United States
Environmental Protection
Agency
Office of
Research and
Development
Office of Solid Waste
and Emergency
Response
EPA/600/5-04/054
May 2004
&EPA Technical Support Center Issue
Fingerprint Analysis of Contaminant Data:
A Forensic Tool for Evaluating Environmental
Contamination
Russell H. Plumb, Jr.*
1.0 Introduction
Several studies have been conducted on behalf of the U.S. Environmental Protection Agency (EPA)
to identify detection monitoring parameters for specific industries.1'2'3'4'5 One outcome of these
studies was the evolution of an empirical multi-variant contaminant fingerprinting process. This
process, Fingerprint Analysis of Leachate Contaminants (FALCON), was developed through the
EPA's Technical Support Center (TSC) in response to the need for identifying the source of
contaminant plumes. FALCON combines data for several contaminants to develop a distinctive
graphical fingerprint or multi-parameter chemical signature. These fingerprint patterns can be used
to characterize the source of a contaminant plume, differentiate the contaminant plume from
background conditions at the source, and monitor the migration of leachate into the environment. It
can be applied to both organic and inorganic contaminants and is effective over a wide range of
contaminant concentrations. This data evaluation process is analogous to using fingerprints to
identify individuals. However, rather than using the size and location of ridges and swirls on the
fingertip, the relative abundance of selected constituents is used to develop distinctive chemical
signatures.
The objective of this paper is to demonstrate that FALCON is a quantitative, defensible fingerprinting
process. A description of the stepwise FALCON technique is provided in Section 2.0. Examples are
presented to illustrate the range of situations in which fingerprinting can be applied to characterize
the occurrence and distribution of environmental contaminants. These examples were developed
using routine monitoring data obtained from a variety of ongoing site characterization and monitoring
programs. Case studies of FALCON applications are presented in Section 3.0.
2.0 FALCON Procedure
The FALCON procedure is a multi-step process of combining data from two or more measurements
to create a distinctive ratio or multi-parameter chemical signature that can be used to characterize a
contaminant plume from a particular source. The following example, using data from a dioxin-furan
contaminated sawmill site, illustrates the individual steps in this fingerprinting process.
Lockheed Martin Environmental Services
1050 East Flamingo Road, Suite E120
Las Vegas, NV89119
Technical Support Center for Monitoring
and Site Characterization
National Exposure Research Laboratory
Environmental Sciences Division
Las Vegas, NV 89193-3478
Office of Superfund Remediation and
Technology Innovation
Office of Solid Waste and Emergency Response
U.S. EPA
Washington, D.C.
121CMB04.RPT * Rev. 8/17/04
-------
Step 1: Data Review
Data are subjected to an initial review to assess the type and quantity of information available. The data are
examined to identify subsets that are representative of the designated source area, background areas, impacted
areas, and other potential sources that may be differentiated by fingerprinting. The data are then examined
to determine whether there are replicate data to assess the reproducibility of any fingerprint pattern that may
be identified. This information may exist as either duplicate analyses, multiple samples from the same area,
or samples from the same location over time. The data are also reviewed to ensure each individual set is
complete and comparable (there are no missing values, results are expressed in the same concentration units,
non-detect limits are specified). Finally, the data are examined for any obvious data quality issues.
Step 2. Data Tabulation
The individual data sets are entered into a spreadsheet to (1) select parameters to be used in the fingerprinting
process, (2) prepare graphical plots of the fingerprint patterns, and (3) estimate the statistical
reproducibility/comparability of identified fingerprint patterns. In this particular example, the laboratory had
reported results for 17 specific dioxin-furan congeners and several "total" congener measurements (e.g., total
hexachlorodibenzodioxin). The "total" measurements were excluded from the fingerprint process because
the results were vague (specific congeners included in the total were not identified and could not be confirmed
with the congener-specific data). Also, Octachlorodibenzodioxin was excluded from the fingerprint because
the extremely high concentrations for this specific congener overwhelmed the concentrations for the
remaining dioxin-furan congeners. After data review and tabulation, 16 dioxin-furan congeners were selected
to develop the sawmill fingerprint. The data from seven soil samples at this site are presented in Table 1.
These samples, with a calculated total dioxin-furan congener concentration ranging from 231 nanograms per
kilogram (ng/kg) to 1,302,460 ng/kg, were specifically selected to demonstrate the capability of this
fingerprinting technique to handle the highly variable contaminant concentrations that can be encountered
in a site investigation.
Step 3. Data Normalization
Data used for contaminant fingerprinting are transformed in a multi-step process. First, individual results
listed as "not detected" in the original data set are replaced with a numerical value. The convention used in
this example is to replace the "ND" result with a value equal to one-half the reported detection limit. Second,
a total concentration is calculated for the parameters used in the fingerprint (fingerprint mass). Finally, the
reported concentration for each fingerprint constituent in that sample is normalized to the calculated
fingerprint mass. Thus, the reported concentrations for the fingerprint constituents are converted to a decimal
percentage of the calculated fingerprint mass. The transformed dioxin-furan data for this example are shown
in Table 2. The substitution of a numeric value for non-detected constituents and the subsequent data
transformation process performs two functions; it permits individual data sets to be plotted on a common y-
axis scale for visual inspection of the fingerprint (Step 4) and it permits individual data sets to be statistically
compared (Step 5).
-------
Table 1. Dioxin-Furan Monitoring Results for Sawmill Soil Samples*
Dioxin-Furan
Congener
2378-TCDD
12378-PeCDD
123478-HxCDD
123678-HxCDD
123789-HxCDD
1234678-HxDD
2378-TCDF
12378-PeCDF
23478-PeCDF
123478-HxCDF
123678-HxCDF
123789-HxCDF
234678-HxCDF
1234678-HpCDF
1234789-HpCDF
OCDF
Sum 1-16
Chemical
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Soil Sampling Locations
SM20
0.95
2.4
2.4
5.4
4.8
160
0.95
2.4
2.4
2.4
2.4
2.4
2.4
11
2.4
26
230.7
SM34
0.95
2.35
3.1
13
11
300
0.3
2.35
2.35
2.35
2.35
2.35
2.35
30
2.35
69
446.1
SM41
180
520
1700
4900
1400
100000
13
13
16
30
150
2.35
85
23000
2300
43000
177309
SM44
1.4
9.8
12
89
35
1100
3.4
2.6
2.3
12
11
2.4
8.8
210
14
450
1964
SM40
170
1800
4000
47000
12000
760000
190
280
540
3300
1600
280
1500
110000
9800
350000
1302460
SM46
16
130
290
2500
630
34000
140
110
120
240
200
9
180
3300
290
8400
50555
SM51
8.4
31
48
350
100
3900
21
6.1
5.5
45
24
7.5
17
600
62
1300
6525.5
* Dioxin-Furan concentrations reported in ng/kg.
ND values have been replaced with a concentration equal to one-half the ND value (in bold).
Table 2. Normalized Dioxin-Furan Monitoring Results for Sawmill Soil Samples
Dioxin-Furan
Congener
2378-TCDD
12378-PeCDD
123478-HxCDD
123678-HxCDD
123789-HxCDD
1234678-HxDD
2378-TCDF
12378-PeCDF
23478-PeCDF
123478-HxCDF
123678-HxCDF
123789-HxCDF
234678-HxCDF
1234678-HpCDF
1234789-HpCDF
OCDF
Chemical
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Soil Sampling Locations
SM20
0.0041
0.0104
0.0104
0.0234
0.0208
0.6935
0.0041
0.0104
0.0104
0.0104
0.0104
0.0104
0.0104
0.0477
0.0104
0.1127
SM34
0.0021
0.0053
0.0069
0.0291
0.0247
0.6724
0.0007
0.0053
0.0053
0.0053
0.0053
0.0053
0.0053
0.0672
0.0053
0.1547
SM41
0.0010
0.0029
0.0096
0.0276
0.0079
0.5640
0.0001
0.0001
0.0001
0.0002
0.0008
0.0000
0.0005
0.1297
0.0130
0.2425
SM44
0.0007
0.0050
0.0061
0.0453
0.0178
0.5602
0.0017
0.0013
0.0012
0.0061
0.0056
0.0012
0.0045
0.1069
0.0071
0.2292
SM40
0.0001
0.0014
0.0031
0.0361
0.0092
0.5835
0.0001
0.0002
0.0004
0.0025
0.0012
0.0002
0.0012
0.0845
0.0075
0.2687
SM46
0.0003
0.0026
0.0057
0.0495
0.0125
0.6725
0.0028
0.0022
0.0024
0.0047
0.0040
0.0002
0.0036
0.0653
0.0057
0.1662
SM51
0.0013
0.0048
0.0074
0.0536
0.0153
0.5977
0.0032
0.0009
0.0008
0.0069
0.0037
0.0011
0.0026
0.0919
0.0095
0.1992
-------
in Table
ctatain lab
Jed as a series of histograms. The x-axis in these plots is an ordered
:0&rlsPiTucnts. The actual order along the x-axis is not critical but it
must be consistent in each of the histograms. The y-axis is a plot of the relative abundance of each
constituent expressed as a decimal percentage of the calculated fingerprint mass. The data transformation
process in Step 3 defines a common y-axis scale of 0.0 to 1.0 decimal percent for each data set since no
constituent can be less than 0 percent of the fingerprint mass, no constituent can represent more than 100
percent of the fingerprint mass, and the sum of all constituents must be equal to 100 percent of the fingerprint
mass. Therefore, each sample histogram can be plotted on the same scale even though the reported concen-
trations may vary by orders of magnitude. A visual inspection of Figure 1 demonstrates that the seven data
sets presented in Table 1 define a single fingerprint pattern characterized by approximately 60 percent con-
stituent 6, approximately 20 percent constituent 16, approximately 10 percent constituent 14, approximately
4 percent of constituent 4 and trace amounts of the remaining 12 constituents. The dioxin-furan congener
number along the x-axis of Figure 1 corresponds to the chemical number listed in Tables 1 and 2.
