BREAK:
Beach, Richard OT1 ilk:
From: Beach, Richard Sent: Monday, April 15, 2002 12:51 PM To: Kevin Russell (E-mail) Subject: Modeling Mtg Followup
Kevin,
As we discussed this morning, please find attached the two presentations that Dick McGrath made last week.
When you get your final presentation files together, please forward them to myself and I will provide them to our team.
By the way, could you also provide Jim Rhea's presentation(s)? We would like to followup on the database inconsistencies, and would like to have a complete accounting for them.
Thanks in advance,
Rick Beach Roy F. Weston, Inc. 610-701-3473 610-701-3125 (fax)
The information contained in this e-mail message is intended only for the personal and confidential use of the recipient(s) named above. This message may be an attorney-client communication and as such is privileged and confidential. If the reader of this message is not the intended recipient or an agent responsible for delivering it to the intended recipient, you are hereby notified that you have received this document in error and that any review, dissemination, distribution, or copying of this message is strictly prohibited. If you have received this communication in error, please notify us immediately by e-mail, and delete the original message.
v/ /xttpR0r&-
Dups, Comps, & Splits: Comparison of EPA and GE
QA Sample Results
Dick McGrath Scott Campbell Sue Stefanosky
Kelly Spittler Roy F. Weston, Inc
April 11, 2002 Draft Confidential - Housatonic River Project
Originals in color.
& Introduction
Principal objective is to present EPA/GE split sample results Will also look at duplicate and "comparability" sample results Total PCB only Three questions of interest: - Magnitude of difference - Variability of difference - Indication of bias
Housatonic River Project pq 2
Originals in color.
Definitions - Soil/Sediment
[Field] Duplicate: Collected for internal QA purposes and analyzed (blind) by same laboratory Comparability Sample: Same procedure as Duplicate. Analyzed by fixed lab as QA check on field lab Split Sample: Aliquot collected following homogenization and provided to GE
Housatonic River Project pg 3
Originals in color.
\ Definitions - Surface Water
[Field] Duplicate: Collected at same time & location. Analyzed (blind) by same laboratory Comparability Sample: Not needed for water Split Sample: Separate container filled at same location and approximately same time and provided to GE (same collection procedure as field dup)
Housatonic River Project pq
Originals in color.
Definitions - Biota (Fish Fillet)
[Field] Duplicate: Second fillet from single specimen removed and analyzed (blind) by same laboratory
Comparability Sample: Not needed for biota
Split Sample: Second fillet from single specimen removed and provided to GE (same procedure as field dup)
Housatonic River
Originals in color.
\ / Sources of Variability
• Small-scale variability in field • Collection procedures • Homogenization effectiveness • Sample handling, storage • Extraction efficiency
• Laboratory instrumentation • Sample quantitation procedures • Error
Housatonic River Project pg 6
Originals in color.
Methods
Relative Percent Difference (RPD)
Chi-Square Test for Goodness of Fit
lousatonic River Project pg 7
Originals in color.
VXJ Relationship Between RPD
^PRO^ and Results Ratio 200%
9 11 13 15 17 19
Ratio Between Results Originals in color.
Housatonic River Project pg 8
\ / I Duplicates - All Media
Medium
Surface Water
Soils & Sediments
Biota Tissue
Total Samples
29
611
38
Total Mean Detects RPD
6 60.05%
428 34.71%
38 28.67%
Housatonic River Project pg 9
Originals in color.
Frequency DistributionSoil & Sediment Dups
140 T
Soils & Sediment Duplicates - RPD 120%
4- 100%
80%
60%
40%
20%
0 01 02 03 04 05 06 07 0.8 09 1 11 12 13 14 15 16 17 18 19 2 More
RPD/100 Housatonic River Project pg 10
Originals in color.
Frequency Distribution Tissue Dups
Tissue Duplicates - RPD
120%
100%
0 0.1 02 0.3 0.4 0.5 06 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 17 1.8 1.9 2 More
RPD/100 Originals in color.
Housatonic River Project pg 1
Comparability Samples Soils & Sediments
Sample Grouping
All data with both results "detect"
Grouped by Concentration: STL result < 7 ppm STL result > 7 ppm but <25 ppm STL result > 25 ppm
Grouped by Terrain Type: Bars and Terraces Floodplain Main Channel Banks SCOX
Sample Size
779
255 248 276
140 211 208 128 53
Percent STL>OS
63.03
49.41 67.74 71.38
60.71 59.24 64.42 68.75 71.70
Percent OS>STL
36.84
50.59 31.85 28.62
39.29 40.76 35.58 30.47 28.3
Mean RPD
59.96%
65.10% 49.79% 64.36%
55.87% 49.61% 75.38% 48.88% 73.88%
Chi-square
53.49
0.035 32.07 50.45
6.43 7.21 17.31 18.91 9.98
P
<.0001
ns <.0001 <.0001
0.01 0.007
<.0001 <.0001 0.002
Housatonic Rivei
Originals in color.
