- Email: [email protected]
A soil sampling intercomparison exercise for the ALMERA network
Journal of Environmental Radioactivity 100 (2009) 982–987
Contents lists available at ScienceDirect
Journal of Environmental Radioactivity journal homepage: www.elsevier.com/locate/jenvrad
A soil sampling intercomparison exercise for the ALMERA network Maria Belli a, Paolo de Zorzi a, Umberto Sansone b, Abduhlghani Shakhashiro b, Adelaide Gondin da Fonseca b, *, Alexander Trinkl b, Thomas Benesch b a b
Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA), Via di Castel Romano 100, I-00128 Roma, Italy International Atomic Energy Agency (IAEA), Agency’s Laboratories Seibersdorf, A-2444 Seibersdorf, Austria
a r t i c l e i n f o
a b s t r a c t
Article history: Received 29 April 2009 Received in revised form 5 August 2009 Accepted 5 August 2009 Available online 26 August 2009
Soil sampling and analysis for radionuclides after an accidental or routine release is a key factor for the dose calculation to members of the public, and for the establishment of possible countermeasures. The IAEA organized for selected laboratories of the ALMERA (Analytical Laboratories for the Measurement of Environmental Radioactivity) network a Soil Sampling Intercomparison Exercise (IAEA/SIE/01) with the objective of comparing soil sampling procedures used by different laboratories. The ALMERA network is a world-wide network of analytical laboratories located in IAEA member states capable of providing reliable and timely analysis of environmental samples in the event of an accidental or intentional release of radioactivity. Ten ALMERA laboratories were selected to participate in the sampling exercise. The soil sampling intercomparison exercise took place in November 2005 in an agricultural area qualified as a ‘‘reference site’’, aimed at assessing the uncertainties associated with soil sampling in agricultural, semi-natural, urban and contaminated environments and suitable for performing sampling intercomparison. In this paper, the laboratories sampling performance were evaluated. Ó 2009 Elsevier Ltd. All rights reserved.
Keywords: Soil sampling ALMERA network Sampling intercomparison Sampling techniques
1. Introduction The ability to provide timely, accurate, and reliable data is central to the work of radio-analytical chemists and physicists. The quality of the analytical data is a key factor in the success of this kind of activity. The process of method development and validation has a direct impact on the quality of these data. In keeping with this policy, the IAEA Chemistry Unit has leading scientific and technical expertise in analytical measurements and radio-analytical chemistry. The Unit undertakes an extensive program of activities, including the organization and conducting of proficiency tests devoted to the analytical step. The sampling is now considered part of a measurement process including the analysis. How the laboratories are performing the sampling in the field or the suitability of different sampling methods to fitting a purpose or, again, the assessment of laboratories performance in sampling, may and has to be checked, as it routinely happens for the analytical activity. In order to achieve a higher degree of harmonization of sampling practices among selected laboratories, a soil sampling exercise with
* Corresponding author. Tel.: þ43 2600 28672. E-mail addresses: [email protected] (M. Belli), [email protected] (P. de Zorzi), [email protected] (U. Sansone), [email protected] (A. Shakhashiro), [email protected] (A. Gondin da Fonseca), [email protected] (A. Trinkl), [email protected] (T. Benesch). 0265-931X/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvrad.2009.08.002
the ALMERA laboratories was conducted at a ‘‘reference site’’. The term ‘‘reference site’’ is defined by IUPAC Recommendation (de Zorzi et al., 2005) as ‘‘an area, one or more of whose element concentration are well characterized in term of spatial and temporal variability’’. Such term implies ‘‘spatial and temporal variability’’ (de Zorzi et al., 2002, 2008a) that replaces the terms ‘homogeneity and stability’ used in the context of reference material. Homogeneity and stability are mandatory requirements for a reference material, as stated by the ISO Guides (ISO, 2006, 2007). In the case of a ‘‘reference site’’, there is not any reference material to be distributed to the laboratories participating in a proficiency test. The measurand is represented by the mass fraction of an element in a sampling target including its spatial and temporal variability and the users are required to demonstrate the ability of their sampling strategy to obtain that quantity value. In the long term after deposition, the behaviour of long-lived radionuclides in soil can be expected to be similar to that of some stable trace elements and the distribution of these trace elements in soil can simulate the distribution of radionuclides. Trace elements in soil, including radionuclides, are mostly associated with finer particlesize fractions (Coughtrey and Thorne, 1983; Spezzano, 2005). Positive correlation of most radionuclides was observed with metals (Navas et al., 2005). Furthermore, sampling for radionuclides in the environment is not unlike sampling for other attributes of environmental media and soil. Sampling procedures for radionuclides derive from techniques used in agriculture and
M. Belli et al. / Journal of Environmental Radioactivity 100 (2009) 982–987
12,5
Cell 1
Cell 2
As
11,5
mg/kg
engineering (HASL, 1997). The radionuclides considered in planning the sampling exercise were those that require radiochemical separation (90Sr, 240Pu, 241Am, 238U) and a test portion, defined as the ‘‘Quantity of material, of proper size for measurement of the concentration or other property of interest, removed from the test sample’’ (de Zorzi et al., 2005), ranging from 10 to 50 g, depending on the activity concentration of the radionuclide.
983
2001 2004 10,5
2. Materials and methods 2.1. Reference site characterization
9,5
Cell 1
Cell 2
Fe
mg/kg
28000
2001 2004 23000
18000
120
Cell 1
Cell 2
Sc
110
mg/kg
The reference site, located in the North Eastern part of Italy (Pozzuolo del Friuli, Udine), was previously characterized for metal mass fraction, adapting a reference sampling scheme previously designed and reported (Desaules et al., 2001). The site of 10,000 m2 was subdivided into 100 sub-areas (cells), of 10 m 10 m. The spatial variability (distribution) of element mass fractions along the site (two dimensional), performing in June 2001 the ‘‘reference sampling’’ (Barbizzi et al., 2004; de Zorzi et al., 2008b), was assessed. A composite sample for each cell was obtained by pooling 25 increments collected from depths 0–20 cm, within the ploughed layer (0–40 cm). The reference sampling was carried out by a single sampling team. Two cells were sampled again and the resulting 25 samples per cell were analysed separately to give information on the within-cell variability. The k0-method of Instrumental Neutron Activation Analysis (k0-INAA) was used for measurements of As, Fe, Sc, and Zn in the soil samples (Jac´imovic´ et al., 2002). These elements show a similar environmental behaviour of many radionuclides. Test portions, removed from the test sample, about 200 mg (one for each test sample) were measured. k0-INAA quality control was performed by using the reference material IAEA Soil-7. Normality of data distribution (coefficient of skewness, kurtosis, Kolmogorov–Smirnov test) of the characterization data was assessed using S-Plus 6 for Windows (Kaluzny et al., 1997). Table 1 reports the assigned values for the elements used in this intercomparison with their expanded uncertainty (U). U is calculated as the experimental standard deviation of the mean values (100 composite sample collected) calculated for a coverage factor k ¼ 2. Contributing to the overall uncertainty is spatial variability over the site, sampling (strategy, sampling device, and sampler), sample preparation (from the primary sample to the test sample) and analysis. 2.2. Temporal variability
2001
100
2004
90
80 120
Cell 1
Cell 2
Zn
110
mg/kg
Temporal variability is due to natural variation over time of the reference site. This variation could be determined by the reference site use, maintenance and weather condition. In the case of this intercomparison, the site was undisturbed for three years. Temporal variability was assessed carrying out a new sampling in selected cells three years after the reference sampling. Sampling and measurement procedures were the same as those applied during reference site characterization. Temporal variability of As, Sc, Fe and Zn are shown in Fig. 1. The values of element mass fractions related to soil samples collected in different time (2001 and 2004) into 2 cells are equivalent according to the criterion by which the difference from two results is less than two times the combined standard uncertainty (ISO, 2000; Lisinger, 2005). The element mass fractions do not vary in a way which affects any comparison between samples collected at different times. These results indicate that the possible causes of natural variation did not lead to instability of the reference site. Assuring a suitable temporal variability, sampling intercomparison exercises can be properly carried out after the reference site characterization.