_c
08
Q U.O
I
*- n R -
Q)
I
SO 4-
<5
0.
"a
.i n ? -
5 u-^
ฃ
n n
i
-g
_ r^ rirflfl 1 rn-Th-n
n
Sm20 - 231 ng/kg
SM34 - 446 ng/kg
D SM41 -177609 ng/kg
SM44- 1964 ng/kg
SM40- 1302406 ng/kg
SM46 - 50555 ng/kg
SM51 - 6226 ng/kg
Jk Jl
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Dioxin-Furan Fingerprint Congener
Regression Analysis Comparison of Sawmill Fingerprint Pattern
SM20 SM34 SM41 SM44 SM40 SM46 SM51
SM20
SM34
SM41
SM44
SM40
SM46
SM51
0.994 0
0
912 0.930
950 0
0
965
997
0.913
0
0
0
951
992
995
0.989
0.998
0.956
0.971
0.959
0.962
0.986
0.985
0.994
0.985
0.991
Average regression analysis comparison = 0.970.
Figure 1. Graphical representation and regression analysis comparison of a dioxin-
furan fingerprint pattern at a sawmill.
-------
Step 5. Statistical Assessment of Pattern Reproducibility
A statistical estimate of the comparability of the individual histograms is calculated using regression analysis.
Each histogram is individually compared with the remaining histograms to calculate an r2 value (regression
coefficient squared). The calculated r2 values fall into the range of 0.00 (the two patterns are totally
dissimilar) to 1.00 (the two patterns are identical) and provides a decimal estimate of the reproducibility or
comparability of two patterns. This example produced 21 histogram comparisons (SM20 vs. SM34, SM20
vs. SM41, etc.) with calculated r2 values ranging from 0.912 to 0.998 that are summarized in the Regression
Analysis Table in Figure 1. The estimated reproducibility of the dioxin-furan fingerprint shown in Figure 1
is 0.970 ฑ 0.027 (97 ฑ 3 percent). Despite the highly variable dioxin-furan concentrations that spanned four
orders of magnitude across the sawmill site and the variable number of non-detected congeners within each
set, the FALCON process identified a single, reproducible chemical signature that characterizes the dioxin-
furan contamination at this facility.
Step 6. Evaluate Remaining Data
Once a source fingerprint has been identified, regression analysis can be used to compare the dioxin-furan
congener distribution at other monitoring locations with the source fingerprint.
These results can be used to:
1. differentiate the source from background conditions,
2. demonstrate whether contamination detected at some distance from a site is related to the source,
3. map contaminant migration away from a source,
4. differentiate multiple sources of the same contaminant, and
5. estimate the mixing ratio of two plumes.
3.0 Contaminant Fingerprinting Case Studies
Case studies are presented to illustrate the versatility of FALCON in a variety of contamination scenarios.
Data used in these case studies were generated using standard analytical techniques in ongoing site
investigation, characterization, and monitoring programs. The intent of these case studies is not to provide
detailed site investigation histories for each example, but simply to illustrate the variety of situations in which
the FALCON fingerprinting process can be applied to characterize the occurrence and distribution of
contaminants.
3. / Source Characterization of Organic Contaminants - Pulp Mill Case Study
Fingerprinting was conducted to differentiate the impacts of a pulp mill from other potential dioxin-furan
sources along the lower Roanoke River in eastern North Carolina. The waste stream from this mill passed
through a baffled settling pond and discharged to a tributary of the Roanoke River. Samples were collected
from the tributary sediments known to be impacted by the pulp mill. Data from these samples were carried
through the FALCON process and produced the dioxin-furan fingerprint pattern shown in Figure 2 (the
fingerprint constituents listed by number along the x-axis are identified in Table 1). The 16 dioxin-furan con-
geners used in this example defined a pattern characterized by 40 to 70 percent constituent 7 (2378-
Tetrachlorodibenzofuran), 10 to 25 percent constituent 6 (1234678-Heptachlorodibenzodioxin), 10 percent
constituent 1 (2378-Tetrachlorodibenzodioxin), and 2 to 17 percent constituent 16(Octachlorodibenzofuran
(OCDF)).
-------
E
V)
1.0
0.8
'S
.2
Q
5
.2 0.6
O)
V
0.
0.4
CS2
CSS
CSS
CS7
CS9
J
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Dioxin-Furan Fingerprint Congener
Regression Analysis for Pulp Mill Fingerprint
CS2 CSS CSS CS7 CSS
CS2
CSS
CSS
CS7
CSS
0.958
0.849
0.948
0.993
0.951
0.826
0.791
0.902
0.985
0.754
Average regression analysis comparison = 0.896.
Figure 2. Dioxin-furan fingerprint pattern for a pulp mill effluent.
There were several site-specific factors that influenced the variability of the individual peak heights and the
fingerprint pattern for this source. First, the discharge practice to the tributary was terminated in 1987.
Changing hydrologic conditions over time (spring runoff, storm events, and tidal cycles) and bioturbation
could have affected the dioxin residuals in the sediments. Second, wood fiber accumulation in the sediments
complicated the sample collection process. However, despite the possible influence of these factors, the
graphical fingerprint for the pulp mill had an estimated reproducibility that ranged from 0.754 (75 percent)
to 0.993 (99 percent) with an average value of 0.896 ฑ 0.08 (90 ฑ 8 percent).
Roanoke sediment samples were collected upriver from the confluence with the tributary that had been
impacted by the pulp mill. These samples defined a second dioxin-furan fingerprint that is contrasted with
the pulp mill fingerprint in Figure 3. The upriver sediment fingerprint is presented as a series of histograms
and the pulp mill fingerprint is presented as an area plot that was calculated as the average of the tributary
samples shown in Figure 2. The upriver sediment fingerprint is characterized by 67 to 78 percent constituent
6 (1234678-HpCDD), 1 to 17 percent constituent 16 (OCDF), 1 to 8 percent constituent 14 (1234678-
Heptachlorodibenzofuran), 3 to 6 percent constituent 4, and trace amount of 12 other congeners. The
reproducibility of the upriver fingerprint pattern ranged from 0.996 (99+ percent) to 0.999 (99+ percent) with
an average value of 0.998 ฑ 0.005 (99+ ฑ 0.5 percent).
-------
1.0
0.8
Q
s
0.6
0.4
0)
D.
To
I*
Q
0.0
i\
Pulp Mill
River 106
River 105
River 108
River 101
River 104
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Dioxin-Furan Fingerprint Congener
Regression Analysis of Dioxin-Furan Fingerprint Patterns
River 106 River 105 River 108 River 101 River 104
Pulp Mill
River 106
River 105
River 108
River 101
River 104
0.063 0.062
1.000
0.062
0.998
0.996
0.061
0.999
0.998
1.000
0.061
0.998
0.997
0.999
0.999
Figures.
Comparison between mill and upriver sediments = 0.062.
Estimated reproducibility of upriver sediment pattern = 0.998.
Comparison of upriver sediment dioxin-furan fingerprint with the pulp mill
dioxin-furan fingerprint.
The presentation in Figure 3 (histogram vs. area plot) permits a rapid visual comparison of the two identified
fingerprint patterns at this site. The dioxin-furan fingerprint pattern for the pulp mill impacted tributary and
the dioxin-furan fingerprint for the upriver Roanoke sediments can be differentiated both visually and
statistically. The Roanoke upriver fingerprint is characterized by a single dominant peak for constituent 6
(1234678-HpCDD) and several minor peaks for constituent 4 (123678-HxCDD), constituent 14 (1234678-
HpCDF), and constituent 16 (OCDF). The pulp mill fingerprint is characterized by a major peak for
constituent 7 (2378-TCDF), and lesser peaks for constituent 1 (2378-TCDD), constituent 6 (1234678-
HpCDD), and constituent 16 (OCDF). The upriver sediment pattern can be distinguished from the pulp mill
pattern by the greater relative abundance of constituent 6 (67 to 78 percent versus 10 to 25 percent), the
virtual absence of constituent 7 (<1 percent versus 42 to 73 percent), and the virtual absence of constituent
1 (< 1 percent versus 10 percent). For the purpose of this paper, an average pulp mill fingerprint was
calculated based on the results presented in Figure 2. This "average" pulp mill fingerprint was compared with
each of the Roanoke upriver sediment samples using regression analysis. As summarized in the regression
table associated with Figure 3, the comparability of the upriver dioxin-furan fingerprint with the pulp mill
dioxin fingerprint ranged from 0.061 (6 percent) to 0.063 (6 percent). A t-Test analysis of the set of r2 values
for the pulp mill and Roanoke sediments comparisons with the set of r2 values for the Roanoke sediment
-------
reproducibility comparisons demonstrates that the pulp mill dioxin has a significantly different fingerprint
pattern at p = 0.05. In this case study, FALCON identified a reproducible (90 ฑ 8 percent) dioxin-furan
fingerprint for the pulp mill source (Figure 2), a second highly reproducible (99 ฑ 0.5 percent) dioxin-furan
fingerprint for the upriver source(s) (Figure 3), and the two fingerprint patterns could be easily differentiated
both graphically and statistically.