> Splits - Surface Water
N GE>EPA (%) EPA>GE (%) Mean RPD Chi-square P Surface Water Splits 190 Surface Water Splits Detects 75 54.67 44.00 87.02% 0.865 ns
Housatonic River
Originals in color.
Frequency Distribution Surface Water Splits
Surface Water Split Samples RPD
T 120%
• 100%
01
More
RPD/100 Originals in color.
Housatonic River Project pg 14
Soils & Sediments
Sample Grouping
Total N of Soil & Sediment Splits Both results >DL
Grouped by Terrain Type Bars and Terraces Main Channel Floodplains Pond Sediment River banks
Grouped by River Reach Reach 4 Reach 5 Reach 6
N
805 562
53 182 73 52 139
106 179 39
GE>EPA (%)
71.53
64.15 68.68 73.97 80.77 73.38
74.53 75.98 82.05
EPA>GE (%)
28.47
35.85 31.32 26.03 19.23 26.62
25.47 24.02 17.95
Mean RPD
64.29%
53.98% 77.36% 39.60% 67.44% 65.10%
85.81% 54.90% 71.13%
Chi-square
312.30
106.98 10.95 76.93 3.53 0.43
17.49 4.08
401.28
P
<.0001
<.0001 <.001 <.0001 0.06 ns
<.0001 <.043 <.0001
Housatonic River Project pg 15
Originals in color.
Frequency Distribution Soil & Sediment Splits
Soils & Sediment Splits - RPD 120%
70 r
100%
80%
60%
40%
20%
More
Originals in color. RPD/100
Housatonic River Project pg 16
Soil & Sediment Splits Fixed Labs Only (both>DL)
Sample Grouping N GE>EPA (%) EPA>GE (%) Mean RPD Chi -square P
Soil/Sed Splits - Fixed Labs Only 127 66.14 33.86 58.33% 13.23 <.001
Soil/Sed Splits - All Results 562 71.53 28.47 64.29% 312.30 <.0001
Housatonic River Project pg 17
Originals in color.
Frequency Distribution Soil & Sediment FL Splits
Soils & Sediments Splits (Fixed Labs) 120%
• 100%
- 80%
- 60%
- 40%
20%
04 0.2 0.5 07 08 0.1 0.3 12 06 1 09 1 1 16 13 14 18 19 15 0 17 2 More
RPD/100 Originals in color.
Housatonic River Project pg 18
v / Magnitude of Differences Soils & Sediments
Geometric Mean of Geometric Mean of GE Concentration EPA Concentration Correlation Percent
Sample Grouping N Percent GE > EPA (ppm) (ppm) Coefficient Difference
Total Soil & Sediment Splits 805 3.26 3.04 0.89 7.01% Both results >DL 562 71.53 10.15 6.92 0.89 46.56%
Grouped by Terrain Bars and Terraces 53 64.15 19.32 15.35 0.31 25.90% Main Channel 182 68.68 7.85 5.56 0.70 41.08% Floodplains 73 73.97 3.32 2.73 0.93 21.37% Pond Sediment 52 80.77 9.03 4.91 0.91 83.98% Rirer banks 139 73.38 17.60 10.63 0.46 65.60%
Grouped by Reach Reach 4 106 74.53 10.12 6.29 0.84 60.72% Reach 5 179 75.98 11.32 7.76 0.53 45.90% Reach 6 39 82.05 21.76 9.64 0.69 125.80%
Originals in color,
Comparison to Other PCB Field Investigations
Tetra Tech PSARA Result Result Difference RPD ABS RPD
2.1 2.7 0.6 -25.00% 25.00% 5.8 5.1 -0.7 12.84% 12.84% 2.7 4.3 1.6 -45.71% 45.71% 8.7 9.1 0.4 -4.49% 4.49% 160 2300 2140 -173.98% 173.98% 60 520 460 -158.62% 158.62%
280 440 160 -44.44% 44.44% 13 12 -1 8.00% 8.00% 2.7 25 22.3 -161.01% 161.01% 11 31 20 -95.24% 95.24% 3 4 1 -28.57% 28.57%
• •• ••• ^ •• ^ ^ •i ^H • ^ •• ••• ••B •••• •••• •••• H
Mean 68.90%
FROM: Indiana Superfund Site(s) 2001 Split Sample Analytical Results Housatonic River Project pg 20
Originals in color.
Splits - Tissue (Fish Fillet)
N GE>EPA (%) EPA>GE {%) Mean RPD Chi -square P
Fish Tissue 18 7 11 41.48% 0.889 ns
Housatonic River
Originals in color.
Frequency Distribution Tissue Splits
Tissue Samples - RPD
120%
5 - 100%
- 80%
O
0)3 3 o£ u.