100
2001 2004
90
2.3. Sampling intercomparison design Each participant of the soil sampling exercise was asked to use their own strategy and procedures (sampling strategy/pattern, type of sampling design, sampling device, sampling depth, etc.) for the assessment of the mean values of trace elements (metals) at the ‘‘reference site’’. The limiting conditions given by the organizers to the laboratories were: 15 laboratory samples as the maximum number of samples to be delivered to the IAEA for treatment and analysis and1 L as the maximum volume of each sample and 3 h as the maximum time period to carry out the sampling. The Chemistry Unit of the Physics, Chemistry and Instrumentation Laboratory in the IAEA’s Seibersdorf Laboratory has established a web-based interface to the Table 1 Reference value assigned to each analyte of the reference sampling site. As
Fe
Sc
Zn
25,570 565
8.6 0.2
91.8 2.1
mg kg1 10.6 0.2
80 Fig. 1. As, Fe, Sc and Zn temporal variability at the reference site.
intercomparison database to collect information from the participants. Each participant provided the description of the sampling devices used and the protocol applied during the intercomparison exercise, the description of the methodology used to prepare the test from each of the samples collected during the sampling exercise and the methodology used to estimate the mean value of the several trace elements in the sampling exercise.
2.4. Methods and strategies applied in the sampling exercise During the soil sampling exercise the participant laboratories adopted strategies which can be classified into the following three main classes: i) systematic by transects triangular grid and diagonals; ii) stratified random and iii) non-systematic/ irregular.
984
M. Belli et al. / Journal of Environmental Radioactivity 100 (2009) 982–987
Table 2 Summary of the sampling procedures and operational remarks. The higher number of asterisks (*) the better the operational performance. Samplers
Sampling pattern
Sampling devices
Type of sample
Sample pretreatment in the field
Number of laboratory samples (1)
Number of increments (2)
Time (3)
Effort
Operational Remarks (4)
1
1
Sampling rings
Composite
No
**
**
*
**
**
2
1
Non systematic Zig-zag Systematic
Single
Yes
*
–
**
*
*
3
2
Shovel / Spatula Corer
Single
No
*
–
**
*
*
4
1
5 6
2 2
7 8 9
1 1 2
10
1
Laborato
g Code
1. 2. 3. 4.
Systematic Random (vertical) Non Systematic Systematic Stratified random Systematic Systematic Stratified random Non Systematic
Spatula / Shovel Shovel Corer
Single
No
*
–
***
**
**
Single Composite
Yes No
* **
– ***
** **
** *
** **
Corer Shovel Corer
Composite Composite Single
Yes Yes No
* * *
** *** –
** ** ***
*** ** **
** ** **
Spatula
Composite
No
**
***
**
***
***
The number of samples collected is qualitatively evaluated: ***(1–5 samples), **(6–10 samples), *(11–15 samples). The number of increments pooled into each composite sample is qualitatively evaluated: ***(1–5 increments), **(6–10 increments), *(11 increments). ***(90 min); **(91–180 min); *(>180 min). The time was corrected by expert judgment considering each team composed by two samplers. The asterisks (*) represent an average of the qualitative judgments concerning number of samples/increments, time consumed and effort.
All sampling was carried out within the ploughed layer (0–40 cm) at a sampling depth ranging from 5 to 25 cm. In Table 2 a summary of the main aspects of the sampling procedures adopted by each laboratory is shown. The qualitative synthetic evaluation of the operational aspects of the sampling strategies and techniques, summarized by operational remarks, takes into account the number of samples and increments collected, the time consumed in the field and the estimated effort. The weight of composite samples ranged from 850 g to 2 kg, while the weight of single samples ranged from 50 g to 1.9 kg. Some laboratories performed part of the sample pre-treatment in the field by stone hand-picking, sieving, coning-quartering, homogenization, reducing the volume, until a maximum sample volume of 1 L (laboratory sample) was obtained. The application of the different sampling strategies and techniques in the case of the sampling intercomparison exercise influences other operational aspects, such as the required number of the samplers (de Zorzi, 2002, 2008a,b) and the time spent for sampling. Time available and number of samples can obviously impact directly on the final cost of sampling and analysis. However, some logistical restrictions must be taken into account in order to better interpret the different operational aspects of the sampling procedures adopted. Some teams were obliged to reduce the number of the samplers, so that their usual sampling procedures were modified. The teams were composed at most of two persons and even considering this better situation some samplers declared having worked under stress and with just sufficient time available. To rule out variability eventually caused by different soil sample preparation techniques and by different analytical laboratories, a single laboratory, the IAEA’s Chemistry Unit, reduced the laboratory samples received from each participant to test samples. The test portions from each test sample were analysed by a single laboratory, the Atomic Energy Commission of Syria, using the INAA technique. IAEA Soil-7 reference material was used as quality control material for INAA measurements of As, Sc, and Zn in the soil samples. IAEA Soil-7 does not report the certified value for Fe, but only an informative value.