Any dioxin-furan contamination washed out of the tributary will mix with dioxin-furan contamination from
the upriver source (s). This will create an unknown mixture of upriver dioxin-furan with pulp mill dioxin-
furan in sediments collected below the tributary confluence. Using the two identified FALCON fingerprints,
it is possible to calculate the expected congener distribution pattern for possible mixtures of the two source
fingerprints. The relative abundance of each congener is known for 100 percent upriver sediment (Figure 3),
is known for 100 percent pulp mill waste (Figure 2), and the expected congener distribution was calculated
for mixtures of 90 percent upriver sediment plus 10 percent pulp mill waste, 80 percent upriver sediment plus
20 percent pulp mill waste, continuing to 10 percent upriver sediment plus 90 percent pulp mill waste. These
results are presented in Table 3 in which each vertical column represents the expected dioxin-furan congener
distribution (fingerprint) for the indicated sediment-pulp mill mixture.
If there were no pulp mill dioxin-furan in the lower river sediments, each downriver station would be
expected to have a congener distribution pattern similar to the upriver fingerprint (Table 3, 100 percent
upriver and 0 percent pulp mill). Therefore, constituent 6 would have a relative abundance of 0.813 (81
percent), constituent 7 would have a relative abundance of 0.016 (2 percent), and the remaining congeners
would have the relative abundance indicated. However, if contaminated tributary sediments were entering
the river, there would be a characteristic shift in the relative abundance of each congener with increasing
amounts of pulp mill dioxin-furan. Thus, the relative abundance of congener 6 would gradually decrease from
0.812 (81 percent) to 0.196 (20 percent) and the relative abundance of congener 7 would simultaneously
increase from 0.016 (2 percent) to 0.546 (55 percent) as the mixture changes from 100 percent upriver
sediment and 0 percent pulp mill waste to 0 percent upriver sediment and 100 percent pulp mill waste. The
relative abundance of each remaining congener in the fingerprint would also be expected to change as
indicated in Table 3.
The actual fingerprint pattern in each downriver sample can be compared to the calculated fingerprint patterns
in Table 3 using regression analysis. This will produce a range of values for each sample as shown in
Table 4. The fingerprint comparison with the best match (maximum r2 value) provides an estimate of the
dioxin-furan mixture at that location. The following examples from Table 4 illustrate the assessment process.
1. The sample from upriver Station R101 produced a match of 0.999 (99+ percent) with the upriver
fingerprint. However, this sample only produced a match of 0.924 (92 percent) with the 70 percent
upriver - 30 percent pulp mill mixture, a match of 0.437 (44 percent) with the 30 percent upriver - 70
percent pulp mill mixture, and a 0.061 (6 percent) match with the pulp mill fingerprint. Since the
maximum fingerprint match for this sample occurred with the 100 percent - 0 percent pulp mill mixture,
this location was considered to be 100 percent upriver dioxin-furan.
2. The sample from downriver Station R118 produced a match of 0.904 (90 percent) with the upriver
fingerprint and also produced a match of 0.905 (90 percent) with the 90 percent upriver sediment - 10
percent tributary sediment mixture. This sample produced a match of 0.859 (86 percent) with the 70
percent upriver - 3 0 percent pulp mill mixture, 0.511(51 percent) with the 3 0 percent upriver - 70 percent
pulp mill mixture, and only 0.198 (20 percent) with the pulp mill fingerprint. Since the maximum
fingerprint match for this sample occurred for both the 100 percent upriver sediment - 0 percent tributary
sediment mixture and the 90 percent upriver sediment - 10 percent tributary sediment mixture, this
location was considered to be 95 percent upriver dioxin-furan and 5 percent pulp mill dioxin-furan.
3. The sample from downriver Station R125 produced a match of 0.718 (72 percent) with the upriver
fingerprint. The fingerprint match increased to 0.846 (85 percent) with the 80 percent upriver sediment -
-------
20 percent pulp mill mixture and reached a maximum value of 0.966 (97 percent) with the 50 percent
upriver - 50 percent pulp mill mixture. The fingerprint match decreased to 0.846 (85 percent) with the
20 percent upriver - 80 percent pulp mill mixture and 0.662 (66 percent) with the pulp mill fingerprint.
This location was considered to have 50 percent upriver dioxin-furan and 50 percent pulp mill dioxin-
furan.
4. The sample from downriver station R140 produced a match of only 0.052 (5 percent) with the upriver
fingerprint. The fingerprint match increased to 0.231 (23 percent) with the 70 percent upriver - 30 percent
pulp mill mixture, 0.696 (70 percent) with the 30 percent upriver - 70 percent pulp mill mixture, and
reached a maximum of 0.944 (94 percent) with the pulp mill fingerprint. Since the maximum fingerprint
match occurred with the pulp mill source fingerprint, this location was considered to be 100 percent pulp
mill dioxin-furan.
The fingerprint analysis results for the lower Roanoke sediment samples are presented in Table 4.
Each of the samples produced a maximum fingerprint match (highlighted and bolded) of 0.832 (83 percent),
or higher, with one of the calculated fingerprint patterns. An inspection of Table 4 shows that 90 to 100
percent of the dioxin-furan contamination upriver from the mill could be attributed to upriver sources.
However, the influence of the mill is evident downriver from the facility. Samples from downriver stations
R120, R127, R128, R143, and R144 have a fingerprint indicating 70 percent upriver dioxin-furan and 30
percent pulp mill dioxin furan. Samples from R133 and R145 have a fingerprint consistent with a 30 percent
upriver - 70 percent pulp mill mixture, and samples from R121 and R140 produced a maximum match with
the pulp mill source fingerprint (0 percent upriver and 100 percent pulp mill). The spatial variability of the
pulp mill dioxin in the downriver samples was attributed to site-specific hydrologic factors such as storm
events, channel scouring, tidal effects and eddying. Overall, fingerprinting results from more than 40
downriver locations (only a portion of which are presented in Table 4) indicate that approximately 35 to 40
percent of the dioxin-furan contamination downriver from the tributary confluence could be attributed to the
pulp mill. This example demonstrates that the FALCON fingerprinting technique is able to (1) develop a
characteristic, reproducible fingerprint for the dioxin-furan source (pulp mill), (2) differentiate pulp mill
dioxin-furan from background and/or upriver sources of the same contaminants, (3) quantitatively estimate
the mixing that has occurred between pulp mill dioxin-furan and upriver sources, and (4) track the migration
of the pulp mill dioxin-furan several miles from the source.
-------
Table 3. Calculated Dioxin-Furan Congener Distributions for Mixtures of Roanoke Sediments and Pulp Mill Effluent
Con-
gener
#
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
100% Upriver
0% pulp
mill
0.003
0.005
0.006
0.018
0.022
0.813
0.017
0.007
0.007
0.008
0.008
0.009
0.008
0.048
0.008
0.034
90% Upriver
10% pulp
mill
0.011
0.005
0.006
0.017
0.021
0.751
0.069
0.006
0.007
0.007
0.007
0.008
0.007
0.047
0.007
0.040
80% Upriver
20% pulp
mill
0.019
0.004
0.005
0.016
0.019
0.690
0.122
0.006
0.007
0.007
0.007
0.007
0.006
0.046
0.006
0.046
70% Upriver
30% pulp
mill
0.028
0.004
0.005
0.016
0.018
0.628
0.175
0.006
0.007
0.007
0.006
0.006
0.006
0.044
0.006
0.053
60% Upriver
40% pulp
mill
0.036
0.004
0.005
0.015
0.017
0.566
0.228
0.005
0.006
0.006
0.005
0.006
0.005
0.043
0.005
0.059
50% Upriver
50% pulp
mill
0.044
0.003
0.004
0.015
0.016
0.504
0.281
0.005
0.006
0.006
0.005
0.005
0.004
0.042
0.005
0.065
40% Upriver
60% pulp
mill
0.052
0.003
0.004
0.014
0.014
0.443
0.334
0.005
0.006
0.006
0.004
0.004
0.004
0.040
0.004
0.071
30% Upriver
70% pulp
mill
0.060
0.003
0.003
0.014
0.013
0.381
0.387
0.004
0.006
0.005
0.003
0.004
0.003
0.039
0.003
0.077
20% Upriver
80% pulp
mill
0.068
0.002
0.003
0.013
0.012
0.319
0.440
0.004
0.006
0.005
0.003
0.003
0.003
0.037
0.003
0.084
10% Upriver
90% pulp
mill
0.076
0.002
0.003
0.012
0.011
0.258
0.493
0.004
0.005
0.004
0.002
0.002
0.002
0.036
0.002
0.090
0% Upriver
100% pulp
mill
0.085
0.002
0.002
0.012
0.009
0.196
0.546
0.003
0.005
0.004
0.002
0.001
0.001
0.035
0.001
0.096
Values represent the estimated abundance of each dioxin-furan congener as a decimal percentage of the fingerprint mass in the sediment-pulp mill mixture.
