Virginia Institute of Marine Science

VIMS team pursues street-level storm-tide predictions

With the official start of the 2008 Atlantic hurricane season on June 1, a team of computer modelers at the Virginia Institute of Marine Science continues to pursue its long-term goal of working with government, academic, and industry partners to provide street-level predictions of storm-tide flooding along the Chesapeake Bay shoreline.

Team leader Harry Wang, an associate professor at VIMS, notes that emergency managers will be able to use this information to alert individual neighborhoods about appropriate protective measures and possible evacuation during hurricanes and nor’easters. Wang estimates that street-level predictions will be possible in five years.

Predicting storm-tide flooding in Chesapeake Bay is made difficult by the countless creeks, coves, and tributaries that form the Bay’s 11,000-mile-plus shoreline. Depending on wind speed and direction, rainfall amounts in a particular watershed, tide levels, and other factors, water levels during a storm can differ significantly along shorelines that lie only a few tens of miles apart.

Given these difficulties, storm-tide predictions are currently limited to broad-brush warnings, such as “a 3-foot storm tide for the lower Bay.”

Wang’s team, which includes VIMS researchers John Boon, Jian Shen, Kyoung-ho Cho, David Forrest, Leonidas Linardakis, Wenping Gong, and Tao Shen, is striving to make storm-tide warnings that are both more accurate and more localized, thereby increasing their utility for those charged with protecting lives, homes, businesses, roads, and utilities around the Bay.

The VIMS team is collaborating on the project with other members of the NOAA-funded Chesapeake Bay Inundation Prediction System. CIPS, which is coordinated by Ms. Elizabeth Smith of Old Dominion University, includes participants from the Virginia Department of Emergency Management; the National Weather Service; the US Geological Survey; NOAA’s Chesapeake Bay Office; the University of Maryland (UMD); and two private companies: Weatherflow, Inc of Poquoson and Noblis of Falls Church.

VIMS’ role in the project is to develop and refine their state-of-the art computer model for predicting storm tides and inundation. A similar effort is underway at the Univ. of Maryland. The strengths of each model will ultimately be combined to produce a real-time “ensemble” model for operational forecasts of Bay flooding.

The storm-tide model being developed at VIMS improves on the National Weather Service’s current “SLOSH” model in several key areas. One is the newer model’s much higher resolution; another is its use of an “unstructured grid.” A third key advance is its larger “domain.”

Wang says that “Combining our higher resolution model grid with LIDAR data allows us to simulate the Bay’s complicated shoreline much more accurately, thus bringing our forecasts potentially down to street level.” LIDAR, for Laser Detection and Ranging, is a new aerial-mapping technology that resolves surface topography on a much finer scale than traditional mapping.

Wang’s team demonstrated the potential of the combined technologies in a recent pilot study of flooding in the Potomac River during Hurricane Isabel. The model reproduced flood levels within a few centimeters of those actually observed.

In order for their model to achieve its full potential, Wang’s team needs to overcome several challenges. One is to obtain the LIDAR elevation data needed to accurately simulate flooding along any particular stretch of shoreline. LIDAR provides a vertical resolution of 15 cm (6 in.) and a horizontal resolution of 1 meter (3 feet). This is a significant improvement on the current SLOSH model, which has a vertical resolution of about a half a meter (20 in.), and is applied to topographic base-maps that denote elevation in terms of football-field-sized squares. Maryland has already used LIDAR to map its entire coastline; Virginia has yet to do so.

Another challenge, says Wang, is “to cut down the run time for operational forecasting.” Currently, the model takes 1 hour to simulate each day. Linardakis, a postdoctoral fellow in high-performance computing at VIMS, is exploring “parallel computing” (in which the model runs on several computers simultaneously) as a means to speed things up. Use of parallel computing will be especially needed if the modelers add waves to their forecast mix. A trial run in 2007 showed that adding wave effects to a storm-tide prediction for the York River made the computer run 100 times more slowly.

A final challenge, shared by storm forecasters everywhere, is accurately timing the arrival of the storm surge with the background astronomical tide. If the surge of water pushed by a storm’s winds arrives at high tide, water levels at some places in the Bay could be almost 4 feet higher than they would be if the same storm surge coincided with a low spring tide.

VIMS Emeritus Professor John Boon is leading the efforts on this front, working to deploy a network of real-time tide gauges at Jamestown Island, the Back River, and other sites around the lower Bay. Eventually, data from these gauges will be transmitted in real-time for incorporation into operational storm-tide forecasts.