2.5. Criteria for evaluation of results The participants’ data were evaluated according to ISO 13528: 2005 and ISO/IEC 17025. The performance of each participating institution was evaluated using the mean values of the measurands, assessed by the methodology provided by the participants, or in the case in which the participants did not report any suggestion for mean value assessment, using the arithmetic averages. Shapiro Wilk and Kolmogorov–Smirnov tests (Millard et al., 2001) were used to check the normality of the distribution of the mean values of all participating institutions. Grubbs’s test was used to identify outliers. According to ISO 13528:2005, the participants sampling performance was evaluated through the bias and the z scoring method using the following equations: D ¼ Xpar Xref
(1)
Xpar X
ref ffi z ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
(2)
u2par þ u2ref
where: D ¼ is the difference between the assigned values and laboratory’s mean value; Xref ¼ Reference value assigned to each measurand of the reference sampling site; Xpar ¼ Participant’s mean value for each measurand; z ¼ Score of the participant; uref ¼ Experimental standard uncertainty for each measurand assigned to the reference site, expressed as standard deviation of the mean; upar ¼ Participant’s experimental standard uncertainty for each Measurand The critical values related to the parameters used for evaluating the laboratory’s b calculated for performance are show in Table 3 and the robust standard deviation s the interlaboratory comparison exercise are shown in Table 4. This value represents the degree of dispersion of the participants’ results calculated on the basis of robust statistics (i.e. including outlier values) according to ISO 13528:2005 (E) by using Algorithm A.
3. Results and discussion
b* BIAS (D) and s
Acceptability
The graphs of the bias values attributed to each laboratory are s and shown in Figs. 2–5. The bias values are assessed versus the 2 b b equal to the robust standard deviation of the b values with s 3s
b jDj 2 s s < jDj 3 sb 2b b jDj > 3 s SCORE (z) jzj 2 2 < jzj 3 jzj > 3
Suitable strategy Warning/Questionable Action
Table 4 s ) of the soil sampling intercomparison exercise. Robust standard deviation ( b
Table 3 Scale of acceptability for bias (d) and for score (z).
Suitable strategy Warning/Questionable Action
b is the robust standard deviation of the soils sampling intercomparison IAEA/SIE/ *s 01 calculated according to ISO 13528:2005, Algorithm A.
Measurand
Robust standard deviation (mg kg1)
As Fe Sc Zn
0.7 1565 0.4 5
M. Belli et al. / Journal of Environmental Radioactivity 100 (2009) 982–987
Bias (D) for As
985
B i as (D) for Zn
2,5
52,5
20,0
10,0
0,5 -0,5
1
2
3
4
5
6
7
8
9
10
-1,5
D (mg/kg)
D (mg/kg)
1,5
0,0 1
2
3
4
5
6
7
8
9
10
-10,0
-2,5
Laboratory code
-20,0
Fig. 2. Bias values (D) for arsenic.
L a bor a tor y c o de Fig. 5. Bias values (D) for zinc.
Bias (D) for Fe
5250
D (mg/kg)
3500 1750 0 1
-1750
2
3
4
5
6
7
8
9
10
-3500 -5250
Laboratory code Fig. 3. Bias values (D) for iron.
Bias (D) for Sc
2,0
Table 5 Application of ISO guide 33 criterion for the assessment of the equivalence of the sampling strategies.
1,0
D (mg/kg)
values, due to different sampling strategies and techniques used in the intercomparison exercise, were not identified. A confirmation of this behaviour can be made by a simulation based on the data set obtained during the characterization of the reference site. The simulation includes the comparison between the mean values obtained for each laboratory calculated for the exercise IAEA/SIE/01 and the mean values calculated on the data set of the reference sampling simulating the same sampling strategies. The comparison, in analogy with the criterion used for the assessment of the temporal variability is made by applying the criterion of Equation (3) below by which the difference Dm of the laboratory’s mean value (e.g. XLab_1) and the corresponding simulated laboratory’s mean value (e.g. XLab_1_simulation) is less than two times the associated combined standard uncertainty uDm (expanded uncertainty UDm) (ISO, 2000; Lisinger, 2005). UDm is
Lab code
Criterion ISO Guide 33
As
Fe
Sc
Zn
1
Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm
1.1 3.3 Pass 0.9 3.3 Pass 1.4 1.8 Pass 0.8 4.9 Pass 1.0 4.4 Pass 1.1 2.6 Pass 0.5 2.6 Pass 2.0 3.1 Pass 0.2 3.4 Pass 1.2 2.8 Pass
2098 11,614 Pass 285 8567 Pass 821 5870 Pass 574 7414 Pass 1461 7037 Pass 6 9099 Pass 1841 7819 Pass 2997 6614 Pass 5534 12,050 Pass 1693 8899 Pass
0.9 4.1 Pass 0.1 2.7 Pass 0.6 2.2 Pass 0.5 2.7 Pass 0.7 2.6 Pass 0.3 3.3 Pass 0.9 2.9 Pass 1.4 2.3 Pass 1.1 3.5 Pass 0.9 3.2 Pass
3 42 Pass 2 32 Pass 4 19 Pass 2 26 Pass 3 23 Pass 51 41 Failed 1 29 Pass 1 35 Pass 3 31 Pass 5 31 Pass
0,0 1
2
3
4
5
6
7
8
9
10
-1,0
2
3
-2,0
Laboratory code Fig. 4. Bias values (D) for scandium.
s values have been calculated according intercomparison exercise. b s and 3 bs values are assigned respectively to ISO 13528:2005. The 2 b to the internal and external additional horizontal lines. Absolute b , while action signals, bias values are acceptable for values 2 s corresponding to sampling strategies to be corrected, is given by s. bias absolute value >3 b In general, the resulting participants’ mean values do not lead to critical bias values. At most 10% of the participant’s bias values (Lab b and in the 9 for Fe, Lab 8 for Sc and Lab 6 for Zn) are higher than 3 s case of Lab 6, this value is clearly associated with an outlier s measurement result for Zn. 60–90% of the bias values are 2 b corresponding to suitable sampling strategies. On the basis of these results significant differences between the laboratories’ mean
4
5
6
7
8
9
10
986
M. Belli et al. / Journal of Environmental Radioactivity 100 (2009) 982–987
ζ- score for As
Absolute value
12
Absolute value
ζ - score for Zn
9,0
8
4
6,0
3,0
0,0
0 1
2
3
4
5
6
7
8
9
1
10
2
3
4
5
6
7
8
9
10
Laboratory code
Laboratory code
Fig. 9. Evaluation of the laboratories performance (z-score values) for zinc. Fig. 6. Evaluation of the laboratories performance (z-score values) for arsenic.
sampling strategy chosen does not influence the final measurements results in terms of mean value. Moreover, the figures show a general underestimation of the mean values obtained within the framework of IAEA/SIE/01. Considering the under control temporal variability previously shown, this behaviour, mostly marked for scandium, cannot be fully explained on the basis of the available information. However, in most of the cases the underestimation observed does not corres. spond bias values higher than 3 b A second test for evaluating the laboratory’s performance was applied on the basis of the z scores. The absolute values of z scores attributed to the laboratories are shown in the graphs of Figs. 6–9. The critical limits of 2 and 3 are assigned respectively to the internal and external additional horizontal lines. For values 2 the participant’s result associated to the sampling strategy adopted could be considered suitable for the objective stated; for value between 2 and 3 the participant’s result could be questionable and for values >3 the sampling strategy would need corrective actions. The z score takes into account the standard uncertainty of the participants’ results and of the assigned value. The score is used when uncertainty of the assigned value is not calculated using the results reported by the participants, as in the case of IAEA/SIE/01 sampling intercomparison exercise. At most 40% of the z scores (As) exceed the critical value of 3. 60–90% of the values are 2 (suitable strategy). The higher percentage of acceptable scores is observed for Zn (80%) while Fe, Sc, and As show comparable percentage of scores 2 and in the range 2 to3. These results confirm what has been observed by evaluating the bias values.