-------
Table 4. Regression Analysis Comparison of Roanoke Sediment Samples with Calculated Dioxin-Furan Fingerprints Patterns
River
Location
Upriver
Station
R101
R103
R106
R107
R109
R110
R115
Upriver
Fingerprint
100 URS
to
OTS
0.999
0.888
0.859
0.863
0.904
0.900
0.998
Calculated composition of upriver sediment - pulp mill mixtures
90 URS
to
10 TS
0.991
0.892
0.854
0.857
0.897
0.894
0.994
80 URS
to
20 TS
0.968
0.883
0.835
0.838
0.875
0.874
0.975
70 URS
to
SOTS
0.924
0.856
0.798
0.800
0.833
0.834
0.935
60 URS
to
40 TS
0.850
0.807
0.740
0.770
0.769
0.772
0.866
50 URS
to
SOTS
0.741
0.733
0.659
0.658
0.682
0.686
0.761
40 URS
to
60 TS
0.599
0.637
0.560
0.558
0.576
0.582
0.622
30 URS
to
70 TS
0.437
0.527
0.451
0.447
0.460
0.467
0.460
20 URS
to
SOTS
0.279
0.414
0.341
0.338
0.346
0.352
0.300
10 URS
to
90 TS
0.149
0.307
0.242
0.239
0.243
0.249
0.167
Pulp Mill
Fingerprint
0 URS
to
100 TS
0.061
0.215
0.160
0.157
0.158
0.164
0.073
Pulp Mill
Downriver
R116
R117
R118
R119
R120
R121
R122
R125
R126
R127
R128
R131
R132
R133
R134
R135
R138
R139
R140
R141
R142
R143
R144
R145
R146
0.911
0.867
0.904
0.916
0.851
0.155
0.500
0.718
0.755
0.832
0.890
0.798
0.839
0.511
0.943
0.819
0.745
0.842
0.052
0.799
0.730
0.738
0.824
0.524
0.655
0.909
0.883
0.905
0.927
0.891
0.216
0.572
0.783
0.807
0.874
0.927
0.844
0.837
0.589
0.967
0.832
0.808
0.851
0.093
0.845
0.789
0.776
0.857
0.598
0.714
0.892
0.844
0.891
0.924
0.924
0.294
0.655
0.846
0.856
0.908
0.956
0.883
0.821
0.674
0.979
0.834
0.869
0.849
0.152
0.884
0.845
0.809
0.882
0.677
0.773
0.857
0.807
0.859
0.903
0.942
0.389
0.748
0.904
0.895
0.929
0.969
0.910
0.788
0.762
0.974
0.820
0.923
0.830
0.231
0.912
0.894
0.829
0.909
0.758
0.826
0.798
0.749
0.804
0.859
0.940
0.497
0.842
0.946
0.917
0.931
0.960
0.919
0.734
0.845
0.945
0.785
0.960
0.789
0.331
0.920
0.926
0.832
0.882
0.833
0.965
0.715
0.668
0.724
0.789
0.911
0.641
0.926
0.966
0.914
0.905
0.924
0.901
0.658
0.915
0.888
0.726
0.974
0.725
0.448
0.904
0.936
0.811
0.847
0.894
0.884
0.612
0.568
0.624
0.695
0.853
0.729
0.982
0.957
0.883
0.851
0.858
0.856
0.563
0.961
0.803
0.645
0.958
0.638
0.574
0.858
0.916
0.765
0.785
0.930
0.876
0.497
0.457
0.511
0.584
0.769
0.829
0.993
0.915
0.823
0.770
0.766
0.783
0.456
0.977
0.695
0.548
0.910
0.536
0.696
0.785
0.866
0.694
0.698
0.937
0.838
0.381
0.347
0.395
0.467
0.666
0.905
0.950
0.846
0.740
0.671
0.657
0.689
0.349
0.960
0.576
0.443
0.935
0.428
0.803
0.692
0.791
0.607
0.597
0.912
0.776
0.274
0.247
0.288
0.355
0.556
0.952
0.863
0.758
0.644
0.564
0.542
0.586
0.252
0.915
0.457
0.341
0.742
0.325
0.887
0.589
0.700
0.512
0.491
0.862
0.695
0.185
0.164
0.198
0.256
0.450
0.972
0.749
0.662
0.545
0.459
0.433
0.484
0.170
0.851
0.348
0.251
0.642
0.235
0.944
0.487
0.602
0.418
0.390
0.795
0.608
Tabulated results are regression analysis comparisons between actual sediment composition and calculated sediment-tributary mixtures.
URS = upriver sediments TS = tributary sediments 0.9305 = maximum fingerprint matches
3.2 Inorganic Source Characterization and Leachate Migration Mapping - Gold Mine Case Study
-------
An industry that displayed significant growth in the last quarter of the twentieth century was gold mining
utilizing cyanide leaching technology. This process1 uses a recirculating cyanide solution to extract gold,
and other metals, from the ore being processed. An alkaline cyanide solution is prepared in an area
referred to as a barren pond. This solution is sprayed on a crushed ore pile to extract gold by
complexation as it percolates through the ore. The gold-rich leachate from this process is collected in a
second pond referred to as the pregnant pond. Gold and other metals are recovered from the leachate and
the cyanide solution is pumped back to the barren pond to repeat the process.
The State of Nevada currently requires heap leaching facilities to conduct quarterly monitoring for 12
geochemical parameters, 28 trace metals, and cyanide at several specified locations (barren pond,
pregnant pond, tailings pond, and ground water monitoring wells).1 Monitoring data from 35 heap
leaching facilities were reviewed to develop a better understanding of the composition of mine leachates
and to identify a shortened list of potential detection monitoring parameters for this industry.1 This study
demonstrated that nine geochemical parameters were always the most abundant constituents in mine
leachates and their relative abundance (expressed as a percentage of total dissolved solids) was a constant
at each mine despite the highly variable concentrations. These parameters were alkalinity, calcium,
chloride, fluoride, magnesium, nitrate, potassium, sodium, and sulfate.
Data for the nine geochemical parameters specified above from the barren pond, the pregnant pond, and
tailings reclaim water at a Nevada heap leaching facility were carried through the fingerprinting process
described in Section 2.0 and plotted as shown in Figure 4. Despite the variability displayed by the
individual parameters (e.g., sulfate ranged from 969 to 1720 mg/L, and sodium ranged from 464 to 985
mg/L), the selected parameters define a distinctive chemical signature in which sulfate represents
approximately 50 percent of the total dissolved solids concentration, sodium represents approximately 18
percent of the total dissolved solids concentration, chloride represents approximately 10 percent of the
total dissolved solids concentration, and the remaining constituents are trace constituents of this pattern.
Although cyanide (CN), copper (Cu), and total trace metals (the sum of all trace metal concentrations,
TTM) are not significant components of the mine leachate fingerprint, they have been included in Figure
4 for comparison. A separate attempt at using trace metals for fingerprinting mine leachates did not
define a distinctive fingerprint.1
The reproducibility of the mine leachate fingerprint at this facility is demonstrated in two ways. First,
based on quarterly samples, the leachate fingerprint had a reproducibility overtime of 0.980 (98 percent)
at the barren pond, 0.986 (99 percent) at the pregnant pond, and 0.985 (99 percent) at the tailings reclaim
pond. Second, the barren pond, the pregnant pond and the tailings reclaim area are all part of a
recirculating system at the mine and a similar fingerprint pattern might be expected at each of these
locations. As demonstrated by the regression analysis comparisons tabulated in Figure 4, the monitoring
data from these different locations defined a consistent source fingerprint with an average reproducibility
of 0.984 (98 percent). Therefore, data overtime at a single location (quarterly monitoring data) and data
from different locations within the leaching circuit define a reproducible geochemical fingerprint at this
facility.
