InfobITS Logo ITS - Fall 2006
In This Issue QuickbITS Archives IT Directory Publishing Info
UH Post-Doc Analyzes Manoa Flood Using
Maui High Performance Computing Center
to previous article to next article
by Dr. Susan Brown

Dr. Guangxia Cao (r) and Prof. Duane Stevens, Department of Meteorology.

Dr. Guangxia (Jax) Cao has used the JAWS supercomputer at the Maui High Performance Computing Center (MHPCC) to model the weather conditions surrounding the flash flood that inundated Manoa Valley on Halloween eve 2004. JAWS is one of the 25 most powerful supercomputers in the world.

During the evening of October 30, 2004, a heavy thundershower on Oahu Island induced a flash flood in Manoa Valley causing severe property loss. The flood damaged cars, ripped through houses and buildings, and inundated several buildings on the UH Manoa campus, causing severe damage to Hamilton Library and the Biomedical Sciences Building.

These types of weather phenomena are difficult to predict using standard global weather models because the island of Oahu can fall within a single computational cell, which averages out all the local terrain effects. To overcome this limitation, researchers at UH Manoa, including Dr. Cao and his advisor, Professor Duane Stevens, perform meso-scale modeling using the Weather Research Forecast (WRF) Model to calculate specific local weather effects on the Hawaiian Islands. 

The total rainfall simulation after a period of 24 hours till 18 ZZ 31 October 2004. Unit: mm.

In this project, Dr. Cao modeled the Halloween eve flood conditions using WRF on JAWS, a 5,120-node Dell cluster at MHPCC. JAWS is currently one of the 25 most powerful supercomputers in the world. Dr. Cao compared different numerical techniques, looked at the effects of including soil moisture, and varied the initial and running conditions to find the best parameters to predict the events for the phenomena of that evening. The illustration to the right displays the calculated total rainfall and the localization in the Manoa area. The results of his research can be used as a basis to improve weather forecast models to predict future catastrophic events.

Dr. Cao's work was supported by a 2006-07 UH MHPCC engagement grant. These grants support UH researchers in their use of MHPCC resources. Dr. Cao shared his work at the major international conference on supercomputing, SC'07, in the MHPCC booth in Reno, Nevada, November 11-16, 2007

 

 
to previous article to next article


Information Technology Services
Maintained by: editor@hawaii.edu
© 2008 University of Hawai'i
Last Updated: January 2008