Quality Assurance and Quality Control Summary Report

Assessing DQI Goals

The critical measurements for the natural weathering tests are total
arsenic, total chromium, and total copper concentrations. Data quality
indicator (DQI) goals for concentration in terms of accuracy, precision,
and completeness, as established in the QAPP for this project, are shown
in Table 4-1. 

Table 4-1.	Data Quality Indicator Goals for Critical Measurements

Analyte	Method	Accuracy (%Recovery)	Precision (%RSD/RPD)	Completeness
(%)

Arsenic (total)	SW-846 Method 6020 (modified)	90-110	10	90

Chromium (total)	SW-846 Method 6020 (modified)	90-110	10	90

Copper (total)	SW-846 Method 6020 (modified)	90-110	10	90



After reviewing sample results, the DQI goals for precision and accuracy
have been revised for concentrations <10 µg/L. Acceptance criteria of
±25% RPD for precision between duplicates and 75-125% recovery for
accuracy will be used for concentrations <10 µg/L.

Precision 

In order to evaluate the precision of a measurement, it is necessary to
make at least duplicate measurements of a relatively unchanging
parameter. Precision can then be expressed as the RPD of the duplicated
measurement. RPD is calculated using Equation 4.1, where Y1 is the
concentration of the first sample and Y2 is the concentration of the
duplicate sample.

 	(Equation 4.1)

A large number of blind field duplicates (wipe samples split following
extraction) were performed and delivered to the laboratory for analysis.
These duplicates were performed at a rate of 7% of the total number of
samples collected and provide an indication of the repeatability of the
analytical method. The DQI goal for precision was set at ±10% RPD. For
the majority of samples, agreement between field duplicates was very
good (i.e., the RPD was small). The DQI goal of ±10% was increased to
±25% for samples with concentrations <10 µg/L because smaller
differences in lower concentrations have a greater effect on RPD. With
the modification, there were still some cases where the DQI goal was
slightly exceeded. In only one set of duplicate samples was RPD >50% for
each analyte; this was the only sample set where the data was qualified
as estimated “J” due to the RPD. A summary of duplicate results is
shown in Table 4-2. Instances where these revised DQI goals were not met
are shown in bold. Completeness summaries for each metal using the
revised DQI goals are shown at the bottom of the table. Achieved
completeness was >80% for all three analytes, which did not meet the DQI
goal of 90% established in the QAPP. There are no acceptance criteria
given in the analytical method for agreement between duplicate samples.
A DQI goal of ±15% RPD may be more realistic for these types of
samples. If DQI goals of ±15% RPD, with ±25% RPD for duplicate
measurements where the mean was <10 µg/L, are used, completeness would
be greater than 90% for all three analytes. All data were used in the
analysis.

Accuracy and Bias

For this project, the accuracy of the measurement is expressed in terms
of recovery of a known spike. Recovery is calculated by Equation 4.2,
where R is the measured concentration and C is the spiked (i.e., target)
concentration.

 	(Equation 4.2)

Spiked samples were performed by the laboratory at a minimum of 3 of 4
concentration levels with the samples from each sampling event. Those
concentration levels were 10,000 µg/L, 1,000 µg/L, 50 µg/L and 1
µg/L. In addition to the laboratory spikes, ARCADIS/EPA provided 1000
µg/L spikes (in triplicate) and submitted these blind to the
laboratory. Spike results are summarized in Table 4-3. Results that do
not meet the DQI goals of 90-110% recovery are indicated in bold (with
the exception of the 1 µg/L spikes). The DQI recovery goal for the 1
µg/L spike sample was increased to 75-125% for the same reason the
precision DQI was increased, as discussed in Section 4.1.1. Completeness
was calculated  for each spiking level and each analyte and is shown at
the end of the summary for each spike level. Bias results generally
improve as the concentration of the spike level is increased, although
there were less spikes performed at the 10,000 µg/L level. In an effort
to improve completeness results, sample groups for the 15-, 20- and
24-month sampling events utilized a higher rate of spike samples than
did the preceding events. 