ζ- score for Fe
Absolute value
6
3
0 1
2
3
4
5
6
7
8
9
10
Laboratory code Fig. 7. Evaluation of the laboratories performance (z-score values) for iron.
ζ - score for Sc
Absolute value
9
6
3
4. Conclusions 0 1
2
3
4
5
6
7
8
9
10
Laboratory code Fig. 8. Evaluation of the laboratories performance (z-score values) for scandium.
calculated for all the elements from the uncertainty of the laboratory’s mean values and the uncertainty of the simulated values.
XLab
1
XLab
1 simulation
¼ Dm < U Dm
(3)
For As, Fe, Sc and Zn all the laboratories pass the criterion stated above, with the exception of Lab 6 for Zn, due to an outlier value (Table 5). As by applying different sampling strategies to the reference sampling data set the criterion was respected by all the laboratories, this means that for this kind of reference site, the
According to the criterion of ISO 13258:2005 for the assessment of the laboratories’ performances, the strategies adopted by all the sampling teams were in general suitable for the purpose, not b . Same conclusions exceeding in most of the cases bias values of 2 s can be made with reference to the z scores. In general, for all the elements the slight differences observed between the laboratories do not appear attributable to the different sampling strategies. The figures show that sampling of top soil in arable, ploughed land is relatively easy, leading to equivalent results between different sampling procedures. On the basis of these results, in a homogeneous agricultural area the sampling strategies chosen by the laboratories can be considered equivalent. Nevertheless, the significant differences are mainly due to some operational aspects. Collecting a high number of samples (composite or singles), using more complicated sampling devices
M. Belli et al. / Journal of Environmental Radioactivity 100 (2009) 982–987
or, in some cases, applying sample pre-treatment in the field lead to different time requirements for performing the sampling, increasing also the total expected costs for sampling and analysis. The general equivalence of the approaches proposed by the laboratories, in terms of agreement of their measurement results with the assigned values, seems to justify, in the use of simple sampling procedures, not including any sample pre-treatment in the field. The results of the intercomparison confirm, on the basis of experimental data, what was probably intuitive. The IAEA/SIE/01 intercomparison exercise approach could be useful for a future testing of different sampling strategies in a more heterogeneous area, such as a semi-natural or a contaminated soil area, with the aim of finding and identifying artificial hot spots. The fundamental requirement of a reference material for chemical analysis is represented by its homogeneity. In the case of a reference site used within the framework of soil sampling intercomparison exercise, aimed at assessing different sampling procedures, the heterogeneity of the sampling target should be desirable. Acknowledgements This work was financially supported under the IAEA subprogramme ‘Supporting Quality in the Analysis of Terrestrial Environmental Samples’. The authors would like to thanks the teams of the 10 Countries selected to participate in the sampling exercise: Republic of Korea, Mexico, Syrian, Ukraine, Islamic Republic of Iran, Slovak, Slovenia, Hungary, Brazil and Lithuania. References Barbizzi, S., De Zorzi, P., Belli, M., Pati, A., Sansone, U., Stellato, L., Barbina, M., Deluisa, A., Menegon, S., Coletti, V., 2004. Characterisation of reference site for quantifying uncertainties related to soil sampling. Environmental Pollution 127, 131–135. Coughtrey, P.J., Thorne, M.C., 1983. Radionuclide Distribution and Transport in Terrestrial and Aquatic Ecosystems. A Critical Review of Data. A.A, Balkema, Rotterdam.
987
Desaules, A., Sprengart, J., Wagner, G., Muntau, H., Theocharopoulos, S., 2001. Description of the test area and reference sampling at Dornach. The Science of Total Environment 264, 17–26. HASL-300, 1997. The Procedures Manual of the Environmental Measurements Laboratory. In: Section 2.4, 28th ed., vol. 1. U.S. Department of Homeland Security, New York. International Organization for Standardization, 2005. General Requirements for the Competence of Testing and Calibration Laboratories. In: ISO/IEC 17025 Geneva. International Organization for Standardization, 2007. International Vocabulary of Metrology – Basic and General Concepts and Associated Terms, third ed.. In: ISO Guide 99 Geneva, Switzerland. International Organization for Standardization, 2006. Reference Materials-General and Statistical Principles for Certification. In: ISO Guide 35 Geneva, Switzerland. International Organization for Standardization, 2005. Statistical Methods For Use In Proficiency Testing By Interlaboratory Comparisons. In: ISO 13528 (E)Geneva, Switzerland. International Organization for Standardization, 2000. Uses of Certified Reference Material. In: ISO Guide 33 Geneva, Switzerland. Jac´imovic´, R., Lazaru, A., Mihajlovic´, D., Ilic´, R., Stafilov, T., 2002. Determination of major and trace elements in some minerals by k0-instrumental neutron activation analysis. Journal of Radioanalytical and Nuclear Chemistry 253, 427–434. Kaluzny, S.P., Vega, S.C., Cardoso, T.P., Shelly, A.A., 1997. S þ Spatial Stats User’s Manual for Windows and Unix. Springer, New York. Lisinger, T., 2005. Application Note 1 – comparison of a measurement result with the certified value. ERM, European Reference Material. Millard, S.P., Nagaraj, K., Neerchal, S.P.N., 2001. Environmental Statistics with S-Plus. CRC Press, USA. Navas, A., Machı`n, J., Soto, J., 2005. Mobility of natural radionuclides and selected major and trace elements along a soil toposequence in the central Spanish Pyrenees. Soil Science 170-9, 743–757. Spezzano, P., 2005. Distribution of pre- and post-Chernobyl radiocaesium with particle size fractions of soil. Journal of Environmental Radioactivity 83, 117–127. de Zorzi, P., Belli, M., Barbizzi, S., Menegon, S., Deluisa, A., 2002. A practical approach to assessment of sampling uncertainty. Accreditation and Quality Assurance 7 (5) 182–188. de Zorzi, P., Barbizzi, S., Belli, M., Ciceri, G., Fajgelj, A., Moore, D., Sansone, U., Van Der Perk, M., 2005. Terminology in soil sampling (IUPAC Recommendation 2005). Pure and Applied Chemistry 77 (5), 827–841. de Zorzi, P., et al., 2008a. Soil sampling strategies: evaluation of different approaches. Applied Radiation and Isotopes 66, 1691–1694. de Zorzi, P., et al., 2008b. A soil sampling reference site: the challenge in defining reference material for sampling. Applied Radiation and Isotopes 66, 1588–1591.
