18018_Singer Furniture_Background Soil Sampling Requirements_19990907

the data so that 0.75 of the data are greater than Q 25 and 0.25 of the data are less than
or equal to o .25; and
Interquartile range (JQR) is the difference between the upper quartile and the lower
quartile (i.e., 1QR = 0 .75 -0 .25).
Data shall be normalized by use of an appropriate transformation, if necessary (See
discussion below for Goodness-of-Fit or Normality Tests), prior to calculation of the
upper and lower cutoffs.
Data not falling between the upper and lower cutoffs shall be reviewed to determine
whether evidence exists to suggest the data are not representative of the background
population.
The owner/operator should check such data for sampling and laboratory errors, field
evidence of waste materials at the sampling locations, and other plausible causes.
Where sufficient evidence indicates that a background sample is not representative of
background conditions, the owner/operator shall discard the datum and obtain a
substitute sample. If no specific error can be documented, the sample must be
retained in the data.
In addition, background samples must be eliminated and replaced with a like number of
samples from uncontaminated areas if: 1) the background samples are taken in areas
known or suspected to be contaminated and whose areal extent of contamination does
not included the closure area, or 2) the background samples have possibly been
affected by RCRA activities conducted in the area undergoing closure. Areas to avoid
for background sampling include but are not limited to:
1) past waste management areas where solid and/or hazardous wastes or
wastewaters may have been placed on the ground, areas of concentrated air
pollutant deposition (from a definable localized source), or areas affected by the
runoff;
2) roads, road side, parking lots, areas surrounding parking lots or other paved
areas, railroad tracks or railway areas or other areas affected by their runoff;
3) storm drains or ditches presently or historically receiving industrial or urban
runoff;
4) spill areas:
5) material handling areas, such as truck or rail car loading areas, or near
pipelines;
6) fill areas;
r
I -
.. '
7) other areas as determined by the HWS.
The owner/operator must use the following statistical procedure to calculate an Action
Level or Upper Confidence Limit (UCL) for the determination of a remediation standard
for each hazardous constituent present in the background soil:
Action Level or UCL "" mean of the background population + two times the standard
deviation of the background population. Soil contamination In the RCRA unit(s)
exceeding the UCL must be remediated.
The above statistical test assumes that the two populations being compared
(background and RCRA unit soil data) are normally distributed. The owner/operator
must first demonstrate, through probability plots and the Shapiro-Wilkes test (Conover.
1980) (or similar test), that both populations are normally distributed before calculating
the UCL.
If it is discovered that the populations are not normally distributed, the owner/operator
shall search for a transformation that makes the populations approximately normal.
The same transformation must be applied to both the background data as well as the
data collected from the RCRA unit(s) In question. If the data cannot be transformed to
normality, the owner/operator must develop an alternative statistical procedure(s) to
calculate an action level, and submit it to HWS for review and approval.
If any hazardous constituent, identified in the waste and included in the list of
constituents submitted by the owner/operator and approved by HWS, is found to be
non-detectable (i.e., below the practical quantitation limit or POL) in the background
soil. then the owner/operator shall use the PQL for the individual constituent as the
clean standard.
Also, the HWS ma~ accept other statistical methods if the owner/operator can
demonstrate that the statistical method proposed is environmentally acceptable and is
technically superior. in order for the HWS to evaluate an alternative statistical method
adequately, the owner/operator must provide considerable background discussion of
the proposed method and its applications, literature references, sample calculations,
and any other information deemed appropriate by the reviewer(s).
For further information on statistical procedures, consult:
1) USEPA, (1986): "Test Methods for Evaluating Solid Waste, Physical/Chemical
Methods, Third Edition."
2) Bickel, P. J., Doksurn, K. A., (1977): Mathematical Statistics: Basic Ideas and
Selected Topics, John Wiley & Sons Inc., New York.
3) Conover. W.J., (1980): Practical Non-parametric Statistics, John Wiley & Sons
Inc., New York.
:""
f
I '
I
I 4) Devore, J., Peck, R., (1986): Statistics: the Exploration and Analysis of Data,
West Publishing Company, St. Paul, Minnesota.
5) Gilbert, K.O., (1987): Statistical Methods for Environmental Pollution Monitoring,
Van Nostran Reinhold, New York.
6) Hoaglin, D.C., Mosteller, F., Tukey, J.W. (1983): Understanding Robust and
Exploratory Data Analysis, Wiley, New York
7) ASTM E-178, Standard Practice for Dealing with Outlying Observations
Situations will exist where the surrounding area or matrix (i.e. ground water, air, soil)
has historically been affected by sources outside of the site under investigation.
Examples of such situations include: acid mine drainage areas, surface waterways with
point/non-point discharge sources, non-potable aquifers (high mineral content, saline)
or where the aquifer is affected by off-site sources (upgradient or up slope
contamination). Specific guidelines cannot be outlined for every site; therefore,
evaluations must be made on a site-by-site basis .
. ·.
---.