Estimation of exposure point concentrations

functional forms, and incorrect boundaries. There are no straightforward

statistical or mathematical techniques for describing this source of uncertainty as

with other forms of uncertainty. The most efficient and appropriate method of

assessing model error is to obtain field verification or conduct an experiment to

validate the model. Table 1 describes the magnitude of uncertainty for these

models as MS because the amount of uncertainty will vary from model to model.

Alternatively, the use of several different models using the same data is a

suitable cross-validation method. Although this does not represent a quantifiable

assessment of uncertainty, the fact that several models all lead to the same

conclusion provides confidence that the models are performing correctly.

Differences in model results can highlight mechanistic or parameter-based

differences in each of the models, which may allow a determination of the

appropriateness of one model over another. Bayesian approaches to comparing

models also exist (Suter 1993).

Fate and transport models: Parameter uncertainty. The fate and transport

models that are typically used in the dredged material management program

obviously contain numerous parameters. Although it is beyond the scope of this

report to examine the uncertainty associated with all of these parameters, the

contribution of a few of the important parameters that are used in many of the

models is considered in the following sections.

Physical-chemical attributes of COCs. Physical-chemical attributes of

COCs are important parameters that are used in many fate and transport models

to predict exposure point concentrations. These parameters, which help predict

partitioning among media, often drive the predicted distribution of a contaminant

in the environment.

Uncertainty in these parameters contributes to uncertainty in resultant

exposure and risk assessments. Typically, physical and chemical property

estimation depends on laboratory conditions (e.g., temperature, pressure, etc.).

Consequently, a significant source of the uncertainty in using physical and

chemical parameters in fate and transport models involves the applicability of a

value for a given situation. For example, if a value is temperature dependent and

there is a large disparity between the laboratory temperature under which the

value was estimated and the expected field temperature, then the results of the

model are likely to be wrong, reflected as uncertainty in the result. However,

simplifying assumptions are inevitable and may be justified.

The contributions of physical and chemical properties to uncertainty in fate

and transport models vary. For example, properties such as molecular weight,

vapor pressure, and water solubility are straightforward to obtain and have less

mathematical effect on the results of fate and transport models than properties

such as Kow and Koc. Therefore, the former properties are ranked low for

magnitude of uncertainty with only moderate difficulty in reducing this

uncertainty. The latter properties are ranked moderate in magnitude of

uncertainty given that published values for particular chemicals can span one to

two orders of magnitude (Mackay, Shiu, and Ma 1992a, 1992b).

Chapter 5 Uncertainty in Tier IV Risk Assessments