Contents lists available at ScienceDirect
Journal of Environmental Radioactivity journal homepage: www.elsevier.com/locate/jenvrad
A soil sampling intercomparison exercise for the ALMERA network Maria Belli a, Paolo de Zorzi a, Umberto Sansone b, Abduhlghani Shakhashiro b, Adelaide Gondin da Fonseca b, *, Alexander Trinkl b, Thomas Benesch b a b
Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA), Via di Castel Romano 100, I-00128 Roma, Italy International Atomic Energy Agency (IAEA), Agency’s Laboratories Seibersdorf, A-2444 Seibersdorf, Austria
a r t i c l e i n f o
a b s t r a c t
Article history: Received 29 April 2009 Received in revised form 5 August 2009 Accepted 5 August 2009 Available online 26 August 2009
Soil sampling and analysis for radionuclides after an accidental or routine release is a key factor for the dose calculation to members of the public, and for the establishment of possible countermeasures. The IAEA organized for selected laboratories of the ALMERA (Analytical Laboratories for the Measurement of Environmental Radioactivity) network a Soil Sampling Intercomparison Exercise (IAEA/SIE/01) with the objective of comparing soil sampling procedures used by different laboratories. The ALMERA network is a world-wide network of analytical laboratories located in IAEA member states capable of providing reliable and timely analysis of environmental samples in the event of an accidental or intentional release of radioactivity. Ten ALMERA laboratories were selected to participate in the sampling exercise. The soil sampling intercomparison exercise took place in November 2005 in an agricultural area qualified as a ‘‘reference site’’, aimed at assessing the uncertainties associated with soil sampling in agricultural, semi-natural, urban and contaminated environments and suitable for performing sampling intercomparison. In this paper, the laboratories sampling performance were evaluated. Ó 2009 Elsevier Ltd. All rights reserved.
Keywords: Soil sampling ALMERA network Sampling intercomparison Sampling techniques
1. Introduction The ability to provide timely, accurate, and reliable data is central to the work of radio-analytical chemists and physicists. The quality of the analytical data is a key factor in the success of this kind of activity. The process of method development and validation has a direct impact on the quality of these data. In keeping with this policy, the IAEA Chemistry Unit has leading scientific and technical expertise in analytical measurements and radio-analytical chemistry. The Unit undertakes an extensive program of activities, including the organization and conducting of proficiency tests devoted to the analytical step. The sampling is now considered part of a measurement process including the analysis. How the laboratories are performing the sampling in the field or the suitability of different sampling methods to fitting a purpose or, again, the assessment of laboratories performance in sampling, may and has to be checked, as it routinely happens for the analytical activity. In order to achieve a higher degree of harmonization of sampling practices among selected laboratories, a soil sampling exercise with
* Corresponding author. Tel.: þ43 2600 28672. E-mail addresses: [email protected] (M. Belli), [email protected] (P. de Zorzi), [email protected] (U. Sansone), [email protected] (A. Shakhashiro), [email protected] (A. Gondin da Fonseca), [email protected] (A. Trinkl), [email protected] (T. Benesch). 0265-931X/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvrad.2009.08.002
the ALMERA laboratories was conducted at a ‘‘reference site’’. The term ‘‘reference site’’ is defined by IUPAC Recommendation (de Zorzi et al., 2005) as ‘‘an area, one or more of whose element concentration are well characterized in term of spatial and temporal variability’’. Such term implies ‘‘spatial and temporal variability’’ (de Zorzi et al., 2002, 2008a) that replaces the terms ‘homogeneity and stability’ used in the context of reference material. Homogeneity and stability are mandatory requirements for a reference material, as stated by the ISO Guides (ISO, 2006, 2007). In the case of a ‘‘reference site’’, there is not any reference material to be distributed to the laboratories participating in a proficiency test. The measurand is represented by the mass fraction of an element in a sampling target including its spatial and temporal variability and the users are required to demonstrate the ability of their sampling strategy to obtain that quantity value. In the long term after deposition, the behaviour of long-lived radionuclides in soil can be expected to be similar to that of some stable trace elements and the distribution of these trace elements in soil can simulate the distribution of radionuclides. Trace elements in soil, including radionuclides, are mostly associated with finer particlesize fractions (Coughtrey and Thorne, 1983; Spezzano, 2005). Positive correlation of most radionuclides was observed with metals (Navas et al., 2005). Furthermore, sampling for radionuclides in the environment is not unlike sampling for other attributes of environmental media and soil. Sampling procedures for radionuclides derive from techniques used in agriculture and
M. Belli et al. / Journal of Environmental Radioactivity 100 (2009) 982–987
12,5
Cell 1
Cell 2
As
11,5
mg/kg
engineering (HASL, 1997). The radionuclides considered in planning the sampling exercise were those that require radiochemical separation (90Sr, 240Pu, 241Am, 238U) and a test portion, defined as the ‘‘Quantity of material, of proper size for measurement of the concentration or other property of interest, removed from the test sample’’ (de Zorzi et al., 2005), ranging from 10 to 50 g, depending on the activity concentration of the radionuclide.
983
2001 2004 10,5
2. Materials and methods 2.1. Reference site characterization
9,5
Cell 1
Cell 2
Fe
mg/kg
28000
2001 2004 23000
18000
120
Cell 1
Cell 2
Sc
110
mg/kg
The reference site, located in the North Eastern part of Italy (Pozzuolo del Friuli, Udine), was previously characterized for metal mass fraction, adapting a reference sampling scheme previously designed and reported (Desaules et al., 2001). The site of 10,000 m2 was subdivided into 100 sub-areas (cells), of 10 m 10 m. The spatial variability (distribution) of element mass fractions along the site (two dimensional), performing in June 2001 the ‘‘reference sampling’’ (Barbizzi et al., 2004; de Zorzi et al., 2008b), was assessed. A composite sample for each cell was obtained by pooling 25 increments collected from depths 0–20 cm, within the ploughed layer (0–40 cm). The reference sampling was carried out by a single sampling team. Two cells were sampled again and the resulting 25 samples per cell were analysed separately to give information on the within-cell variability. The k0-method of Instrumental Neutron Activation Analysis (k0-INAA) was used for measurements of As, Fe, Sc, and Zn in the soil samples (Jac´imovic´ et al., 2002). These elements show a similar environmental behaviour of many radionuclides. Test portions, removed from the test sample, about 200 mg (one for each test sample) were measured. k0-INAA quality control was performed by using the reference material IAEA Soil-7. Normality of data distribution (coefficient of skewness, kurtosis, Kolmogorov–Smirnov test) of the characterization data was assessed using S-Plus 6 for Windows (Kaluzny et al., 1997). Table 1 reports the assigned values for the elements used in this intercomparison with their expanded uncertainty (U). U is calculated as the experimental standard deviation of the mean values (100 composite sample collected) calculated for a coverage factor k ¼ 2. Contributing to the overall uncertainty is spatial variability over the site, sampling (strategy, sampling device, and sampler), sample preparation (from the primary sample to the test sample) and analysis. 2.2. Temporal variability
2001
100
2004
90
80 120
Cell 1
Cell 2
Zn
110
mg/kg
Temporal variability is due to natural variation over time of the reference site. This variation could be determined by the reference site use, maintenance and weather condition. In the case of this intercomparison, the site was undisturbed for three years. Temporal variability was assessed carrying out a new sampling in selected cells three years after the reference sampling. Sampling and measurement procedures were the same as those applied during reference site characterization. Temporal variability of As, Sc, Fe and Zn are shown in Fig. 1. The values of element mass fractions related to soil samples collected in different time (2001 and 2004) into 2 cells are equivalent according to the criterion by which the difference from two results is less than two times the combined standard uncertainty (ISO, 2000; Lisinger, 2005). The element mass fractions do not vary in a way which affects any comparison between samples collected at different times. These results indicate that the possible causes of natural variation did not lead to instability of the reference site. Assuring a suitable temporal variability, sampling intercomparison exercises can be properly carried out after the reference site characterization.