Table 4-2.	Precision of Duplicate Samples

Sample/Duplicate (Group)	As (µg/L)	RPD	Cr (µg/L)	RPD	Cu (µg/L)	RPD

SS-859/626 (WIPE-05)	3.6/4.0	10.5	5.7/5.7	0	23/23	0

SS-860/633	490/490	0	690/690	0	190/190	0

SS-861/635	5.9/5.4	8.8	10/9.2	8.3	40/37	7.8

SS-862/647	300/270	10.5	390/370	5.3	190/180	5.4

SS-863/656	34/36	5.7	8.1/8.8	8.3	31/33	6.3

SS-864/666	73/74	1.4	17/18	5.7	40/40	0

SS-865/674	120/140	15.4	220/260	16.7	130/150	14.3

SS-866/687	150/150	0	260/260	0	180/180	0

SS-867/690	810/810	0	1700/1700	0	660/660	0

SS-868/700	83/83	0	120/120	0	74/73	1.4

SS-872/712 (WIPE-06)	72/81	11.8	120/140	15.4	87/98	11.9

SS-873/714	290/300	3.4	410/410	0	200/200	0

SS-874/718	58/56	3.5	27/26	3.8	71/69	2.9

SS-875/719	360/340	5.7	110/100	9.5	110/110	0

SS-876/720	190/190	0	210/210	0	140/150	6.9

SS-877/721	130/120	8	140/140	0	92/89	3.3

SS-878/724	460/440	4.4	150/150	0	160/150	6.5

SS-879/727	8.9/20	76.8	2.2/5.3	82.7	13/31	81.8

SS-880/730	5.1/5.1	0	5.0/5.2	3.9	59/59	0

SS-881/734	3.9/4.1	5	4.7/4.8	2.1	62/64	3.2

SS-886/837 (WIPE-07)	160/170	6.1	300/310	3.3	99/100	1

SS-887/849	69/69	0	46/45	2.2	73/71	2.8

SS-888/852	3/2.8	6.9	8.7/8.4	3.5	35/33	5.9

SS-885/824	1.7/1.7	0	2.6/2.6	0	51/53	3.8

SS-1274/1071 (WIPE-08)	880/890	1.1	1100/1100	0	190/190	0

SS-1275/1084	25/26	3.9	7.3/7.6	4	66/71	7.3

SS-1276/1090	200/200	0	240/250	4.1	110/110	0

SS-1277/1091	180/190	5.4	310/320	3.2	170/180	5.7

SS-1278/1107	150/150	0	140/140	0	220/230	4.4

SS-1279/1116	6/5.3	02.4	11/10	9.5	150/150	0

SS-1280/1126	130/130	0	150/160	6.5	280/300	6.9

SS-1281/1131	150/150	0	350/340	2.9	250/260	3.9

SS-1282/1149	530/530	0	830/840	1.2	390/400	2.5

SS-1283/1153	700/680	2.9	1100/1100	0	490/500	2

SS-1284/1160	<0.5/<0.5	0	<0.5/<0.5	0	1.5/1.8	18.2

SS-1285/1161	35/33	5.9	39/37	5.3	130/130	0

SS-1286/1171	8.3/7.9	4.9	13/12	8	140/140	0

SS-1287/1172	2.5/2.8	11.3	2.6/2.9	10.9	110/120	8.7

SS-1288/1173	2.5/2.6	3.9	8.3/8.3	0	170/180	5.7

SS-1289/1179	210/210	0	550/570	3.6	220/250	12.8

SS-1290/1180	<0.5/<0.5	0	0.95/0.84	12.3	2.3/2.9	23.1

SS-1291/1181	3.7/3.6	2.7	5.7/5.7	0	120/130	8

SS-1292/1208	39/37	5.3	55/54	1.8	93/93	0

SS-1293/1210	30/30	0	57/58	1.7	77/82	6.3

SS-1294/1215	19/19	0	17/16	6.1	120/120	0

SS-1295/1222	3/3	0	6.2/6.6	6.2	170/170	0

SS-1296/1223	180/190	5.4	290/310	6.7	140/140	0

SS-1297/1230	370/350	5.6	740/700	5.6	300/300	0

SS-1298/1234	120/110	8.7	230/220	4.4	160/150	6.5

SS-1299/1247	41/44	7.1	41/46	11.5	80/87	8.4

SS-1300/1258	18/17	5.7	13/12	8	110/100	9.5

SS-1587/1324 (WIPE-10)	9.2/8.0	14	15/13	14.3	90/81	10.5

SS-1588/1333	240/240	0	440/440	0	230/220	4.4

SS-1589/1344	53/48	9.9	27/24	11.8	100/91	9.4

SS-1590/1348	77/72	6.7	48/44	8.7	210/190	10

