13.01.2013 Views

Dames & Moore, 1999 - USDA Forest Service

Dames & Moore, 1999 - USDA Forest Service

Dames & Moore, 1999 - USDA Forest Service

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

7.2.5 SOURCES OF UNCERTAINTY<br />

Limitations associated with any risk assessment have a number of components. including degree of<br />

success in meeting objectives, the range of conditions over which conclusions can be applied, and the<br />

certainty with which conclusions can be drawn (EPA 1989a). The conclusions of a risk assessment are<br />

useful only once they have been placed in perspective relative to the uncertainties associated with the<br />

evaluation.<br />

7.2.5.1 General Components of Uncertainty<br />

Toxicology and risk assessment are relatively new sciences, and, as with all risk assessments. there are a<br />

variety of uncertainties in this baseline ERA. The ERA process relies on assumptions that have varying<br />

degrees of accuracy and validity. Uncertainty in risk estimation has both qualitative and quantitative<br />

components. In general, the uncertainty surrounding a risk estimate consists of:<br />

real variation, reflecting actual ranges in biological responses<br />

lack of certainty, regarding basic physical, chemical, and biological properties and<br />

processes (e.g., bioavailability, bioaccumulation)<br />

assumptions in the models used to approximate key input values (e.g., dose-response an<br />

exposure "models" used for criteria derivation)<br />

measurement error<br />

It is important to understand that uncertainty includes both real variation (reflecting actual, mechanistic<br />

biological response ranges and variability in ecosystem conditions) and error. Thus, because biological<br />

systems are inherently uncertain and variable, some component of variability in risk estimation is due to a<br />

realistic reflection of ecological conditions, while another component is due to "error" or uncertainty<br />

introduced by the overall analytical process. However, it is critically important to understand that natural<br />

ecosystem variability represents an important source of uncertainty to the evaluation of the ecosystem.<br />

While this parameter is paramount to the making of risk management decisions, it cannot be controlled.<br />

Therefore, "error" is the component to be minimized, because this encompasses undesirable uncertainty<br />

that has been introduced by the assessment process and can. to some degree, be controlled.<br />

Database<br />

The analytical database has inherent uncertainties. For example, sampling was generally concentrated in<br />

areas of known elevated concentration so that estimates of the actual underlying distribution of PCOCs<br />

made from the dataset may be biased conservatively. In addition, when analytes were not detected in<br />

environmental media, these data points were not included in the statistical analysis. Although this<br />

conclusion is based upon the available evidence, it is possible that metals could have been present at<br />

concentrations below the method detection limit. This was not felt to impact the conclusions reached<br />

because the detection limits were always lower than the concentrations known to cause adverse effects.<br />

Nevertheless, this method introduces uncertainty which cannot be quantified. An attempt to limit this<br />

source of uncertainty was made by using the median concentrations rather than the means. Medians are an<br />

expression of central tendency which is less biased by either high or low values and is more appropriate for<br />

animals which may or may not contact a specific area where data were collected.<br />

G:\wpdurW~~\boIdm-2\ni74.doc 7-7 1<br />

17693405419Uuly 27.<strong>1999</strong>;5:16 PMDRAFT FINAL RI REPORT

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!