100
2001 2004
90
2.3. Sampling intercomparison design Each participant of the soil sampling exercise was asked to use their own strategy and procedures (sampling strategy/pattern, type of sampling design, sampling device, sampling depth, etc.) for the assessment of the mean values of trace elements (metals) at the ‘‘reference site’’. The limiting conditions given by the organizers to the laboratories were: 15 laboratory samples as the maximum number of samples to be delivered to the IAEA for treatment and analysis and1 L as the maximum volume of each sample and 3 h as the maximum time period to carry out the sampling. The Chemistry Unit of the Physics, Chemistry and Instrumentation Laboratory in the IAEA’s Seibersdorf Laboratory has established a web-based interface to the Table 1 Reference value assigned to each analyte of the reference sampling site. As
Fe
Sc
Zn
25,570 565
8.6 0.2
91.8 2.1
mg kg1 10.6 0.2
80 Fig. 1. As, Fe, Sc and Zn temporal variability at the reference site.
intercomparison database to collect information from the participants. Each participant provided the description of the sampling devices used and the protocol applied during the intercomparison exercise, the description of the methodology used to prepare the test from each of the samples collected during the sampling exercise and the methodology used to estimate the mean value of the several trace elements in the sampling exercise.
2.4. Methods and strategies applied in the sampling exercise During the soil sampling exercise the participant laboratories adopted strategies which can be classified into the following three main classes: i) systematic by transects triangular grid and diagonals; ii) stratified random and iii) non-systematic/ irregular.
984
M. Belli et al. / Journal of Environmental Radioactivity 100 (2009) 982–987
Table 2 Summary of the sampling procedures and operational remarks. The higher number of asterisks (*) the better the operational performance. Samplers
Sampling pattern
Sampling devices
Type of sample
Sample pretreatment in the field
Number of laboratory samples (1)
Number of increments (2)
Time (3)
Effort
Operational Remarks (4)
1
1
Sampling rings
Composite
No
**
**
*
**
**
2
1
Non systematic Zig-zag Systematic
Single
Yes
*
–
**
*
*
3
2
Shovel / Spatula Corer
Single
No
*
–
**
*
*
4
1
5 6
2 2
7 8 9
1 1 2
10
1
Laborato
g Code
1. 2. 3. 4.
Systematic Random (vertical) Non Systematic Systematic Stratified random Systematic Systematic Stratified random Non Systematic
Spatula / Shovel Shovel Corer
Single
No
*
–
***
**
**
Single Composite
Yes No
* **
– ***
** **
** *
** **
Corer Shovel Corer
Composite Composite Single
Yes Yes No
* * *
** *** –
** ** ***
*** ** **
** ** **
Spatula
Composite
No
**
***
**
***
***
The number of samples collected is qualitatively evaluated: ***(1–5 samples), **(6–10 samples), *(11–15 samples). The number of increments pooled into each composite sample is qualitatively evaluated: ***(1–5 increments), **(6–10 increments), *(11 increments). ***(90 min); **(91–180 min); *(>180 min). The time was corrected by expert judgment considering each team composed by two samplers. The asterisks (*) represent an average of the qualitative judgments concerning number of samples/increments, time consumed and effort.
All sampling was carried out within the ploughed layer (0–40 cm) at a sampling depth ranging from 5 to 25 cm. In Table 2 a summary of the main aspects of the sampling procedures adopted by each laboratory is shown. The qualitative synthetic evaluation of the operational aspects of the sampling strategies and techniques, summarized by operational remarks, takes into account the number of samples and increments collected, the time consumed in the field and the estimated effort. The weight of composite samples ranged from 850 g to 2 kg, while the weight of single samples ranged from 50 g to 1.9 kg. Some laboratories performed part of the sample pre-treatment in the field by stone hand-picking, sieving, coning-quartering, homogenization, reducing the volume, until a maximum sample volume of 1 L (laboratory sample) was obtained. The application of the different sampling strategies and techniques in the case of the sampling intercomparison exercise influences other operational aspects, such as the required number of the samplers (de Zorzi, 2002, 2008a,b) and the time spent for sampling. Time available and number of samples can obviously impact directly on the final cost of sampling and analysis. However, some logistical restrictions must be taken into account in order to better interpret the different operational aspects of the sampling procedures adopted. Some teams were obliged to reduce the number of the samplers, so that their usual sampling procedures were modified. The teams were composed at most of two persons and even considering this better situation some samplers declared having worked under stress and with just sufficient time available. To rule out variability eventually caused by different soil sample preparation techniques and by different analytical laboratories, a single laboratory, the IAEA’s Chemistry Unit, reduced the laboratory samples received from each participant to test samples. The test portions from each test sample were analysed by a single laboratory, the Atomic Energy Commission of Syria, using the INAA technique. IAEA Soil-7 reference material was used as quality control material for INAA measurements of As, Sc, and Zn in the soil samples. IAEA Soil-7 does not report the certified value for Fe, but only an informative value.
2.5. Criteria for evaluation of results The participants’ data were evaluated according to ISO 13528: 2005 and ISO/IEC 17025. The performance of each participating institution was evaluated using the mean values of the measurands, assessed by the methodology provided by the participants, or in the case in which the participants did not report any suggestion for mean value assessment, using the arithmetic averages. Shapiro Wilk and Kolmogorov–Smirnov tests (Millard et al., 2001) were used to check the normality of the distribution of the mean values of all participating institutions. Grubbs’s test was used to identify outliers. According to ISO 13528:2005, the participants sampling performance was evaluated through the bias and the z scoring method using the following equations: D ¼ Xpar Xref
(1)
Xpar X
ref ffi z ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
(2)
u2par þ u2ref
where: D ¼ is the difference between the assigned values and laboratory’s mean value; Xref ¼ Reference value assigned to each measurand of the reference sampling site; Xpar ¼ Participant’s mean value for each measurand; z ¼ Score of the participant; uref ¼ Experimental standard uncertainty for each measurand assigned to the reference site, expressed as standard deviation of the mean; upar ¼ Participant’s experimental standard uncertainty for each Measurand The critical values related to the parameters used for evaluating the laboratory’s b calculated for performance are show in Table 3 and the robust standard deviation s the interlaboratory comparison exercise are shown in Table 4. This value represents the degree of dispersion of the participants’ results calculated on the basis of robust statistics (i.e. including outlier values) according to ISO 13528:2005 (E) by using Algorithm A.