SS-1591/1352	150/140	6.9	30/26	14.3	41/36	13

SS-1592/1371	92/91	1.1	14/14	0	36/33	8.7

SS-1593/1376	140/140	0	41/39	5	81/72	11.8

SS-1594/1383	2300/2300	0	2700/2600	3.8	390/360	8

SS-1595/1386	12/11	8.7	11/9.2	17.8	72/67	7.2

SS-1596/1407	730/700	4.2	100/950	5.1	490/430	13

SS-1597/1415	100/98	2	160/160	0	140/130	7.4

SS-1598/1479	480/490	2.1	630/640	1.6	350/360	2.8

SS-1600/1487	77/73	5.3	170/160	6.1	150/140	6.9

SS-1601/1495	470/480	2.1	560/570	1.8	230/240	4.3

SS-1602/1502	450/490	8.5	1100/1260	13.6	420/460	9.1

SS-1603/1512	110/110	0	240/240	0	160/160	0

SS-1604/1514	31/29	6.7	87/81	7.1	190/180	5.4

SS-1605/1519	59/65	9.7	270/290	7.1	140/150	6.9

SS-1606/1529	99/100	1	22/22	0	32/33	3.1

SS-1607/1531	22/23	4.4	6.5/6.9	6	33/34	3

SS-1608/1534	2.6/2.4	8	5.3/5.0	5.8	83/82	1.2

SS-1609/1536	2.2/2.1	4.7	8.3/7.8	6.2	91/88	3.4

SS-1610/1540	<1.0/<1.0	0	<1.0/<1.0	0	2.4/3.0	22.2

SS-1611/1566	7.7/7.6	1.3	9.6/9.2	4.3	94/88	6.6

SS-1612/1568	130/120	8	49/43	13	140/120	15.4

SS-1613/1576	280/300	6.9	73/80	9.2	140/160	13.3

SS-1614/1578	56/54	3.6	57/56	1.8	170/160	6.1

WB-278/893 (WIPE-11)	8900/9400	5.5	8800/9100	3.4	5300/5300	0

WB-348/894	11000/9700	12.6	12000/9800	20.2	7500/6300	17.4

WB-353/895	6200/6000	3.3	7100/6600	7.3	4400/4000	9.5

WB-373/896	8500/8600	1.2	7800/8000	2.5	5500/5500	0

WB-383/897	8800/8600	2.3	8500/8300	2.4	4600/4400	4.4

WB-394/898	8900/10000	11.6	8500/9600	12.2	4900/5400	9.7

WB-409/899	600/600	0	680/680	0	410/380	7.6

WB-445/900	5300/5500	3.7	6700/6900	2.9	4500/4600	2.2

WB-445/901	7500/7700	2.6	7400/7700	4	4500/4600	2.2

WB-449/902	2200/2300	4.4	3000/3100	3.3	1600/1700	6.1

WB-443/903	11000/8400	26.8	11000/8100	30.4	7400/5400	31.3

WB-437/904	1800/1200	40	1900/1200	45.2	1300/860	40.7

WB-260/905	5400/3500	42.7	6600/6500	1.5	3600/5500	41.8

WB-196/906	4000/4300	7.2	5100/5400	5.7	2700/2700	0

WB-115/907	4900/4900	0	4800/4700	2.1	2800/2600	7.4

WB-118/908	5800/5200	10.9	7000/6000	15.4	4100/3400	18.7

WB-149/909	6600/6700	1.5	8400/8400	0	5200/6100	15.9

WB-173/910	5900/6500	9.7	7200/7700	6.7	4200/4300	2.4

WB-239/911	10000/9800	2	11000/10000	9.5	7000/6200	12.1

WB-255/912	12000/11000	8.7	13000/12000	8	7200/6200	14.9

WB-146/920 (WIPE-12)	130/91	35.3	140/95	38.3	110/77	35.3

WB-154/921	4200/4700	11.2	4800/5500	13.6	2900/3200	9.8

WB-215/922	5900/5000	16.5	6600/5600	16.4	3500/3000	15.4

SS-1884/1842 (WIPE-13)	30/28	6.9	14/14	0	29/27	7.1

SS-1885/1713	33/31	6.3	30/38	20.3	540/480	11.8

SS-1886/1727	72/78	8	120/130	8	340/370	8.5

SS-1887/1815	1300/1200	8	2200/2100	4.7	820/780	5

SS-1888/1851	590/520	12.6	1500/1300	14.3	1300/1100	16.7

SS-1889/1797	7.5/7.6	1.3	7.3/7.4	1.4	140/140	0

