of NOAA tide station time series. Such efforts are essential to evaluate current and future SLOSH upgrades, as well as to allow robust comparisons with other hydrodynamic modeling systems.

An example of the forecast guidance provided by AutoSurge for three different SLOSH basins (eht2 – Cape Hatteras, NC; orf – Norfolk, VA; and ny3 – New York, NY) is shown in Fig. 1.

FIG. 1. Evolution of the ensemble water levels generated by AutoSurge from different ATCF models/aids over time in the eht2, orf, and ny3 basins during Hurricane Irene (2011).

The ensemble members' (red dots) predicted maximum and minimum (yellow shade) water level spread, average (red middle line) and standard deviation (or 68% confidence interval assuming a normal distribution, pink shade) from different ATCF trackers are displayed. The average value and the standard deviation are highest for the eht2 basin and smallest for the ny3 basin. The NHC storm surge predictions from the forecast advisories at 24 and 12 hours before landfall are shown as well.

As part of the storm validation procedure, hindcast simulations of Hurricane Irene were run with the AutoSurge hindcast option selected. This mode of operation uses best track information exclusively. The simulations were run after the storm had passed through all available basins as part of a post-storm analysis. The maximum simulated water levels using the best track are shown in Fig.1.

Comparisons between observed and modeled water levels and their evolution in time for different SLOSH basins will be described. The performance of the simulations will be quantified by presenting the root mean square error and correlations of the observed and SLOSH-simulated surface water elevations.

AutoSurge is currently undergoing rigorous testing in NHC's Storm Surge Unit during the 2011 hurricane season and will likely be deployed for operational use during the 2012 hurricane season.

References

Forbes, C. and Rhome, J. 2011: An Automated Operational Storm Surge Prediction System for the National Hurricane Center, Estuarine and Coastal Modeling XII, M. L. Spaulding [ed], ASCE, submitted.

Taylor, A. A. and B. Glahn, 2008: Probabilistic guidance for hurricane storm surge, 88^th Annual Meeting of the American Meteorological Society (20-24 January 2008), New Orleans, LA, 7.4, 8 pp.