Environmental Engineering Reference
In-Depth Information
14.8.2.3 Software
SSDs can be derived using any software package with which one can fit distri-
butions, and there are many. Examples of freely available software specifically
for SSDs are
ETX
(Van Vlaardingen et al.
2004
),
MS-Excel spreadsheets
for
Weighted SSDs (Duboudin et al.
2004
), an executable named
OMEGA
(Beek et al.
2002
), and web-based platforms like
www.risicotoolboxbodem.nl
and WEBFRAM-
SSD modeling (
http://defrarisk.cadmusweb.com/Generic1.aspx
). Both ETX and the
WEBFRAM SSD software run three Goodness-of-Fit tests (Anderson-Darling,
Cramer-Von Mises, and Kolmogorov-Smirnov), which address the first issue by
focusing on the over-all fit for the whole range of input data.
14.9 Statistical Issues in SSD Modeling and Interpretation
So far, we have introduced the concept of distribution-based modeling, two con-
ceptual ways to use SSDs in Ecological Risk Assessment, software tools, and
validation. In practice, SSD-modeling also requires a framework within which
results are collected and interpreted. In this Section, we address statistical issues
in SSD modeling and interpretation. The next Section discusses other issues, for
example related to exposure conditions and to handling mixtures.
14.9.1 Minimum Data Numbers and (Mis)Fit
Many authors have discussed the minimum amount of data needed to do SSD-
modeling. Some agencies have adopted specific minimum numbers of species while
others derive SSD-output based on cases with few input data.
Some regulatory frameworks formally require a minimum data set for standard
setting. An example of the
Guidance
on the use of SSDs for deriving standards
for the EU is as follows: “
Confidence can be associated with a PNEC
[i.e., the
Predicted No Effect Concentration, like an HC5]
derived by statistical extrapolation
if the database contains at least 10 NOECs (preferably more than 15) for different
species covering at least eight taxonomic groups. Deviations from these recommen-
dations can be made, on a case-by-case basis, through consideration of sensitive
endpoints, sensitive species, mode of toxic action and/or knowledge from structure-
activity considerations
” (European Communities
2003
). In the U.S.A., data from
eight different genera of fish and invertebrates are required for setting National
Water Quality Standards (Stephan et al.
1985
).
We do not discuss the justification for these chosen values, because they are part
of specific regulatory soil protection contexts, and they are for the derivation of
generic legal standards (Y
→
X), not necessarily for the more specific use of SSDs
(X
Y).
As an alternative set of criteria for cases when there is not (yet) a regulatory
fixation of methods, we propose a two-way approach to help in deciding whether
an SSD-model is appropriate for a problem definition. First, in a statistics-based
→
Search WWH ::
Custom Search