SS-1890/1671	870/760	13.5	1800/1700	5.7	650/610	6.3

SS-1891/1653	1000/940	6.2	1760/1600	9.5	950/940	1.1

SS-1892/1624	360/3440	5.7	420/460	9.1	260/250	3.9

SS-1893/1772	4000/3700	7.8	5600/5200	7.4	2200/2100	4.7

SS-1894/1749	85/75	12.5	120/110	8.7	170/180	5.7

SS-1895/1869	310/350	12.1	230/250	8.3	330/340	3

SS-1896/1882	140/140	0	270/290	7.1	330/340	3

SS-1897/1641	14/14	0	20/21	4.9	310/290	6.7

SS-2180/2062 (WIPE-14)	1070/1100	2.8	1040/1060	1.9	238/241	1.3

SS-2181/2145	621/648	4.3	516/554	7.1	465/492	5.6

SS-2182/2048	2120/2130	0.5	2480/2620	5.5	407/422	3.6

SS-2183/2148	23.7/23.6	0.4	13.7/15.6	13	246/252	2.4

SS-2184/2072	1.9/2.3	19	2.2/3.1	34	125/130	3.9

SS-2185/2026	16/17	6.1	18.3/18.6	1.6	350/355	1.4

SS-2186/2136	1.2/2.5	70.3	2.6/2.9	10.9	110/118	7.0

SS-2187/2028	461/462	0.2	287/290	1	130/128	1.6

SS-2188/1978	5750/6100	5.9	6620/6660	0.6	727/726	0.1

SS-2189/1930	152/156	2.6	188/199	5.7	92.7/95.9	3.4

SS-2458/2219 (WIPE-2700)	5.7/6.3	10	4.4/4.4	0	46/49	6.3

SS-2459/2250	8.2/7.6	7.6	23/20	14	350/320	9

SS-2460/2291	29/28	3.5	27/26	3.8	340/330	3

SS-2461/2329	690/670	2.6	1100/1000	9.5	880/820	7.1

SS-2462/2378	6.2/7	12.1	17/20	16.2	240/270	11.8

SS-2463/2382	450/410	9.3	510/450	12.5	480/420	11.8

SS-2464/2409	1/1	0	1.6/1.8	11.8	60/63	4.9

SS-2465/2412	1/1	0	1.9/1.7	11.1	69/66	4.4

SS-2466/2426	2000/2000	0	2100/2100	0	510/520	1.9

SS-2467/2445	310/290	6.7	220/210	4.7	130/120	8

SS-2750/2478 (WIPE-7269)	3100/3000	3.3	3900/3500	10.8	1300/1300	0

SS-2752/2518	210/230	9.1	220/220	0	200/200	0

SS-2753/2692	17/17	0	17/17	0	280/290	3.5

SS-2754/2575	2800/2800	0	3000/2800	6.9	1200/1200	0

SS-2756/2546	11/12	8.7	5.8/5.6	3.5	56/56	0

SS-2758/2576	1500/1500	0	2200/2100	4.7	1900/1900	0

SS-2760/2608	16/16	0	29/30	3.4	410/430	4.8

SS-2761/2716	540/540	0	1200/1200	0	880/890	1.1

SS-2762/2644	220/250	12.8	406/520	12.2	630/700	10.5

SS-2764/2668	2400/2800	15.4	2900/3400	15.9	950/1100	14.6

Completeness

Using DQI of ±10%

Using DQI of ±15%	

124/145 = 85.5%

136/145= 93.8%	

117/145 = 80.7%

131/145= 90.3%	

119/145 = 82.1%

134/145= 92.4%



Table 4-3.	Spike Results

Analytical Group	Arsenic	Chromium	Copper

	Obtained	%Recovery	Obtained	%Recovery	Obtained	%Recovery

10,000 µg/L Spike Level (DQI Goal 90-110%)

WIPE-10	9700	97	10000	100	9700	97

WIPE-13	10000	100	12000	120	10000	100

WIPE-14	10000	100	11200	112	10000	100

WIPE-14 (duplicate)	10000	100	11200	112	10100	101

WIPE-2700	8700	87	8900	89	8900	89

WIPE-2700 (duplicate)	9500	95	9800	98	9900	99

WIPE-7269	8800	88	9100	91	9200	92

WIPE-7269 (duplicate)	9400	94	10000	100	9900	99

Completeness	75%

50%

87.5%

1,000 µg/L Spike Level (DQI Goal 90-110%)

