|Tom Warner. (Photo by Carlye Calvin.)|
Over the past decade, more than 10% of the 700,000 U.S. veterans of the 1991 Gulf War have reported a variety of muscular and neurological ailments they feel might be a result of exposure to toxins and/or vaccines during their service. After the war, the Pentagon acknowledged that troops might have been exposed to tiny amounts of sarin, a potent nerve gas. New ensemble modeling in RAP suggests that if, in fact, Iraqi toxins were released during and after the war, the gas could have dispersed in a variety of possible directions and concentrations.
Tom Warner and Rong-Shyang Sheu (both of RAP) worked with a variety of experts from academia, business, and the military on the simulations, which were funded by the Department of Defense's Office of the Special Assistant for Gulf War Illnesses. This study was a follow-on to work carried out during the last few years by RAP at the office's request. Those first simulations covered 6 days each, but a meteorological panel that included UCAR president Rick Anthes recommended more in-depth study, with a variety of models and techniques, to get at the uncertainty aspect of the dispersion. The result was RAP's 60-day reanalysis, spanning the Gulf War as well as the period of oil-well fires that followed.
To better quantify the uncertanties that cropped up in the model solutions, Tom and his colleagues zeroed in on Al Muthanna, a site in central Iraq where toxins may have been released on 7 February 1991. Using MMM and SCD computers, they linked the Penn State/NCAR Mesoscale Model, version 5 (MM5), to the Second-order Closure Integrated Puff (SCIPUFF) model. SCIPUFF tracks the dispersion of an overlapping set of three-dimensional "puffs," such as chemical clouds.
Lacking hard data on any toxins released, the modelers simulated a kilogram (2.2 pounds) of inert gas released a meter (3.3 feet) above ground in a cube 5 m (16.4 feet) square. Weather data were also quite sparse, so the team compared Al Muthanna to a nearby point, Hafar Al-Batin, with surface and radiosonde reports and a similar climate. Around these two points, the MM5 resolution narrowed to 3.3 kilometers (2 miles). The spread of the gas was modeled in 12 ensemble runs. Each one used a different blend of model ingredients to simulate the same event, drawing one each from three large-scale analyses, three boundary-layer parameterizations, and two surface-physics schemes.
|As can be seen from the panel above, the 12 members of the ensemble model gave very different predictions for plume dispersal. Each box shows the predicted plume trajectory over time.|
The rationale for using ensemblesone of the first times they've been used to attain this level of mesoscale detailwas to create a set of probabilities of exposure at various distances from Al Muthanna. The data and models agree that weak northwest winds prevailed. The winds shifted to easterly in the southern part of the model area by 24 hours, but this shift varied in its onset and extent among the ensemble members. There were other striking differences:
Aside from the insight it provides on the Gulf War situation, the modeling exercise breaks new ground in the use of mesoscale ensembles for reanalysis and for coupling atmospheric and dispersion models. The model showed "fairly complex transport and dispersal processes, even though we tend to think of deserts as fairly uniform and boring," says Tom. He adds, "We learned a lot about boundary-layer variability for the entire desert." The findings have been presented at an atmospheric transport and dispersion conference at George Mason University; one journal article has been published and another one is currently in review.
Edited by Bob Henson,
Prepared for the Web by Jacque Marshall
Last revised: Thu Apr 19 15:33:03 MDT 2001