3. Results and discussion
b* BIAS (D) and s
Acceptability
The graphs of the bias values attributed to each laboratory are s and shown in Figs. 2–5. The bias values are assessed versus the 2 b b equal to the robust standard deviation of the b values with s 3s
b jDj 2 s s < jDj 3 sb 2b b jDj > 3 s SCORE (z) jzj 2 2 < jzj 3 jzj > 3
Suitable strategy Warning/Questionable Action
Table 4 s ) of the soil sampling intercomparison exercise. Robust standard deviation ( b
Table 3 Scale of acceptability for bias (d) and for score (z).
Suitable strategy Warning/Questionable Action
b is the robust standard deviation of the soils sampling intercomparison IAEA/SIE/ *s 01 calculated according to ISO 13528:2005, Algorithm A.
Measurand
Robust standard deviation (mg kg1)
As Fe Sc Zn
0.7 1565 0.4 5
M. Belli et al. / Journal of Environmental Radioactivity 100 (2009) 982–987
Bias (D) for As
985
B i as (D) for Zn
2,5
52,5
20,0
10,0
0,5 -0,5
1
2
3
4
5
6
7
8
9
10
-1,5
D (mg/kg)
D (mg/kg)
1,5
0,0 1
2
3
4
5
6
7
8
9
10
-10,0
-2,5
Laboratory code
-20,0
Fig. 2. Bias values (D) for arsenic.
L a bor a tor y c o de Fig. 5. Bias values (D) for zinc.
Bias (D) for Fe
5250
D (mg/kg)
3500 1750 0 1
-1750
2
3
4
5
6
7
8
9
10
-3500 -5250
Laboratory code Fig. 3. Bias values (D) for iron.
Bias (D) for Sc
2,0
Table 5 Application of ISO guide 33 criterion for the assessment of the equivalence of the sampling strategies.
1,0
D (mg/kg)
values, due to different sampling strategies and techniques used in the intercomparison exercise, were not identified. A confirmation of this behaviour can be made by a simulation based on the data set obtained during the characterization of the reference site. The simulation includes the comparison between the mean values obtained for each laboratory calculated for the exercise IAEA/SIE/01 and the mean values calculated on the data set of the reference sampling simulating the same sampling strategies. The comparison, in analogy with the criterion used for the assessment of the temporal variability is made by applying the criterion of Equation (3) below by which the difference Dm of the laboratory’s mean value (e.g. XLab_1) and the corresponding simulated laboratory’s mean value (e.g. XLab_1_simulation) is less than two times the associated combined standard uncertainty uDm (expanded uncertainty UDm) (ISO, 2000; Lisinger, 2005). UDm is
Lab code
Criterion ISO Guide 33
As
Fe
Sc
Zn
1
Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm Dm UDm
1.1 3.3 Pass 0.9 3.3 Pass 1.4 1.8 Pass 0.8 4.9 Pass 1.0 4.4 Pass 1.1 2.6 Pass 0.5 2.6 Pass 2.0 3.1 Pass 0.2 3.4 Pass 1.2 2.8 Pass
2098 11,614 Pass 285 8567 Pass 821 5870 Pass 574 7414 Pass 1461 7037 Pass 6 9099 Pass 1841 7819 Pass 2997 6614 Pass 5534 12,050 Pass 1693 8899 Pass
0.9 4.1 Pass 0.1 2.7 Pass 0.6 2.2 Pass 0.5 2.7 Pass 0.7 2.6 Pass 0.3 3.3 Pass 0.9 2.9 Pass 1.4 2.3 Pass 1.1 3.5 Pass 0.9 3.2 Pass
3 42 Pass 2 32 Pass 4 19 Pass 2 26 Pass 3 23 Pass 51 41 Failed 1 29 Pass 1 35 Pass 3 31 Pass 5 31 Pass
0,0 1
2
3
4
5
6
7
8
9
10
-1,0
2
3
-2,0
Laboratory code Fig. 4. Bias values (D) for scandium.
s values have been calculated according intercomparison exercise. b s and 3 bs values are assigned respectively to ISO 13528:2005. The 2 b to the internal and external additional horizontal lines. Absolute b , while action signals, bias values are acceptable for values 2 s corresponding to sampling strategies to be corrected, is given by s. bias absolute value >3 b In general, the resulting participants’ mean values do not lead to critical bias values. At most 10% of the participant’s bias values (Lab b and in the 9 for Fe, Lab 8 for Sc and Lab 6 for Zn) are higher than 3 s case of Lab 6, this value is clearly associated with an outlier s measurement result for Zn. 60–90% of the bias values are 2 b corresponding to suitable sampling strategies. On the basis of these results significant differences between the laboratories’ mean
4
5
6
7
8
9
10
986
M. Belli et al. / Journal of Environmental Radioactivity 100 (2009) 982–987
ζ- score for As
Absolute value
12
Absolute value
ζ - score for Zn
9,0
8
4
6,0
3,0
0,0
0 1
2
3
4
5
6
7
8
9
1
10
2
3
4
5
6
7
8
9
10
Laboratory code
Laboratory code
Fig. 9. Evaluation of the laboratories performance (z-score values) for zinc. Fig. 6. Evaluation of the laboratories performance (z-score values) for arsenic.
sampling strategy chosen does not influence the final measurements results in terms of mean value. Moreover, the figures show a general underestimation of the mean values obtained within the framework of IAEA/SIE/01. Considering the under control temporal variability previously shown, this behaviour, mostly marked for scandium, cannot be fully explained on the basis of the available information. However, in most of the cases the underestimation observed does not corres. spond bias values higher than 3 b A second test for evaluating the laboratory’s performance was applied on the basis of the z scores. The absolute values of z scores attributed to the laboratories are shown in the graphs of Figs. 6–9. The critical limits of 2 and 3 are assigned respectively to the internal and external additional horizontal lines. For values 2 the participant’s result associated to the sampling strategy adopted could be considered suitable for the objective stated; for value between 2 and 3 the participant’s result could be questionable and for values >3 the sampling strategy would need corrective actions. The z score takes into account the standard uncertainty of the participants’ results and of the assigned value. The score is used when uncertainty of the assigned value is not calculated using the results reported by the participants, as in the case of IAEA/SIE/01 sampling intercomparison exercise. At most 40% of the z scores (As) exceed the critical value of 3. 60–90% of the values are 2 (suitable strategy). The higher percentage of acceptable scores is observed for Zn (80%) while Fe, Sc, and As show comparable percentage of scores 2 and in the range 2 to3. These results confirm what has been observed by evaluating the bias values.