WIPE-05	1100	110	1000	100	980	98

WIPE-06	1000	100	980	98	920	92

WIPE-07	1100	110	1000	100	950	95

WIPE-08	980	98	1000	100	980	98

WIPE-08*	970	97	1000	100	970	97

WIPE-08*	940	94	1000	100	950	95

WIPE-08*	980	98	1000	100	970	97

WIPE 11	1100	110	1000	100	930	93

WIPE-11 (duplicate)	1100	110	1000	100	980	98

WIPE-12	1100	110	990	99	970	97

WIPE-13	1000	100	1100	110	990	99

WIPE-14	1070	107	1130	113	1060	106

WIPE-14 (duplicate)	874	87	1120	112	1020	102

WIPE-2700	980	98	990	99	970	97

WIPE-2700 (duplicate)	970	97	990	99	980	98

WIPE-7269	930	93	920	92	920	92

WIPE-7269 (duplicate)	940	94	920	92	920	92

Completeness	94%

88%

100%

50 µg/L Spike Level (DQI Goal 90-110%)

WIPE-05	50	100	52	104	49	98

WIPE-06	58	116	52	104	51	102

WIPE-07	45	90	47	94	43	114

WIPE-08	46	92	50	100	47	94

WIPE-10	52	104	49	98	48	96

WIPE 11	51	102	53	106	51	102

WIPE-11 (duplicate)	54	108	51	102	50	100

WIPE-12	54	108	52	104	50	100

WIPE-13	64	128	57	114	59	118

*WIPE-14	5	10	8.3	17	6.3	13

WIPE-14 (duplicate)	46.7	93	54.5	109	52.1	104

WIPE-2700	55	110	57	114	58	116

WIPE-2700 (duplicate)	50	100	51	102	51	102

Completeness	83.3%

83.3%

75.0%

1 µg/L Spike Level (DQI Goal 75-125%)

WIPE-05	2.3	230	3.4	340	4.6	460

WIPE-06	0.95	95	0.9	90	1.2	120

WIPE-07	0.97	97	0.84	84	1.3	130

WIPE-08	1.0	100	1.0	100	1.2	120

WIPE-10	1.1	110	<1.0	100	<1.0	100

WIPE 11	1.2	120	1.1	110	1.2	120

WIPE-11 (duplicate)	0.5	50	0.5	50	0.5	50

WIPE-12	1.1	110	0.96	96	1.3	130

WIPE-13	1.3	130	1.0	100	1.0	100

WIPE-14	<1.0	100	<1.0	100	2.5	280

WIPE-14 (duplicate)	1.3	130	3	300	2.7	270

WIPE-2700	1	100	1	100	1	100

WIPE-2700 (duplicate)	1	100	1	100	1	100

Completeness	69.2%

76.9%

53.8%

* This sample was incorrectly spiked and thus not used in the
determination of completeness.

Completeness

The ratio of the number of data points taken that met DQI goals to the
total number of data points planned is defined as data completeness.
Completeness goals of >90% were not achieved for a number of
measurements. These are summarized in Tables 4-2 and 4-3. Data results
suggest that the DQI goal of ±10% for precision between duplicates may
have been too ambitious. There is no specific acceptance criteria given
in the analytical method for precision between duplicate samples. If the
DQI goal were slightly increased to 15% RPD, completeness goals would
have been met for all metals. Also, the analytical method cites
acceptance criteria for recovery of spiked blanks as 85-115% which is
slightly higher than the DQI goal of 90-110%. Using this criteria,
completeness of accuracy results would improve. DQI goals should be
reviewed and revised as appropriate for future studies.

Quality Control Checks

A variety of control samples were taken as described in Section 4.1-4.4
of the QAPP (Appendix A) and Section 2.13.4 of this report. Results from
the checks are described in the following subsections. 

