Western consortium's model smokes out fire behavior
- By Trudy Walsh
- Jul 23, 2004
Desert Research Institute analysts can now view their fire and smoke forecast models as 2-D maps.
When something 'goes up in smoke,' where does it go?
The Desert Research Institute, a Reno, Nev., group that's part of the University and Community College System of Nevada, is using modeling software to answer this and other questions about fire and smoke behavior.
The researchers want better forecasts of so-called prescribed burning for management objectives, such as eradicating an invasive species or improving the health of an ecosystem, said Tim Brown, an associate research professor.
'You don't want the smoke to wind up in a valley or a hospital or any place where people are especially sensitive to smoke,' Brown said.
The program serves a consortium of wildfire and air quality agencies called the California and Nevada Smoke and Air Committee. In addition to the institute, CANSAC members include the Forest Service, Bureau of Land Management, National Park Service, Fish and Wildlife Service, and California Forestry and Fire Protection Division.
The project kicked off June 1, Brown said, and 'at this point we might say it's in the evaluation stage.' The fire behavior models will be used by members of CANSAC, meteorologists trained in fire weather forecasting and National Weather Service analysts.
The institute's 450 faculty members, graduate students and researchers are using Mesoscale Model Version 5, or MM5, modeling software developed jointly by Pennsylvania State University and the National Center for Atmospheric Research. MM5 is written mostly in Fortran.
MM5 runs on a 64-bit Linux-based SGI Altix 3000 supercomputer with Intel Itanium processors and SGI's InfiniteStorage system.
The researchers will simulate mesoscale atmospheric circulation for the western United States and parts of southwestern Canada and northwestern Mexico. The Altix 3000 calculates conditions at 33 atmospheric levels, from the ground up to about 70,000 feet. The twice-daily runs take at least 7G of data, Brown said.
The researchers visualize the MM5 output on two-dimensional maps from a Fortran-based graphics package called Read/Interpolate/Plot, developed by NCAR and the University of Washington.
Trudy Walsh is a senior writer for GCN.