12
-------
J2 08
Q u.o-
H
5
o> n R.
ercenta
D C
ป. C
u- vj.t
"5
'o
ฎ n 9-
n n
D Barren Q1 EH Barren Q2 EH Barren Q3 EH Barren Q4
EH Pregnant Q1 EH Pregnant Q2 EH Pregnant Q3 EH Pregnant Q4
H failings Q1 I failings Q2 I failings Q3 I failings Q4
111
mm ~ jH
Alk Ca Cl F Mg NO3 K Na SO4 CN
Mine Leachate Fingerprint Constituents
Cu TTM
Regression Analysis of Gold Mine Tailings Leachate Fingerprint
Background Ground Water
Barren Pond
Pregnant Solution
Tailings Reclaim
Barren
Pond
0.435
Pregnant
Pond
0.460
0.983
Tailings
Reclaim
0.406
0.972
0.997
Estimated reproducibility of mine leachate fingerprint = 0.984.
Comparison between mine leachate and ground water = 0.433.
Figure 4. Geochemical fingerprint at a gold mine using cyanide heap leaching.
A further evaluation of mine site monitoring records demonstrated that the same set of geochemical
parameters used to fingerprint leachate at gold mines can also be used to fingerprint ground water in the
vicinity of the mine.1 The ground-water fingerprint developed from two years of quarterly monitoring at
an upgradient location is contrasted with the heap leaching fingerprint in Figure 5. The quarterly
upgradient ground water data are presented as a series of histograms and the mine leachate fingerprint is
presented as an area plot. The first observation is that the ground water data define a single fingerprint
pattern characterized by 25 to 30 percent sodium, 20 percent sulfate, 20 percent alkalinity (Alk in the
figures), and 10 to 15 percent chloride. The estimated reproducibility of this chemical signature for
ground water upgradient of the mine site is 0.941 (94 percent). The second observation is that the
ground-water fingerprint developed with the FALCON procedure is visually distinct from the mine
leachate fingerprint. Specifically, the mine leachate is enriched in sulfate and depleted in alkalinity when
compared to the regional ground water. Despite the fact that the fingerprint constituents are naturally
occurring substances and their concentrations are highly variable, the mine leachate and the regional
ground water each have a distinctive, reproducible fingerprint that can be differentiated from each other
visually and with regression analysis. In this example, there was only a 43 percent comparability between
the two fingerprint patterns.
13
-------
M
Q
o
3,
s
0)
Q.
W
'5
0
Q
1.0
0.8
0.6
0.4
12/91
11/92
3/92
2/93
6/92
5/93
8/92
Barren Pond
Ca
Cl F Mg N03 K Na SO4 CN
Mine Leachate Fingerprint Constituents
Cu TTM
Regression Analysis Matrix for Gold Mine Upgradient Ground Water
8/91 3/92 6/92 9/92 11/92 2/93 5/93
Barren Pond
8/91
3/92
6/92
9/92
11/92
2/93
5/93
0.493 0.489 0.566
0.999 0.978
0.982
0.609
0.948
0.944
0.960
0.475
0.924
0.919
0.882
0.908
0.558
0.980
0.980
0.984
0.975
0.948
0.414
0.963
0.966
0.922
0.835
0.857
0.908
Average upgradient fingerprint comparison = 0.941.
Mine fingerprint comparison with upgradient ground water = 0.515.
Figure 5. Comparison of a ground-water fingerprint with the mine leachate fingerprint.
Once a source fingerprint pattern has been established for the mine, it can be used to monitor leachate
migration into the environment. This is accomplished by comparing the geochemical pattern at each
downgradient monitoring location with either the source fingerprint or the upgradient ground-water
fingerprint. If the location was not impacted by fugitive mine leachate, it would be expected to have a
geochemical fingerprint that more closely resembled the upgradient fingerprint rather than the mine
leachate fingerprint. However, if fugitive leachate was impacting a location, the fingerprint would more
closely resemble the mine leachate fingerprint rather than the upgradient fingerprint. Fingerprint analyses
for several monitoring wells in the vicinity of a heap leaching facility are summarized in Table 5. An
inspection of the results indicates that the wells fall into one of three categories. One category consists of
upgradient wells that have a geochemical fingerprint that matches the upgradient fingerprint (80 to 99+
percent) but is different from the mine leachate fingerprint (36 to 63 percent). A second category consists
of downgradient wells with a geochemical fingerprint similar to upgradient ground water (84 to 88
percent) and different from the mine leachate fingerprint (48 to 60 percent). The third category consists
of wells with a geochemical fingerprint that has a poor match with the upgradient ground-water
fingerprint (44 to 75 percent) and a stronger similarity to the mine leachate fingerprint (80 to 93 percent).
14
-------
The compositional shift from alkalinity-rich and sulfate-poor to sulfate-rich and alkalinity-poor that has
occurred in the category three wells is consistent with mine leachate migrating into that area. Once the
impacted wells have been identified through the fingerprinting process, the results can be plotted and
contoured to delineate specific areas impacted by fugitive mine leachate. This example demonstrates that
the FALCON process develops reproducible fingerprints for mine leachate that can be visually and
statistically differentiated from background ground-water conditions in the vicinity of the mine. Also, it
demonstrates that the mine leachate fingerprint retains its distinctive chemical identity as it migrates
through the ground water and can be used as an internal tracer to detect and map its presence some
distance from the designated tailings disposal area.
Table 5. Comparison of Ground-water Fingerprints with a Mine Leachate Fingerprint
Classification
Category 1
Category 2
Category 3
Sampling
Location
Well 2
Well 4
Well 22
Mine Site
Well 37
Well 42
Well 38
Well 45
Well 1
Well 9
Well 12
Well 15
Well 39
Well 44
Well 40
Well 61
Well 66
Site Location
Upgradient
Upgradient
Upgradient
Downgradient
Downgradient
Downgradient
Downgradient
Downgradient
Downgradient
Downgradient
Downgradient
Downgradient
Downgradient
Downgradient
Downgradient
Downgradient
Fingerprint Comparison
Mine Source
0.433
0.627
0.357
1.000
0.480
0.492
0.602
0.531
0.835
0.898
0.844
0.874
0.866
0.932
0.801
0.885
0.897
Background GW
1.000
0.937
0.800
0.429
0.885
0.844
0.880
0.882
0.463
0.492
0.747
0.621
0.584
0.441
0.592
0.610
0.381
Tabulated values are regression analysis comparisons between ground water and the
indicated fingerprint pattern.
3.3 Early Detection of Leachate Migration - Copper Mine Case Study
A review of monitoring results from several copper mines demonstrated that the same set of geochemical
parameters used at gold heap leaching facilities would also be useful for fingerprinting leachates at copper
mines.1 As indicated in Figure 6, tailings pond data over a period of 14 years produced a series of
histograms that defined a single chemical fingerprint characterized by 55 to 60 percent sulfate, 15 percent
calcium, 10 percent magnesium, 10 percent total trace metals (TTM), and 5 percent alkalinity. Despite
variability in the ore being processed and the long period of record, the tailings pond at this facility had a
single characteristic fingerprint with an estimated reproducibility of 0.970 (97 percent). In addition, this
pattern was distinctively different from the alkalinity-rich, sulfate-poor background ground-water
fingerprint in the vicinity of the mine (fingerprint comparison = 0.016 (2 percent)).1 Fingerprint
comparisons were used to identify several downgradient wells that had been impacted by mine leachate as
demonstrated in the previous example. The monitoring results from one impacted well at this copper
mine provided a set of data that illustrates another possible application of the FALCON procedure.
Relative abundance (actual concentration divided by TDS) for the geochemical fingerprint parameters at
well 1225 at this site are plotted as a function of time in Figure 7. Between 1970 and 1976, the dominant
geochemical parameter in the ground water at this location was alkalinity while sulfate was almost
15
-------
negligible. This alkalinity-rich, sulfate-poor pattern, that is typical of most unimpacted freshwater
systems, had a 99 percent match with the ground water upgradient of the mining facility and less than a 5
percent match with the tailings pond leachate at the copper mine. However, the ground water
composition at this location began to change around 1976. There was a gradual reduction in the relative
abundance of alkalinity and a concurrent increase in the relative abundance of sulfate. This
compositional shift, that is consistent with a sulfate-rich, alkalinity poor leachate entering the area,
continued through 1984 when the Well 1225 fingerprint produced a 95 percent match with the tailings
pond leachate. Even though sulfate and alkalinity are naturally occurring substances, fingerprint analysis
clearly identifies mine leachate as the causative factor for the ground-water changes observed at this
location. Thus, the fact that the geochemical fingerprint retains its distinctive identity as it migrates
through the ground water could have been used to provide an early warning of mine leachate migration.
e
s
0)
3
a>
a.