Blanks during Sampling Events

During actual sampling events, two types of blank samples were submitted
by ARCADIS/EPA to the laboratory for analysis. Blind field blanks were
submitted at a frequency of one every 20 samples and consisted of the
extraction fluid from a clean wipe (i.e., not wiped across a board).
Reagent blanks were also submitted at a frequency of one per sample
delivery group and consisted of an aliquot of the nitric acid reagent
used for extracting samples. The majority of results from the blind
field blank samples resulted in non-detects for each metal of interest.
When there were detects (>1µg/L), concentrations were always < 10 µg/L
and always less than the action levels. In some cases, samples with
reported concentrations of the metals less than the action level
associated with these field blanks were qualified as not detected and
flagged “U”. These results are summarized in the individual
validation reports included in Appendix R. None of the metals of
interest were detected in any of the reagent blanks submitted to the
laboratory for analysis. No blank corrections were performed on reported
sample concentrations.

Laboratory Control Samples

A series of laboratory control samples were analyzed with each batch of
samples as defined by the analytical method (EPA Method 200.8). Internal
laboratory QC checks included laboratory reagent blanks at a frequency
of one every 20 samples, laboratory fortified blanks at a frequency of
one per batch and post-digestion spikes which were performed at a
frequency of one every 10 samples. Results of laboratory control samples
were summarized in the analytical reports and also in the internal
validations performed by ARCADIS/EPA. Failure to meet acceptance
criteria resulted in data flagged as estimated “J”, or non-detects,
“U”. 

Data Validation Summary

The subcontract laboratory was required to submit calibration and QC
data along with each data package. All data packages received by
ARCADIS/EPA were internally validated by a qualified staff scientist
according to the QA/QC criteria set forth in the U.S. EPA Contract
Laboratory Program National Functional Guidelines for Inorganic Data
Review, July 2002 (NFG). When parameters called out in the NFG were
different from those established in the QAPP or the analytical method
(EPA Method 200.8), the more stringent criteria were used. Validation
reports were prepared for each sample delivery group and the reported
data were qualified as appropriate. These reports are included in
Appendix R.

Deviations from the QAPP

Deviations from the original QAPP and the reasons or justification for
them are listed below:

Laboratory cites use of Method 200.8-Determination of Trace Elements in
Waters and Wastes by Inductively Coupled Plasma/Mass Spectrometry
instead of SW-846 Method 6020 as referenced in the QAPP. Method 200.8 is
a more detailed and specific method than 6020 and there are no technical
differences between the two.

Field blanks at a rate of 5% were not included in the first two batches
of samples sent to the laboratory as stated in the QAPP Section 4.4.
This was a mistake by the sampling team. Subsequent batches contained
the appropriate number (or more) of field blanks. Based on the results
of subsequent field blanks, it is not believed that the first batches of
samples were compromised due to insufficient number for field blanks.

Blind spiked samples were not submitted to the laboratory at the
frequency described in Section 4.4 of the QAPP. Only one set of
triplicate spikes at 1,000 µg/L were submitted. There are a number of
laboratory control samples required by the method that are performed
with each batch. It was determined that one set of blind spiked samples
was sufficient.

The laboratory performed post-digestion spikes at a 10% frequency, but
did not do standard matrix spikes and matrix spike duplicates, since
ARCADIS/EPA did the sample extraction and digestion.

The wash technique used to prequalify the coating brushes (described in
Section 2.6) was slightly different than that specified in the QAPP.
Specifically, the brushes were washed in 40 ml of DI water instead of
200 ml. Additionally, the wash water was digested in a manner similar to
that used for preparing wipe samples for analysis. The QAPP did not
specify any digestion of the wash water. 

The technique used to digest raw coating samples (described in Section
2.10) was modified slightly as follows:

Hydrofluoric acid was used to digest the coatings in addition to the
nitric acid.

The HF had to be neutralized before analysis using boric acid.

Samples were brought up to 100 ml volume with 10% nitric before
analysis.

Wood moisture content was not measured during the wipe events, because
of concerns about damaging and compromising coatings using the wood
moisture probe.

Audits

The ARCADIS QA Officer performed an internal technical systems audit on
the sampling portion of this project on September 10 and 11, 2003. The
following findings were reported to the ARCADIS Work Assignment (WA)
Manager. All problems found during the audit were corrected the same
day.

Sampling staff were not wearing gloves on the first day of sampling.
Corrective actions were taken and for all subsequent sampling the
sampling staff used double gloves and changed the outer pair after each
sample was taken.

Sampling apparatus was not being decontaminated after each board.
Corrective action was implemented that apparatus would be wiped down
after each board using wipes wetted with DI water.

There were some initial problems with the wipe staying on the apparatus
when wiping boards with a rough surface. The corrective action taken was
to orient the wipe such that corners are facing board when wiping, which
helps the wipe stay in place.