ζ- score for Fe
Absolute value
6
3
0 1
2
3
4
5
6
7
8
9
10
Laboratory code Fig. 7. Evaluation of the laboratories performance (z-score values) for iron.
ζ - score for Sc
Absolute value
9
6
3
4. Conclusions 0 1
2
3
4
5
6
7
8
9
10
Laboratory code Fig. 8. Evaluation of the laboratories performance (z-score values) for scandium.
calculated for all the elements from the uncertainty of the laboratory’s mean values and the uncertainty of the simulated values.
XLab
1
XLab
1 simulation
¼ Dm < U Dm
(3)
For As, Fe, Sc and Zn all the laboratories pass the criterion stated above, with the exception of Lab 6 for Zn, due to an outlier value (Table 5). As by applying different sampling strategies to the reference sampling data set the criterion was respected by all the laboratories, this means that for this kind of reference site, the
According to the criterion of ISO 13258:2005 for the assessment of the laboratories’ performances, the strategies adopted by all the sampling teams were in general suitable for the purpose, not b . Same conclusions exceeding in most of the cases bias values of 2 s can be made with reference to the z scores. In general, for all the elements the slight differences observed between the laboratories do not appear attributable to the different sampling strategies. The figures show that sampling of top soil in arable, ploughed land is relatively easy, leading to equivalent results between different sampling procedures. On the basis of these results, in a homogeneous agricultural area the sampling strategies chosen by the laboratories can be considered equivalent. Nevertheless, the significant differences are mainly due to some operational aspects. Collecting a high number of samples (composite or singles), using more complicated sampling devices
M. Belli et al. / Journal of Environmental Radioactivity 100 (2009) 982–987
or, in some cases, applying sample pre-treatment in the field lead to different time requirements for performing the sampling, increasing also the total expected costs for sampling and analysis. The general equivalence of the approaches proposed by the laboratories, in terms of agreement of their measurement results with the assigned values, seems to justify, in the use of simple sampling procedures, not including any sample pre-treatment in the field. The results of the intercomparison confirm, on the basis of experimental data, what was probably intuitive. The IAEA/SIE/01 intercomparison exercise approach could be useful for a future testing of different sampling strategies in a more heterogeneous area, such as a semi-natural or a contaminated soil area, with the aim of finding and identifying artificial hot spots. The fundamental requirement of a reference material for chemical analysis is represented by its homogeneity. In the case of a reference site used within the framework of soil sampling intercomparison exercise, aimed at assessing different sampling procedures, the heterogeneity of the sampling target should be desirable. Acknowledgements This work was financially supported under the IAEA subprogramme ‘Supporting Quality in the Analysis of Terrestrial Environmental Samples’. The authors would like to thanks the teams of the 10 Countries selected to participate in the sampling exercise: Republic of Korea, Mexico, Syrian, Ukraine, Islamic Republic of Iran, Slovak, Slovenia, Hungary, Brazil and Lithuania. References Barbizzi, S., De Zorzi, P., Belli, M., Pati, A., Sansone, U., Stellato, L., Barbina, M., Deluisa, A., Menegon, S., Coletti, V., 2004. Characterisation of reference site for quantifying uncertainties related to soil sampling. Environmental Pollution 127, 131–135. Coughtrey, P.J., Thorne, M.C., 1983. Radionuclide Distribution and Transport in Terrestrial and Aquatic Ecosystems. A Critical Review of Data. A.A, Balkema, Rotterdam.
987
Desaules, A., Sprengart, J., Wagner, G., Muntau, H., Theocharopoulos, S., 2001. Description of the test area and reference sampling at Dornach. The Science of Total Environment 264, 17–26. HASL-300, 1997. The Procedures Manual of the Environmental Measurements Laboratory. In: Section 2.4, 28th ed., vol. 1. U.S. Department of Homeland Security, New York. International Organization for Standardization, 2005. General Requirements for the Competence of Testing and Calibration Laboratories. In: ISO/IEC 17025 Geneva. International Organization for Standardization, 2007. International Vocabulary of Metrology – Basic and General Concepts and Associated Terms, third ed.. In: ISO Guide 99 Geneva, Switzerland. International Organization for Standardization, 2006. Reference Materials-General and Statistical Principles for Certification. In: ISO Guide 35 Geneva, Switzerland. International Organization for Standardization, 2005. Statistical Methods For Use In Proficiency Testing By Interlaboratory Comparisons. In: ISO 13528 (E)Geneva, Switzerland. International Organization for Standardization, 2000. Uses of Certified Reference Material. In: ISO Guide 33 Geneva, Switzerland. Jac´imovic´, R., Lazaru, A., Mihajlovic´, D., Ilic´, R., Stafilov, T., 2002. Determination of major and trace elements in some minerals by k0-instrumental neutron activation analysis. Journal of Radioanalytical and Nuclear Chemistry 253, 427–434. Kaluzny, S.P., Vega, S.C., Cardoso, T.P., Shelly, A.A., 1997. S þ Spatial Stats User’s Manual for Windows and Unix. Springer, New York. Lisinger, T., 2005. Application Note 1 – comparison of a measurement result with the certified value. ERM, European Reference Material. Millard, S.P., Nagaraj, K., Neerchal, S.P.N., 2001. Environmental Statistics with S-Plus. CRC Press, USA. Navas, A., Machı`n, J., Soto, J., 2005. Mobility of natural radionuclides and selected major and trace elements along a soil toposequence in the central Spanish Pyrenees. Soil Science 170-9, 743–757. Spezzano, P., 2005. Distribution of pre- and post-Chernobyl radiocaesium with particle size fractions of soil. Journal of Environmental Radioactivity 83, 117–127. de Zorzi, P., Belli, M., Barbizzi, S., Menegon, S., Deluisa, A., 2002. A practical approach to assessment of sampling uncertainty. Accreditation and Quality Assurance 7 (5) 182–188. de Zorzi, P., Barbizzi, S., Belli, M., Ciceri, G., Fajgelj, A., Moore, D., Sansone, U., Van Der Perk, M., 2005. Terminology in soil sampling (IUPAC Recommendation 2005). Pure and Applied Chemistry 77 (5), 827–841. de Zorzi, P., et al., 2008a. Soil sampling strategies: evaluation of different approaches. Applied Radiation and Isotopes 66, 1691–1694. de Zorzi, P., et al., 2008b. A soil sampling reference site: the challenge in defining reference material for sampling. Applied Radiation and Isotopes 66, 1588–1591.