O
1.0
0.8
D Leachate 12/80 D Leachate 3/81
Leachate 9/81 D Leachate 3/82
Leachate 1/82 D Leachate 3/83
Leachate 2/84 D Leachate 3/84
Leachate 6/84 Leachate 7/84
Leachate 6/81
Leachate 6/82
Leachate 6/83
Leachate 5/84
D Upgradient GW
Ca
Cl F Mg NO3 K Na
Mine Leachate Fingerprint Constituents
SO4 TTM
12/80
Regression Analysis Matrix for Copper Mine Tailings Pond Fingerprint
Tailings Pond Sampling Date
3/81 6/81 9/81 3/82 6/82 12/82 3/83 6/83 2/84 3/84 5/84
6/84 7/84
12/80
3/81
6/81
9/81
3/82
6/82
12/82
3/83
6/83
2/84
3/8
5/84
6/84
7/84
0.997 0.995 0.994 0.997 0
0.996 0.998 1.000 0
0.998 0.998 0
0.999 1
0
994
998
998
000
999
0
0
0
0
0
0
999
996
984
992
996
992
0.985
0.995
0.992
0.996
0.995
0.997
0.984
0.952
0.967
0.968
0.973
0.967
0.974
0.949
0.981
0.929
0.947
0.946
0.954
0.947
0.954
0.926
0.964
0.996
0.919
0.939
0.936
0.946
0.938
0.946
0.916
0.958
0.994
0.999
0.911
0.933
0.929
0.941
0.932
0.941
0.908
0.954
0.991
0.998
0.999
0.936
0.957
0.953
0.965
0.956
0.965
0.933
0.976
0.986
0.986
0.983
0.986
0.937
0.959
0.954
0.967
0.958
0.968
0.935
0.980
0.986
0.984
0.982
0.985
0.999
Figure 6.
Average mine leachate fingerprint reproducibility = 0.970.
Comparison of a copper mine tailings pond fingerprint with the background ground-water
fingerprint.
16
-------
CO
Q
1.0
0.8
o
S) 0.6
d)
a
o
n
'5
0.4
0.2
0.0
Sulfate
TTM
D Na + K
D Cl + F + NO3
D Ca + Mg
Alkalinity
6/70 11/70 6/71 11/72 12/74 7/76 12/78 2/80 12/80 9/81 6/82 6/83 11/84
Well 1225 Sampling Date
Ground-Water Conditions at Copper Mine Well 1225
Sampling Location
Tailings Pond Leachate
Well 122510/70
Well 12253/71
Well 122511/72
Well 12256/73
Well 122512/74
Well 122510/75
Well 12257/76
Well 12255/78
Well 122512/78
Well 122511/79
Well 12252/80
Well 12255/80
Well 12258/80
Well 122512/80
Well 12253/81
Well 12256/81
Well 12259/81
Well 12253/82
Well 12256/82
Well 122512/82
Well 12256/83
Well 122512/83
Well 12255/84
Well 122511/84
Well 12255/85
PH
Standard Units
2.0
nd*
nd*
nd*
nd*
nd*
nd*
nd*
nd*
nd*
nd*
nd*
nd*
nd*
nd*
nd*
nd*
7.4
7.0
6.7
7.2
7.4
7.4
7.7
7.1
6.9
Total Metals
(mg/L)
1963
0.291
0.372
0.467
0.410
0.296
0.341
0.357
0.697
0.383
0.354
0.350
0.342
0.322
0.354
0.295
0.302
0.110
0.030
0.150
1.190
1.110
0.000
0.470
1.570
1.550
Fingerprint Match
Decimal Percent
1.000
0.033
0.027
0.031
0.037
0.017
0.043
0.038
0.018
0.075
0.026
0.375
0.576
0.755
0.717
0.898
0.805
0.731
0.890
0.890
0.946
0.867
0.950
0.920
0.945
0.949
Figure?.
* nd = no data
Geochemical fingerprint analysis of ground water downgradient of a
copper mine tailings basin.
17
-------
A review of additional monitoring data from this location (Figure 7) provides further insight into the
capabilities of FALCON as a mechanism to provide early warning or early detection of leachate
migration. The tailings pond at this facility has a total trace metals concentration (sum of 17 trace metals
being routinely monitored) of 1963 mg/L and apH of 2.0. Between 1970 and 1975, while the
geochemical fingerprint was essentially identical to upgradient (background) conditions, the total trace
metals concentration at this well was in the range of 0.291 to 0.467 mg/L. While the tailings pond
fingerprint was being established at Well 1225 between 1976 and 1982, the total trace metals
concentration remained in the range of 0.030 to 0.697 mg/L. After the tailing leachate fingerprint was
fully established, the total trace metals concentration had only increased slightly to 0.470 to 1.570 mg/L.
Also, the pH at this location was still in the range of 6.7 to 7.7 between 1981 and 1985. (The pHdata
from this set between 1970 and 1981 was missing. However, the pH at upgradient locations and other
impacted downgradient locations was in the range of 6.7 to 7.7 between 1970 and 1985.) These field
results, and an additional laboratory study1, demonstrate that the geochemical fingerprint migrates faster
through the ground water than other inorganic parameters of environmental concern that may be in mine
leachate. Thus, in addition to providing a reliable characterization of mine leachates at their source and
acting as an internal tracer, a geochemical fingerprint acts as a good indicator parameter because it
migrates faster than mine leachate constituents of higher regulatory concern (pH and trace metals).
Fingerprint analysis can provide an early warning so that remedial action can be initiated while a
contaminant migration problem is smaller and more manageable.
This example demonstrates that the inorganic fingerprints can be used to characterize mine leachates at
their source and to differentiate the leachate from background conditions. In addition, since the
fingerprint retains its identity and migrates faster than other contaminants of higher regulatory concern,
the fingerprint can function as an effective detection monitoring parameter.
3.4 Ground Water and Surface Water Mixing - Molybdenum Mine Case Study
This facility is located adjacent to the Red River in New Mexico and monitoring was conducted at the
mine site, several ground water and seep locations, and the nearby river. The same set of geochemical
parameters used in the gold mine and copper mine case studies were also useful at this site. Although the
concentrations of the individual fingerprint parameters varied by a factor of 10 across the site, the
monitoring results defined a single mine leachate pattern characterized by 60 percent sulfate, 20 percent
calcium, 10 percent alkalinity, and small amounts of the remaining geochemical parameters (Figure 8).
The regression analysis comparison of eight source samples produced r2 values ranging from 0.9618 to
0.9999. The estimated reproducibility of this source fingerprint was 0.9894 (99 percent).
Monitoring was conducted at several ground water and surface seep locations between the mine
operations and the river. Regression analysis of the geochemical pattern at these locations produced
fingerprint comparisons of 94 to 99 percent with the source fingerprint. In addition, surface water data
from several gulches that drain from the mine site towards the Red River also produced geochemical
fingerprints with a strong similarity to the mine leachate (> 90 percent). Fingerprint analysis indicated
that fugitive mine leachate was migrating through the ground water and may be entering the nearby river.
18
-------
I.U
.** no
(0 U.o
Q
14-
O
0
O) n c
Percenta
3 C
ป C
Decimal
3 C
0 4
On
.U
r
k_
1 1 1 1 1 1 m !~~i~~ r~i \ \
Alk Cl F
n
j_,.
S04
[
CE
rri
I
i
iniiH r
K Mg
81
n S2
n S3
S4
S5
n se
S7
D S8
lltol
Na
Mine Leachate Fingerprint Constituents
Regression Analysis of Molybdenum Mine Fingerprint
Molybdenum Mine Sampling Locations
S1 S2 S3 S4 S5 S6 S7 S8
S1
S2
S3
S4
S5
S6
S7
S8
1.000 0.997 0
0.994 0
0
977
973
990
0
0
1
996
994
000
0.989
0.984
0.981
0.987
0.962
0.988
0.998
0.998
0.996
0.984
0.995
0.975
0
0
0
0
0
0
1
997
995
997
989
996
974
000
Average fingerprint reproducibility = 0.989.
Figure 8. Geochemical fingerprint of molybdenum mine leachate.
The monitoring program at this mine site included 16 sampling stations in the nearby river. Data from
three locations upriver from the mine site defined a geochemical fingerprint pattern that is contrasted with
the mine leachate fingerprint in Figure 9. The river pattern, presented as a series of histograms, is
characterized by 45 to 60 percent alkalinity, approximately 25 percent calcium, approximately 10 percent
chloride, and 5 to 15 percent sulfate. This alkalinity-rich, sulfate-poor pattern upriver of the mine is
typical of most fresh water systems and can be clearly differentiated from the sulfate-rich, alkalinity-poor
mine fingerprint of the mine leachate that is presented as an area plot in Figure 9. The estimated
reproducibility of the Red River fingerprint at these three upriver stations was 0.9804 (98 percent) and the
comparability with the molybdenum mine fingerprint was only 0.0189 (2 percent). As in previous
examples, the two inorganic fingerprint patterns can be readily distinguished from each other both
visually and with regression analysis.
19
-------
CO
Q
o
ffi
O)
S
o>
a
d)
0.
re
1.0
0.8
0.6
0.4
0.2
0.0
D Tailings Leachate
Upriver RR01
DUpriver RR02
D Upriver RR04
Alk Cl F SO4 Ca K Mg Na
Mine Leachate Fingerprint Constituents
Fingerprint Analysis in Red River Near Molybdenum Mine
Station
RR04
RR02
RR01
RR03
RR05
RR06
RR09
RR11
RR08
RR07
RR10
RR12
RR13
RR14
RR15
RR16
Location
Upriver from mine
Upriver from mine
Upriver from mine
Adjacent to mine
Adjacent to mine
Adjacent to mine
Adjacent to mine
Adjacent to mine
Adjacent to mine
Adjacent to mine
Adjacent to mine
Adjacent to mine
Adjacent to mine
Downriver from mine
Downriver from mine
Downriver from mine
Background
Fingerprint
Match
0.987
0.988
0.976
0.906
0.875
0.767
0.912
0.969
0.832
0.856
0.824
0.458
0.698
0.465
0.488
0.469
Leachate
Fingerprint
Match
0.011
0.011
0.012
0.063
0.082
0.192
0.064
0.015
0.133
0.111
0.140
0.496
0.262
0.484
0.467
0.477
Percent
Leachate
in River
0
0
10
20
20
30
20
10
25
20
25
40
30
40
40
40
Figure 9. Comparison of mine leachate fingerprint with the background Red River
fingerprint.
