Imagine an accidental release of radiation from a nuclear power plant in the United States. People would have to be quickly evacuated from areas through which harmful concentrations of radiation might pass. Could we predict where the plume might go, how fast, and at what concentrations? We might know the basic weather conditions in terms of wind direction and speed, but we might not know much more than that. Predictions might easily be made for a relatively flat terrain, but what about a complex terrain such as the Front Range of the Rocky Mountains where the Rocky Flats nuclear weapons plant is located? What about in the valley and ridge country around Oak Ridge, Tennessee, with its concentration of nuclear facilities?
Peaks and valleys. A simulation using the regional atmospheric modeling system shows wind vectors in the Eldorado Canyon (EC), Coal Creek Canyon (CC), Ralston Creek Canyon (RC), and Rocky Flats (RF).
With the death toll from the 1986 Chernobyl accident in the Ukraine now estimated at 125,000 people, there is no question about the importance of understanding how radiation and other hazardous substances might act if released into the atmosphere. Emergency preparedness, and the more general concerns of air quality and global climate change, are the focus of the Atmospheric Studies in Complex Terrain (ASCOT) program developed by the Office of Health and Environmental Research in the U.S. Department of Energy.
ASCOT evolved in the late 1970s from a combination of air quality and meteorological programs being conducted by the DOE in conjunction with the planned exploitation of geothermal and oil shale deposits in mountainous regions of the American West. The DOE has long recognized the need for an understanding of the meteorological processes at work in the planetary boundary layer (PBL)--the depth of atmosphere just above the surface of the earth. The transport and diffusion of atmospheric contaminants are critically dependent on the characteristics of the PBL, which in turn are strongly dependent on topographical factors.
Historically, ASCOT research has been closely tied to DOE activities. Other agencies, such as the EPA, also need to understand how air flow and temperature fields exacerbate air pollution in regions of complex terrain, many of which are classified as nonattainment areas under the Clean Air Act. Organizations studying global climate change have also recognized that they must improve their methods of representing regional phenomena such as wind, temperature, cloud cover, and energy exchange patterns, which may be strongly influenced by terrain. ASCOT has the potential to address all of these issues.
The two principal goals of ASCOT are to characterize, understand, and predict boundary layer structure and evolution over complex terrain and to develop methods for applying this knowledge to DOE tasks that involve site safety, air quality, and climate change. Since its inception, ASCOT has included a mix of measurement, analysis, and modeling activities. Half a dozen DOE laboratories, as well as laboratories at the National Oceanic and Atmospheric Administration (NOAA) and various universities, have cooperated in field experiments that have provided the data for subsequent analyses and modeling studies.
"The level of cooperation among the various institutions involved in ASCOT has been outstanding," says Chris Doran, scientific director of ASCOT based at Pacific Northwest Laboratories in Richland, Washington. "We have developed a high level of expertise in this field, which could be applied to a variety of related areas of study."
Funding for ASCOT was cut 50% in fiscal year 1995, and Doran says future funding prospects are unclear. Not all of the goals and objectives set for ASCOT have been achieved. However, significant progress has been made in analyzing the behavior of the PBL over complex terrain.
Recent Field Experiments
In keeping with its goal of supporting the DOE's mission, ASCOT's recent field experiments have focused on the department's Oak Ridge complex in the Tennessee Valley and the Rocky Flats Plant along the Front Range of the Rocky Mountains near Denver, Colorado. Both the Tennessee Valley and the Front Range areas exhibit a complex mix of local and regional air circulation patterns that are strongly affected by terrain and thermal forcing, which complicates the development of reliable transport and diffusion models.
The Tennessee Valley is a broad area between the Cumberland Mountains and the Great Smoky Mountains. Field tests have shown there is a multilayered wind flow in the Tennessee Valley. Winds flowing across the mountains from the northwest can result in winds within the valley being driven at right angles (up-valley) by the large-scale pressure gradient along the valley axis. The valley floor is corrugated by parallel ridges 100-150 m high. At the corrugation level, winds may travel in yet other directions, driven by energy-exchange processes, such as the heating and cooling of valley walls.
The Front Range of Colorado features mountains rising 1500-2000 m above the adjacent plains. Here, the diurnal cycle of heat fluxes is typically larger than in the Tennessee Valley. The complexity of the terrain results in a host of locally generated air flows in valleys and over the slopes and plains.
"One complicating factor is that many scales of motion may contribute to a particular situation, and there is usually no absolute separation of one scale from another," says Doran. "Local flows are rarely completely decoupled from larger- scale flows. Nevertheless, it is often possible to find conditions in which thermal forcing dominates."
Analyses and Modeling
For the purpose of measuring, analyzing, and modeling factors affecting the PBL in these two geographical areas, ASCOT teams have installed a variety of state-of-the-art measuring devices, in addition to using information provided by the National Weather Service. Data have been obtained through a combination of surface stations, aircraft flights, tethered and free-flying instrumented balloons, sound detection and ranging, and tracer sampling. Through cooperation with NOAA's Environmental Technology Laboratory, ASCOT researchers obtained additional information from infrared light detection and ranging and from radar wind profilers and radio acoustic sounding systems.
Something in the air. Multiple scales of air motion in the Front Range of Colorado make it difficult to predict where pollutants might flow.
TYpically, intensive measurements were made for periods of two to four weeks. A network of continuously operating instruments also provided long-term data from these areas. The intensive measurements were used to establish a climatology or catalog of flow patterns. Researchers then characterized the general features of the PBL structure and the local circulation.
Detailed analyses ranged from the application of simple conceptual models to the use of sophisticated three-dimensional, full-physics numerical models. Several models used in ASCOT were developed or modified specifically for the program. Others were already in existence.