2
• 60%
- 40%
1 - 20%
0 02 04 06 08 1 12 14 16 18 2 More 0%
RPD/100 f Originals in color.l
Housatonic River Project pg 22
Summary of Split Sample Analysis
Numerous sources of variability expected to produce different results Values reported appear to be reasonable and within expected ranges GE reported concentrations often higher than EPA results - difference is statistically significant for soils/seds
Housatonic River Project pg 23
Originals in color.
Review of Changes to Soil and Sediment TOC Data
Scott Campbell Dick McGrath
Roy F. Weston, Inc. April 11, 2002
Housatonic River Project
Originals in color.
v Initial Discovery
Routine database audit: - Out-of-range data
- Suspect data points
- Follow-up action
TOC values greater than 30% - Review of hardcopy reports - Laboratory contacted
Originals in color.
A..;/,
TOO Methods
Initial instrumentation and method - Fall of 1998 to late April 1999 - Ol analytical 1010 NDIR TOO Detection System
- 500 mg sample output as mg of TOC - Concentrations converted to dry-wt TOC -Dilutions of 2X and 5X
- Approximately 1630 samples analyzed - Pre-drying step initiated
Housatonic River
Originals in color.
^ TOC Methods (Concluded) • New Method
- Late April 1999 to August 2000
- CE instruments NC 2500 elemental GC- TCD analyzer
- 25 mg of sample output as %Carbon
- Pre-dried sample analyzed - Approximately 2400 samples analyzed
Housatonic River Project pg 4
Originals in color.
^ Incorporating a Dilution Factor to Initial Method
Correct formula - TOCS = (1000 * TOG, / 0.5) / S / D
- Dilution Factor of 0.5 or 0.2 (fractional) - Some sample batches had integer DFs
incorrectly entered in the denominator - Dividing by the integer value had the effect of
decreasing the TOC value
Housatonic River Project pg 5
Originals in color.
j Incorporating a DF to New Method
Correct Formula -TOCS =(1000*TOCC/(0.025*S))*D
- DF calculated in the numerator
- Effectively increases the TOC result
- DF as an integer is appropriate - Incorrect Formula -TOCS =(1000*TOCC/(0.025*S))*D
• Parenthesis preceding 0.025 omitted
• Expression evaluated from left to right
Housatonic River Project pg 6
Originals in color.
., Effects on Results from Calculation Modification
Dry weight of 100% introduces no error Dry weight decrease introduces a induced error that becomes larger in a non-linear fashion A 95% solids values induces a 10.8% decrease in reported TOC content At 80% solids the calculated TOC is less than 75% of the true value
Housatonic River Project pg
Originals in color.
, Calculation Effects by Method
Initial Method - Two different calculations
- Results potentially have either calculation error
New Method - Results can only have error involving percent
solids calculation
Housatonic River Project pg 8
Originals in color.
J
Corrective Actions
Identification of all TOC Sample Delivery Groups (SDG) TOC formula reviewed for each SDG SDG separated into three groups
Unaffected Percent solid miscalculation Dilution factor miscalculation
Housatonic River Project pg 9
color.
I
Summary TOC Changes
4,185 potentially effected TOC results - 340 required no change
- Approximately 2,000 required < 5% change
- 500 results more than doubled
-Approximately 250 required 5X change or greater
Housatonic River Project pg 10
Originals in color.
, Database TOG Changes by Reach
Reach 1 and 2 Combined - Nearly 30% of results with less than 5% change - Approximately 18% of results changed by a
factor of 2X or more
Reach 3 - No samples analyzed for TOC by these
methods
Housatonic River Project pg 11
Originals in color.
, Database TOC Changes by Reach (Continued)
Reach 4 - Nearly 80% of results less than 5% increase - Approximately 8% of results with 2X or greater
corrections Reach 5 - Nearly 67% of results less than 50% increase - Approximately 8% of results with 5X or greater
corrections Reach 6 - Similar to overall pattern with corrections
greater than 5% but less than 2X
Housatonic River Project pg 12
Originals in color.
, Database TOC Changes by Reach (Concluded)
Reach 7 (including Rising Pond) - Nearly 90% increased by less than 5%
Reach 9 - No sample requiring greater than a 10%
increase
Reference Locations - Majority of samples increased by more than 5%
- Nearly a third of samples more than tripled in value
Housatonic River Project pg 13
Originals in color,
, Database TOC Changes by Selected Terrain
Floodplain Samples - Nearly 25% increased by a factor of 5X - More than half of the samples more than
doubling - One third of results increased by 5% or less
Main Channel Samples - Nearly half of the samples changing by less
than 5% - Approximately 30% of sample increased by a
factor of 2X or more
Housatonic River Project pg 14
Originals in color.
j Database TOC Changes by Selected Terrain (concluded)
SCOX Samples (Reach 5) - Greatest changes observed - Less than 25% of results increased by 5% - Nearly 60% of results increased by 2X or more - One third of results increase by a factor of 5X or more
Woods Pond Samples - More than 20% of samples had no change - Approximately one-third of samples doubling - Nearly 20% of samples increased by a factor of 5X or
more
Housatonic River Project pg 15
Originals in color.