An evaluation of the data from the remaining river sampling locations adjacent to and downriver from the
mine site demonstrated that the geochemical composition of the river was being altered. At stations
adjacent to the mine site (RR03 to RR13), the comparability of the fingerprint with the upriver fingerprint
ranged from 46 to 97 percent. At the three stations downriver from the mine site (RR14 to RR16), the
comparability with the upriver fingerprint was reduced to 47 to 49 percent. These compositional changes
were due to a reduced relative abundance of alkalinity and an increased relative abundance of sulfate as
20
-------
the river flowed past the mining operation. This shift is consistent with a sulfate-rich, alkalinity-poor
leachate characteristic of the mining operation (Figure 8) entering the river and is reflected by the
increased comparability of the river fingerprint with the mine leachate fingerprint. The upriver stations
only had a 1 percent match with the mine leachate fingerprint. However, as the river passed the site, the
leachate fingerprint match increased to 6 to 50 percent at adjacent sampling locations and 47 to 48 percent
at downriver sampling locations.
Fingerprint analysis indicated that mine leachate was migrating through the ground water and reaching
the Red River in sufficient quantity to alter the geochemical composition. An additional capability of the
FALCON technique is that the surface water to mine leachate dilution ratio can be estimated. As in the
sediment dioxin case study (Section 3.1), the two source fingerprints can be used to calculate the expected
composition for various mixtures of river water and mine leachate (i.e., 90 percent river + 10 percent
mine leachate, 80 percent river water + 20 percent mine leachate, etc.). The actual inorganic composition
at each river location can then be compared to each of the calculated mixtures to produce a range of
fingerprint match values as shown in Table 6. For example:
1. The sample from station RR4 produced a fingerprint match of 0.987 (99 percent) with the upriver
fingerprint. As the calculated amount of mine leachate in the mixture increased above 0 percent, the
fingerprint match decreased. Since the best match (maximum r2) occurred with the upriver
fingerprint, this location was considered to be 100 percent upriver water.
2. The sample from station RR12 produced a fingerprint match of 0.458 (46 percent) with the upriver
fingerprint. The fingerprint match increased to 0.772 (77 percent) for the 80 percent upriver + 20
percent mine leachate mixture and reached a maximum of 0.992 (99 percent) for the 60 percent
upriver + 40 percent mine leachate mixture. As the calculated amount of mine leachate in the mixture
increased above 40 percent, the fingerprint match decreased. This location was considered to be 60
percent upriver water and 40 percent mine leachate.
Each of the remaining samples was evaluated in a similar manner and the calculated fingerprint mixture
that produced the best fingerprint match is highlighted and bolded in Table 6. It should be noted that the
maximum fingerprint match for each river sample was 0.981 (98 percent), or higher.
As shown in Table 6 and summarized in Figure 9, the three upriver stations produced a best fingerprint
match for mixtures that contained 0 to 10 percent mine leachate. The stations adjacent to the mine site
produced the best fingerprint match for mixtures that contained 10 to 40 percent mine leachate and the
three downriver locations produced the best match for a mixture of 40 percent mine leachate and 60
percent river water. The fluctuations at the adjacent river stations listed in Table 6 (e.g., RR 11 has a
lower percentage of leachate than RR6) are due to runoff entering the river from unimpacted tributaries.
In this example, the FALCON technique was able to develop a characteristic fingerprint for the mine
leachate and differentiate mine leachate from background ground water conditions and surface water. In
addition, based on the properties of the fingerprint, it was possible to track the migration of mine leachate
through the ground water, into a nearby river, and develop a quantitative estimate of the mixing that was
occurring between river water and fugitive mine leachate.
21
-------
Table 6. Comparison of Red River Geochemical Fingerprints with Secondary Mine Leachate Fingerprints
River
Location
Up river
Adjacent
Downriver
Station
Number
RR4
RR2
RR1
RR3
RR5
RR6
RR9
RR11
RR8
RR7
RR10
RR12
RR13
RR14
RR15
RR16
Identified
River
Fingerprint
100% UR
0% ML
0.987
0.988
0.976
0.906
0.875
0.767
0.912
0.969
0.832
0.856
0.824
0.458
0.698
0.465
0.488
0.469
Calculated composition of surface water - mine leachate mixtures.
90% UR
10% ML
0.953
0.954
0.999
0.966
0.943
0.869
0.972
0.995
0.918
0.935
0.913
0.592
0.813
0.599
0.622
0.602
80% UR
20% ML
0.847
0.848
0.968
0.991
0.983
0.967
0.997
0.969
0.988
0.993
0.986
0.772
0.937
0.776
0.797
0.778
70% UR
30% ML
0.677
0.679
0.860
0.937
0.945
0.994
0.943
0.866
0.981
0.973
0.984
0.916
0.996
0.917
0.933
0.916
60% UR
40% ML
0.467
0.468
0.660
0.800
0.822
0.932
0.806
0.690
0.887
0.865
0.894
0.992
0.969
0.989
0.997
0.986
50% UR
50% ML
0.268
0.269
0.475
0.615
0.646
0.796
0.619
0.487
0.727
0.697
0.737
0.979
0.859
0.972
0.971
0.966
40% UR
60% ML
0.124
0.124
0.296
0.432
0.466
0.631
0.435
0.307
0.551
0.518
0.562
0.898
0.710
0.887
0.880
0.880
30% UR
70% ML
0.041
0.042
0.166
0.285
0.317
0.478
0.287
0.175
0.397
0.365
0.408
0.787
0.561
0.775
0.762
0.767
20% UR
80% ML
0.001
0.006
0.084
0.179
0.208
0.354
0.180
0.091
0.279
0.249
0.288
0.676
0.435
0.663
0.648
0.655
10% UR
90% ML
0.001
0.001
0.036
0.109
0.132
0.261
0.109
0.041
0.193
0.167
0.201
0.578
0.336
0.565
0.549
0.558
Identified
Leachate
Fingerprint
0% UR
100% ML
0.011
0.011
0.012
0.063
0.082
0.192
0.064
0.015
0.133
0.111
0.140
0.496
0.262
0.484
0.467
0.477
Tabulated results are regression analysis comparisons between river samples and the calculated surface water-mine leachate mixture.
UR = Upriver ML = Mine Leachate 0.9830 = maximum leachate matches
-------
(0
s-
0
Dl
0.20
0.15
0.10
*
I
0
0
i 0.05
(0
'5
0
Q
0.00
Landfill
D Salvage Yard
I 3 I 5 | 7 | 9 |i1|13|15|17|19|21|23|25|27|29|3'l|33|35|37|39|4'l|43| bUVI^Isi |S3
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52
PAH Fingerprint Constituents
PAH Fingerprint Analysis of Tributary Sediments
Station
T6
T3
T4
T5
T17
T15
S9
T9
T1
T2
T10
T14
S7
S8
S10
S11
Landfill
Match
Decimal
Percent
0.874
0.886
0.900
0.901
0.930
0.982
0.984
0.990
0.254
0.113
0.247
0.154
0.379
0.529
0.283
0.138
Salvage
Yard Match
Decimal
Percent
0.657
0.433
0.429
0.508
0.203
0.380
0.359
0.347
0.964
0.843
0.781
0.850
0.986
0.859
0.986
0.872
Total PAH
(ng/kg)
3951
3075003
99022
91739
4845764
4135
12256
14671
210573
23743530
1842
58481394
332628
354305
328377
174072
Figure 10. Fingerprint analysis of PAH contaminated sediments.
23
-------
3.5 Mixing of Similar Contaminant Plumes - PAH Case Study
Two potential sources of poly cyclic aromatic hydrocarbon (PAH) are separated by a small tributary that
drains into a river. One potential source is a landfill that included waste tar among its contents. The
second potential source was a salvage yard that processed crude tar wastes to recover creosote, phenol,
and other chemicals. Wastes from this salvage yard were disposed of on-site. Sediment sampling of the
tributary that runs between these two operations produced total PAH concentrations ranging from less
than 5 mg/kg to greater than 50,000 mg/kg.