In addition to the systems audit, three blind audit samples were
submitted to STL for analysis. Results from these samples were presented
in Table 2-20 and met all DQI goals.

An audit of data quality was performed on the information contained in
the database through Sampling Event #4 (July 2004). Database data was
checked for accuracy by comparing entries with raw data sheets, project
notebooks and laboratory reports. The ARCADIS QA Officer performed this
audit by randomly selecting samples representing approximately 10% of
the total number of entries contained in the database. The following
parameters were verified for 77 randomly selected specimens:

Specimen label

Board label

Minideck label

Specimen type

Date sampled

Date analyzed

Lab sample ID

Laboratory batch number

Arsenic concentration

Chromium concentration

Copper concentration

Core wood analysis results (when applicable)

Moisture content

Baseline arsenic, chromium and copper

Coating ID

In addition to the 77 samples randomly selected, 100% of the data
recorded for coating volume and coating mass for each minideck was
recalculated. Any discrepancies were immediately reported to the ARCADIS
work assignment leader (WAL) and database manager (Krich Ratanaphruks)
and corrected. An internal report was submitted to the ARCADIS WAL. The
following findings were cited:

Specimen A-BG-BL6 in database should actually be A-L-BL5. All other
information related to this sample is correct.

There is no baseline data for A-BG-M4 contained in the database.

In the file Baseline Samples-ver5.xls for sample C-BO, there are two BL4
entries. One should be BL5.

The concentration in the database for C-BT-M2 is an average of BL3 and
Bl4 and should be an average of BL2 and BL3.

There were discrepancies (affecting approximately 20%) in the way
moisture averages were calculated. The same logic was not applied to
every specimen.

There were numerous errors made in volume (affecting approximately 30%)
and weight (affecting approximately 15%) calculations for initial
coating of minidecks. Errors were caused by mistakes in subtraction and
difficulty reading handwriting in notebooks. For this reason, 100% of
the information was recalculated using the information documented on
original data sheets and in the project notebook. Corrected values were
submitted to the Database Manager.

The majority of these findings were minor and corrected in the database
immediately. As a result of the audit findings, a new column for
moisture data for each specimen was added to the database that is an
average of all measurements made for the entire board. This method of
calculating moisture eliminates having to determine which two moisture
measurements to use in the average to apply to each specimen. No
systematic corrective actions were implemented for the coating volume
and mass finding because these measurements will not be performed again
unless new minidecks are prepared. In the event new coatings are
evaluated, changes in how volumes and weights are recorded will be
implemented.

An EPA audit was performed November 18 through 30, 2004 at the time of
the collection and processing of the 15-month samples. The findings of
this audit state that the auditor observed no deviations from the QAPP.
Minor comments of the auditor have been incorporated into appropriate
sections of this report.

Other Data Corrections

Several other corrections to the dataset (via the database) were made
over the course of the study as errors were discovered. Nearly all of
the corrections were to address minor data entry errors that were easily
identified, confirmed and corrected. Most commonly, these involved the
transposition of a letter in the PSA ID; for example, typing in A-AF-M1
instead of A-F-M1. These types of errors were easily identified by
cross-referencing the minideck ID versus the list of specimens on each
minideck (i.e., using Table 2-7). 

To further check the dataset and identify suspected errors, a number of
checks were used:

The database output was sorted in several ways to ensure that the
expected number of samples for each specimen were represented.

Several quantitative measures (e.g., averages and standard deviations
for each coating) were used to confirm that the database data was
entered into the statistical model properly.

The Baseline and each time series dataset was manually reviewed to
identify inconsistencies or transpositions of data.

Summary statistics and plots were scrutinized for unusual or outlier
datapoints.

One datapoint strongly suspected to be in error was changed. It was the
sampling event #3 (time = 7 month) sample for specimen C-BX-M1 on
minideck 7-B. The reported data for this PSA sample showed a two-log
reduction in DA as compared with all of the other sampling events for
this sample, both pre- and post-event #3. Upon comparison with the raw
data, a peculiarity was noted with the next sample taken during event #3
which was from an untreated control board also on minideck 7-B. The
reported results for this sample showed unusually high DA. This evidence
indicates that the labeling of these two samples (C-BX-M1 and the
untreated control for 7-B) got mixed up, although we were unable to
determine how, when or where the error occurred (e.g., lab versus
field). Nevertheless, the circumstantial evidence suggesting such an
error was so strong that the two results were swapped.

