The updated site-specific risk assessment (uSSRA) of the proposed National Bio- and Agro-Defense Facility (NBAF) has modified the dispersion modeling and analysis of the 2010 SSRA. It has responded to some of the previous criticisms by conducting sensitivity studies with the Second-order Closure Integrated Puff Model (SCIPUFF) to examine the effects of uncertainty in meteorological variables and model parameters on predicted doses; accounting for livestock distribution in computing the integrated dose (the dose is related to the risk of infection at any specified location relative to the release location); and using three different methods to estimate the risk of infection resulting from a dose of foot-and-mouth disease virus (FMDv).
The uSSRA uses SCIPUFF to estimate the exposure of a potential FMDv release from the NBAF and uses the North American Animal Disease Spread Model (NAADSM) to model airborne spread of FMDv once an infection is initiated. SCIPUFF is a Lagrangian air dispersion model that uses Gaussian puffs to model 3-D time-dependent dispersion of concentrations; it is available both in commercial and Environmental Protection Agency no-cost versions.
SCIPUFF is appropriate for modeling airborne transport and dispersion of potential releases from the NBAF, but the uSSRA’s description of its application and the associated results are difficult to follow even for experts in
transport modeling. The uSSRA used a meteorological classification scheme referred to as self-organizing maps (Kohonen, 1982) to generate meteorological inputs required to run SCIPUFF. Because this approach is generally not used in air pollution modeling, it is difficult to evaluate the validity of the claim that 95% of the meteorological conditions result in no infection. The graphics used to present the results convey little information. With such insufficient information, the committee found it impossible to evaluate the validity of the results.
The NAADSM is used “to simulate the spread and control of foreign animal diseases in a population of susceptible livestock herds” (www.naadsm.org). The NAADSM analysis includes a pathway to model the spread of infection through airborne transport of virus particles. The airborne spread model has two options to describe the probability of airborne spread of infection between two premises. The first option assumes that the probability of infection at a premise declines linearly with distance from the source of infection; this option was used in the 2010 SSRA. The second option assumes exponential decline with distance and is used in the uSSRA (p. 440); exponential decline is further explained in the NAADSM user’s guide. The linear model typically leads to lower probabilities of spread over shorter distances and higher probabilities over longer distances.
The uSSRA states that the adoption of the exponential option was based on FMDv dispersion modeling results as illustrated in Figure 6.1.4-19 (p. 442). However, the uSSRA does not show how these results are derived from the results presented in the cited references (Garner and Cannon, 1995; Sørensen et al., 2000) or those from SCIPUFF as described in Volume I of the uSSRA. Furthermore, it is unclear how these results were used to specify the parameters of the airborne spread equation on p. 67 of the NAADSM user’s guide.
Figure 6.1.4-19 of the uSSRA indicates that the uptake of plaque-forming units by cattle falls off by an order of magnitude when the distance increases by a factor of 2.5 from 2 to 5 km. That result is inconsistent with the statement made in the uSSRA that according to Garner and Cannon (1995) the risk of infection is expected to fall off linearly with distance under stable atmospheric conditions. It might be more appropriate to assume that the risk of infection is inversely proportional to the distance from the source because the risk of FMDv exposure is high when the atmospheric boundary layer is stable (Garner and Cannon, 1995). Under these conditions, the shallow boundary layer limits vertical dispersion, and the growth
of horizontal plume spread is at most linear with distance from the source (Venkatram and Wyngaard, 1988).
The uncertainty in the NAADSM airborne spread model suggests the need to conduct sensitivity analyses to examine the effects of both model formulation and parameter values on FMDv spread and hence on the economic impact of FMDv release from the NBAF. A sensitivity analysis should have been included in the uSSRA, as it would also indicate the role of airborne spread relative to other modes of spread.
Garner, M.G., and R.M. Cannon. 1995. Potential for windborne spread of foot-and-mouth disease virus in Australia. A report prepared for the Australian Meat Research Corporation, Bureau of Resource Sciences: Canberra, Australia. 88 pp.
Kohonen, T. 1982. Self-organized formation of topologically correct feature maps. Biol Cybern 43:59-69.
Sørensen, J.H., D.K.J. Mackay, C.Ø. Jensen, and A.I. Donaldson. 2000. An integrated model to predict the atmospheric spread of foot-and-mouth disease virus. Epidemiol Infect 124(3):577-590.
Venkatram, A., and J. Wyngaard (Editors). 1988. Lectures on Air Pollution Modeling. Boston: American Meteorological Society. 390 pp.
This page intentionally left blank.