One benefit of ASCOT is that existing models have been validated. "One of the remarkable aspects of this program has been the number of people who have gone to great pains to test these models and be honest about how they performed," says Doran.
The less sophisticated models used in ASCOT rely on simple data, typically wind speed and direction, taken from observation sites within the test area. A release of a gas is then simulated and the model is used to predict where the gas is transported, when it arrives, and in what concentrations. The simplest models do not attempt to recreate the many interactive forces at work in the PBL. Initial meteorological conditions may be assumed to persist.
"With these simple models, you are relying on the representativeness of your observations to determine the wind field," says Robert Addis, an ASCOT modeler with the Savannah River Technological Center. "If your observations are not at the right height or are not detailed enough, you may get inaccurate readings of phenomena such as drainage jets that may be occurring within the area. Such a model cannot interpret where the wind field is not being measured. Further, if your initial conditions change, the accuracy of the predictions will suffer. You can't necessarily assume that the wind speeds you put in initially will be applicable two or three hours later."
More advanced models take into account a wide variety of additional factors such as air temperature, humidity, cloud cover, time of day, turbulence, soil moisture, and underlying canopy. Using advanced physics, the models solve equations of motion, state, and energy for the atmosphere over the study area. Advanced models should outperform simple diagnostic models in a real world situation.
"If you have adequate physics in the model, you have a better chance of at least capturing the main physics of the flow in an area of complex terrain," says Addis. "This is important for emergency preparedness planning. In an emergency situation, you don't usually have a large number of observations already in hand. Very likely, you only have a limited amount of data on meteorological conditions. Theoretically, a predictive model with good physics has a better chance of accurately characterizing a wind field using minimal observations than does a diagnostic model. The question is, does the more advanced model predict the future any better?"
Results
Field experiments releasing tracer gas (sulfur hexafluoride or SF6) were performed at the Front Range in 1991 to test how well the models work. Sulfur hexafluoride mimics the behavior of a nonbuoyant (i.e., not pushed aloft by a flame), nondepositing (i.e., not containing particles that drop out of the plume) pollutant such as might occur with the release of radioactivity or toxic gas from a nuclear power plant or chemical or nuclear weapons storage or manufacturing facility. However, the models can be used to imitate the behavior of any gaseous pollutant, according to J.T. Lee, meteorology team leader for ASCOT programs at Los Alamos National Laboratory.
The accuracy of results from the models tested varied widely. Although not all predictive models performed well, scientists found the best of the predictive models performed better than the diagnostic ones under actual conditions.
"For some of the tests, the best models were able to predict concentrations of pollutants within a factor of 2 out to 16 kilometers from the point of release," Lee says. "That's good enough to be applicable for emergency response situations."
The best models were those that use four-dimensional data assimilation, which allows modelers to take observations as they come in and "nudge" the solution toward what is actually occurring in the atmosphere. Lee says this method can be helpful in directing researchers where to locate the observation network. It also lets scientists know the degree of confidence they should put in the models--an important factor for emergency preparedness.
Lee says the models developed under ASCOT can be applied to sites in other areas of complex terrain. HOTMAC (Higher Order Turbulence Model for Atmospheric Circulation), a model developed at Los Alamos National Laboratory, is currently being used by the U.S. Army for emergency preparedness planning in Utah. The RAMS (Regional Atmospheric Modeling System) developed by Colorado State University is being used at the Savannah River nuclear plant in Georgia.
Lee cautions that the models cannot be used for an emergency response without local meteorological data to feed into the model. The more data that can be obtained, the better the model will perform. Depending on the complexity of the terrain, emergency preparedness planners may be advised to erect their own data-gathering instrumentation around a site to supplement information available from local weather services.
Of course, where and how far a release of radioactive or toxic gas will travel from Oak Ridge, Rocky Flats, or any other site is highly dependent on the meteorological conditions at that time. Planned evacuation routes may not be changed as a result of having the ASCOT models. However, the models give emergency preparedness planners another tool to predict where a plume might travel over time and to warn people in the affected areas to take appropriate action.
"In sum, we have found that predicting the movement of gaseous plumes in complex terrain is difficult, but not impossible," says Addis. "What would be a good study is to remove half or three quarters of the observations we used in our tracer experiments and see how well the models do. This would be more akin to what happens in an emergency situation. You will have to respond quickly based on a very limited amount of data."
Cutbacks in funding are preventing ASCOT teams from conducting additional studies to assess and improve the accuracy of their models. What is more worrisome, some scientists say, is that the public and private sectors appear reluctant to use the more successful models developed under ASCOT. In light of what happened at Chernobyl, this seems particularly foolhardy.
"We have some truly fantastic models, but emergency managers with DOE and the private sector are abandoning them in favor of simplicity," says Addis. "America is going backwards in emergency management and planning to the straight line Gaussian models used in the 1950s--models that do not take into account the effects of complex terrain or changing meteorological conditions. The Europeans are going forward with the deployment of sophisticated models. As a result of Chernobyl, they understand what can happen when the best science is ignored."
John Manuel
American Meterological Society. Atmospheric studies in complex terrain: part l. J Appl Meterol 28(6) 1989.
American Meterological Society. Atmospheric studies in complex terrain: part ll. J Appl Meterol 28(7) 1989.
Blumen W. Atmospheric processes over complex terrain. Boston, MA: American Meterological Society, 1990.
DOE. Multi-year strategic plan for the atmospheric studies in complex terrain. DOE/ER-0549T. Washington, DC: United States Department of Energy, 1992.
Last Update: May 16, 1997