Great Lakes Storm Surge Study

Principle Investigator: Mary A. Cialone
HPC System: Diamond
Sponsor: Federal Emergency Management Agency (FEMA)

Video

Visualization Details:

The Data Analysis and Assessment Center (DAAC) was requested to visualize the results of an ADCIRC storm-event simulation. The simulation modeled an actual storm that lasted 12 days, from 25 November to 07 December 1990. The boundaries of the simulation were restricted to the coast lines of these two Great Lakes for simplicity. A customized Visualization Toolkit (VTK) reader was used to translate the ADCIRC data into Legacy VTK format, which enabled it to be read into ParaView for processing. The mesh consisted of 409,264 nodes and 777,626 cells, and was greatly refined along the coastline edges of the Great Lakes. Water-level solutions, as well as other data parameters, were recorded every 5 minutes over the 12-day event, yielding 3456 time-steps. The total size of the data was 46 GBytes.

Some visualizations require stretching an X-, Y-, or Z-dimension of a dataset in order to detect small features that otherwise would go unnoticed. This is typical in oceanic datasets where square miles of water area are in view, but surface characteristics, based on a few feet, are the focus of the study. In these cases, the height of the surface geometry needs to be uniformly up-scaled to see these variations.

The DAAC employed a fast, dynamic method for selecting an effective exaggeration amount by utilizing some image-processing capabilities of AutoDesk's 3D Studio Max rendering software. Our technique used a series of high-resolution height-mapped images rather than traditional fixed polygonal geometry. In 3D Studio Max, we selected the flat water suface and deformed it by activating a displacement modifier to warp the water surface geometry. The geometry modifier worked in combination with greyscale of the height-mapped images, pure black warped the surface geometry downward, pure white warped it upwards, fifty- percent gray left the geometry unchanged. The degree of exaggeration was a matter of adjusting the deformation settings of the geometry modifier. This method allows playback preview over all time steps of the data (in wireframe) before any images are actually rendered. Refining the amount of exaggeration and previewing the results was accomplished in seconds. Once the exaggeration amount was selected, the images were rendered.

Simulation Details:

The U.S. Federal Emergency Management Agency (FEMA) Region V has undertaken the task of updating the Great Lakes flood risk maps and has contracted with the U.S. Army Engineer Research and Development Center (ERDC) to develop wind, wave, and water-level climates to define the storm-event population that will be used to estimate the potential for coastal flooding in Lake Michigan. The techniques and procedures developed by ERDC will then be applied to the remaining Great Lakes by other FEMA contractors. The approach being taken is to simulate wave and storm-surge conditions associated with 150 meteorological events via computer modeling. These modeled storm water-level and wave data form the foundational database to be used in updating the Great Lakes flood risk maps.

The goals of the visualizations were to relate insights gained from the simulation. This study focused on two main areas of interest. First, understanding how oscillatory modes can be induced by wind and atmospheric pressure gradients over the Great Lakes regions, including how those modes affect water-level fluctuations across these large lakes. A second point of interest included analyzing flood threats resulting from water-level fluctuations. Analyzing the first focus point required a view of all Lake Michigan and Lake Huron from a distant viewpoint. The second focus of the study needed visualizations isolated to the Green Bay area where maximum flooding occurred.