The FALCON procedure was applied to a set of soil samples collected adjacent to the tributary. This
effort identified two PAH fingerprint patterns as shown in Figure 10. For brevity, the 53 standard PAH
compounds are not listed by name but simply referred to by compound number. One pattern was
characterized by a relatively dominant peak for PAH compound 33, smaller peaks for PAH compounds
39, 40, 45, 46, and 20, and the low abundance of PAH compounds 1 to 19. The distinguishing
characteristics of the second fingerprint pattern were the relatively large peaks for PAH compounds 1 and
20, several peaks for PAH compounds 2 to 10, and a relatively lower abundance of PAH compounds 39
to 53. The comparability of the two identified PAH fingerprints was only 0.329 (33 percent).
The PAH distribution in collected tributary sediment samples was compared to the two identified PAH
fingerprints. One set of samples (T6, T3, T4, T5, T17, T15, S9 and T9) produced strong matches (0.874
to 0.990) with the fingerprint relatively enriched in compounds 33 to 53 and a poor match (0.203 to
0.657) with the fingerprint enriched in compounds 1 to 19. The second set of samples (Tl, T2, T10, T14,
S7, S8, S10, and Sll) produced a strong match (0.781 to 0.986) with the fingerprint enriched in PAH
compounds 1 to 20 and a poor match (0.113 to 0.529) with the fingerprint enriched in PAH compounds
33 to 53. Based on the fingerprint assessment, tributary sediments closest to the bank had a low total
PAH concentration and a PAH fingerprint enriched with compounds 1 to 20. The mid-channel sediments
had a very high total PAH concentration (two orders of magnitude greater than the sediments closest to
the bank) and a PAH distribution relatively enriched in compounds 33 to 53.
4.0 FALCON Capabilities
FALCON is a flexible data analysis technique of combining data from two or more contaminants to
develop a distinctive chemical signature. The resultant FALCON fingerprint, based on the relative
abundance rather than actual concentrations of the individual contaminants, provides a mechanism to
identify the source and monitor the environmental behavior of fugitive emissions and leachates. These
source patterns provide a visual characterization of contaminants in liquid and solid matrices and
generally have a reproducibility of 90 to 99 percent.
The FALCON process can assist in the evaluation and interpretation of site characterization and
monitoring data in several ways:
1. The process produces a characteristic fingerprint that associates a contaminant with a particular
source. The data normalization process permits a direct visual comparison of fingerprint patterns
despite highly variable contaminant concentrations and the use of regression analysis provides a
statistical estimate of the reproducibility or comparability of two patterns. Distinctive fingerprints
can be developed to characterize organic and inorganic contamination due to spills, leaks, and landfill
leaching. In addition to the dioxin-furan, mine site, and PAH examples illustrated in this report, the
FALCON technique has also been used to characterize halogenated organic solvent spills, gasoline
and diesel fuel spills, and landfill leachates.
24
-------
2. The fingerprints can differentiate a source fingerprint from background conditions. As illustrated in
the gold mine (Section 3.2), copper mine (Section 3.3), and the molybdenum mine (Section 3.4)
examples, each source had a characteristic fingerprint that could be visually and statistically
differentiated from background ground water and upriver surface water conditions in the vicinity of
each facility.
3. FALCON fingerprints provide a mechanism to differentiate multiple sources of the same
contaminants. As illustrated in the dioxin example (Section 3.1) and the PAH example (Section 3.5),
distinctive fingerprints can be developed for each source.
4. FALCON fingerprints retain their chemical identity and act as an internal tracer as they migrate
through the environment. Therefore, fingerprint patterns at established monitoring locations can be
directly compared with the source fingerprint as demonstrated with the gold mine example (Section
3.2) to verify the source of detected contaminants and to map the areas that have been impacted by a
specific potentially responsible party.
5. FALCON fingerprints can also be used as a tool to characterize the environmental behavior of
contaminant plumes. Fingerprint analysis provided a quantitative estimate of the Roanoke sediment
contamination that can be specifically attributed to a pulp mill even though there are multiple sources
of dioxin-furan contamination to the river (Section 3.1). The molybdenum mine case study (Section
3.4) demonstrated that a source fingerprint could be tracked through the ground water and used to
estimate the rate of mixing between mine leachate and surface water. Thus, in addition to source
characterization, the fingerprints can be used to characterize the migration of contaminants between
environmental phases (ground water, surface water, and sediments) and to quantify the extent of
mixing or dilution that has occurred between two source fingerprints. This capability provides a
mechanism to apportion responsibility for environmental degradation between potentially responsible
parties.
6. Fingerprints also have a potential application as detection monitoring parameters to provide early
warning of leachate migration. This capability is illustrated with the Coppermine case study
(Section 3.4) in which the tailings pond fingerprint was initially detected and then fully developed
before the more hazardous constituents associated with mine leachate were detected downgradient
from the mine site. Fingerprint analysis would have provided an early warning to implement a
corrective action or remediation program while the developing problem was smaller and more
manageable. The factors that would permit the FALCON process to be used in this capacity are (a)
the fingerprint characterizes the source and differentiates it from background conditions, (b) the
fingerprint retains its chemical identity as it migrates away from the source, and (c) the fingerprint
migrates faster than the contaminants of higher regulatory concern.
7. Each of the case studies mentioned in this report were developed with routine monitoring data. It is
not necessary to use special analytical techniques, that may not produce data compatible with
historical site records, in order to use the FALCON technique.
The previous discussion focused on the use of contaminant fingerprinting to evaluate site-specific data.
The FALCON process can also provide the technical basis for the development of industry-specific
monitoring strategies. For example, a review of the gold mine (Section 3.2), copper mine (Section 3.3),
and molybdenum mine (Section 3.4) case studies reveals that the same small set of geochemical
parameters defined a characteristic, reproducible fingerprint at each mine as well as the ground water and
surface water in the vicinity of the mine. This observation has been verified by evaluating data from
more than 30 additional mines.1 These results suggest the possibility of a uniform two-phased monitoring
program at mining facilities based on the set of geochemical parameters that characterize mine leachates.
The first phase would utilize the fingerprint parameters to characterize and detect the leachate. As long as
25
-------
leachate is not detected, contaminants are not entering the ground water and the facility would remain at
this monitoring level. However, once the leachate fingerprint is detected, the second phase would be
triggered to more fully characterize the slower migrating contaminants that may be present. The factors
that support consideration of this approach are that the fingerprint is a reliable indicator of the leachate,
the fingerprint retains its identity as it migrates, and the fingerprint migrates faster than other leachate
constituents (as illustrated in the copper mine case study, Section 3.3). This approach could reduce
monitoring costs and provide a uniform monitoring strategy that would be easier to implement, evaluate,
and enforce.
5.0 Summary
FALCON is an empirical data assessment and visualization tool that produces contaminant fingerprint
patterns. This technique combines data from two or more parameters to produce visually distinctive and
reproducible fingerprints. The resultant fingerprints can be used to:
1. Characterize contaminants at their source,
2. Compare and evaluate background levels of contaminants with anthropogenic sources,
3. Establish an internal tracer to monitor the migration of a contaminant plume through ground water,
surface water, and sedimentary environments, and
4. Differentiate two sources of the same contaminant and estimate the relative mixing that has occurred
between two contaminant plumes.
Additional information on the FALCON procedure can be obtained from Gareth Pearson, Director of the
Technical Support Center, U.S. EPA National Environmental Research Laboratory, Las Vegas, Nevada
(pearson.gareth@epa.gov).
6.0 References
1. Plumb Jr., R. H. 1999. Characterization of Mine Leachates and the Development of a Ground-Water
Monitoring Strategy for Mine Sites. EPA/600/R-99/007. Office of Research and Development, U.S.
Environmental Protection Agency, Washington, D.C. February.
2. Plumb Jr., R. H. and Engelmann, W. H. 1991. Groundwater Monitoring Strategy for Incinerated
Municipal Waste Ash Monofills. HMC-Northeast '91 Conference Proceedings, Hazardous Materials
Control Research Institute, Greenbelt, Maryland, pp 353-359.
3. Plumb Jr., R. H. 1990. Development of a Ground-Water Monitoring Strategy for Incinerated
Municipal Waste Landfills. Report No. EPA-600/X-90/110. Environmental Monitoring Systems
Laboratory, U.S. Environmental Protection Agency, Las Vegas, Nevada. 55 p. May.
4. Plumb Jr., R. H. and Pitchford, A. M. 1985. Volatile Organic Scans: Implications for Ground Water
Monitoring. Proceedings Conference on Petroleum Hydrocarbons and Organic Chemicals in Ground
Water: Prevention, Detection, and Restoration. Pp 207-222. American Petroleum Institute/National
Water Well Association, Dublin, Ohio. November.
26
-------
5. Plumb Jr., R. H. and Fitzsimmons, C. K. 1984. Performance Evaluation of RCRA Indicator
Parameters. First Canadian/American Conference on Hydrogeology, pp 129-136. Banff, Alberta,
Canada. Alberta Research Council/National Water Well Association. June 22 - 26.
7.0 Notice
The U.S. Environmental Protection Agency (U.S. EPA) through its Office of Research and Development
(ORD) funded and managed in the research described here under assistance agreement number DW
47939416 with the Government Services Agency (GSA) to Lockheed Martin Environmental Services. It
has been subjected to the Agency's peer and administrative review and has been approved for publication
as an EPA document.
27
------- |