Table 8. Sensitivity Analysis of TBP Input Parameters, Averaged Over All PCDD/F Congener/Homologue Groups

ERDC TN-DOER-R5

September 2004

Sensitivity Analysis. Sensitivity indices calculated for the four TBP input parameters using

the available data indicated that Cs and lipid each contributed 21 percent to TBP uncertainty,

TOC contributed 26 percent, while BSAF contributed 32 percent (Table 8). BSAFs included in

these analyses ranged from a minimum of 0.0033 for 1,2,3,4,6,7,8-HpCDF, to a maximum of

7.45 for 2,3,7,8-TCDF. Mean BSAFs were in the range 0.02 to 1.28.

Table 8

Sensitivity Analysis of TBP Input Parameters, Averaged Over All PCDD/F Congener/Homologue

Groups

Parameter

TBP min

TBP mean

TBP max

Sensitivity Index

Percent Contribution to TBP Uncertainty

0.762

1.942

3.793

1.404

TOC

4.318

1.942

1.025

1.696

Lipid

0.440

1.942

3.172

1.407

BSAF

0.528

1.942

5.825

2.108

TBP calculated using minimum value of selected parameter and means of other parameters.

TBP calculated using means of all parameters.

TBP calculated using maximum value of selected parameter and means of other parameters.

Congener/homologue mean BSAFs.

CONCLUSIONS: This technical note illustrates two methods for estimating TBP uncertainty in

evaluations involving the potential for bioaccumulation of PCDD/Fs in benthic organisms. The

RSS method provides a large uncertainty interval that incorporates error due to the model

(method error) and measurement error associated with each model input (propagated error). The

bootstrap method constructs a statistical distribution for TBP that can then be used to determine

descriptive statistics (mean, confidence interval, etc.) and conduct tests of significance. The RSS

method produced an uncertainty interval with a range two to five times the magnitude of TBP,

while the bootstrap uncertainty interval range was about twice the magnitude of TBP. Both types

of uncertainty interval were similar in identifying the PCDD/F congeners or homologue groups

for which TBP underestimated observed bioaccumulation.

Using individual congener or homologue group mean BSAFs in calculating TBP was preferable

to using an overall mean BSAF from all PCDD/Fs. The overall BSAF resulted in TBPs that

significantly overestimated observed bioaccumulation of some PCDD/F congeners using the

bootstrap method, or in uncertainty intervals that were too broad to identify significant

differences between predicted and observed tissue concentrations using the RSS method.

Whereas the bootstrap method requires SAS programs or specialized software such as

Resampling Stats (www.resample.com), the RSS method can be implemented using a calculator

or spreadsheet software and the step-by-step procedure outlined above. TBP and its uncertainty

interval, generated using either of these methods, have been shown to provide reasonably

accurate estimates of actual PCDD/F congener bioaccumulation in sediment-exposed benthos,

when congener-specific BSAFs are used in the calculations.

TBP is implemented in Tier II of the four-tiered dredged material evaluation process

(USEPA/USACE 1991, 1998). TBP can be useful in identifying sediments for which expensive

Tier III or Tier IV evaluations are unnecessary, either because the sediment is acceptably clean

for unrestricted open-water disposal (as evidenced by the TBP upper uncertainty limit), or