Sampling: o,en the source of most uncertainty in analy6cal measurements Prof Michael H. Ramsey School of Life Sciences, University of Sussex, Brighton, UK [email protected]L.S. Theobald Lecture for 2012, Analy9cal Research Forum, GSK/University of HerEordshire, 8 th July 2013
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Sampling: o,en the source of most uncertainty in analy6cal measurements
Prof Michael H. Ramsey School of Life Sciences, University of Sussex, Brighton, UK [email protected] L.S. Theobald Lecture for 2012, Analy9cal Research Forum, GSK/University of HerEordshire, 8th July 2013
Overview – ques9ons addressed 1. Is uncertainty of measurement (U) – inevitable? evaluaHon beneficial? 2. Where does the measurement process start? 3. What are the sources of U? 4. What are methods for esHmaHng U from sampling & analysis
– with case studies in different media (e.g. soil, food) 5. How large is U, and what proporHon arises from sampling?
– for food, soil, gas, water -‐ with ex situ and in situ measurements on range of analytes
6. How much U is acceptable? – are measurements fitness-‐for-‐purpose?
7. How can U be changed to achieve fitness-‐for-‐purpose? 8. ImplicaHons for scope and role of analyHcal chemistry? 9. Conclusions
Uncertainty: inevitable? -‐ and evalua6on beneficial?
• It is well established that any analyHcal measurement result is only an esHmate of the true value of the analyte concentraHon.
• Knowing the REALISTIC value of this uncertainty gives vital benefits to users who interpret the measurements e.g. – to test a scienHfic hypotheses (e.g. H1: contaminaHon has declined) – to make a management decisions (probabilisHcally)
• e.g. ac9on required
Uncertainty of measurement (U)
‘An esHmate a]ached to a test result which characterises the range of values within which the true value is asserted to lie’ (ISO 3534-‐1: 3.25, 1993*) • e.g. U’ = 20% on measurement of 390 mg/kg Pb in soil
– Range of U’ = 390+20% =468 mg/kg to 390-‐20% =312 mg/kg
• We can never know the true value (could be 460 mg/kg) • Effect on comparison with threshold value of 450 mg/kg
i.e. Measured value 390 mg/kg is below the threshold value True value 460 mg/kg is above the threshold value
Measured value gives ‘false nega9ve’ classifica9on • All we need to know is how far from the truth we might be
– e.g. with 19/20 chance if being right = 95% confidence • How can we esHmate uncertainty of measurements – including that from
sampling? –described below
• Recent defini9on of U more cryp9c: ‘Parameter, associated with the result of a measurement, that characterizes the dispersion of the values that could reasonably be a]ributed to the measurand’ (GUM_JCGM_100, 2008)
� 390
468
312
x
Aims for U in Analy6cal Chemistry
• AnalyHcal chemists generally aim to:-‐ – reduce U of measurements, by – improving the precision, and reducing bias in analyHcal methods
• This aim has two problems: it assumes:-‐ – no other significant sources of U – a minimum level of uncertainty is always desirable
-‐ even when cost of pursuit may be disproporHonate to benefits produced -‐ review these assump9ons
Where does the measurement process start?
• The soluHon requires a fresh look at the ‘measurement process’ • Measurement process aims to:
– esHmate true value of analyte concentraHon (i.e. the measurand) in – some mass of material (i.e. the sampling target).
• Measurement process therefore starts when: – porHon of material (i.e. a sample) is selected – to represent the sampling target
• e.g. an area of land, or a batch of product – unless the sampling target is defined as the lab sample (in some regula9ons)
• Therefore U not just from analyHcal process – – must include contribuHons from primary sampling and – all of the physical preparaHon of the sample
• e.g. prior to receipt by the lab • More evident for in situ measurements
– e.g. placing PXRF implicitly involves taking a sample • analysts is also acHng as the ‘sampler’ • has implicaHons for the scope of analyHcal science
Sampling as part of the Measurement Process
Sampling
Physical sample preparation
Analysis
Sampling Target Collection of a single sample, or several increments combined into composite sample
Primary Sample Comminution and/or splitting
Sub-sample Further comminution and/or splitting
Laboratory sample
Physical preparation, e.g. drying, sieving, milling, splitting, homogenisation
Test sample Selection of test portion for chemical treatment preceding chemical analysis
Test portion Chemical treatment leading to analytical determination
Test solution Analytical determination of analyte concentration
Process step Form of material
Description of process step
More careful use of the word ‘sample’
TradiHonal view of measurement process (clear boxes)
Who is responsible for primary sampling? -‐ how is quality of sampling ensured?
• AnalyHcal chemists tradiHonally considered measurement process to begin when sample received by the lab
• So theory and pracHce of sampling lek to other people. • Different philosophy and approach to ensuring ‘quality’ has
therefore developed for sampling • e.g. ‘correct’ sampling giving rise to a ‘representaHve’ sample (Gy 1992*)
– not compaHble with approach used for chemical analysis. – e.g. ‘correct’ analyHcal methods are not assumed to give true values of analyte concentraHon.
• Needs new integrated approach for whole measurement process**
*Gy, P.M. (1992) Sampling of heterogeneous and dynamic material systems; theories of heterogeneity, sampling and homogenizing. Elsevier, Amsterdam pp 653. (page 567) ** Ramsey M.H. and Boon K.A. (2010). Geostandards and GeoanalyHcal Research, 34, 293-‐304. doi 10.1111/j.1751-‐908X.2010.00104.x
Sources of uncertainty: heterogeneity and ambiguity
• All sampling targets are heterogeneous to some extent, • All measurement procedures have some level of ambiguity
– e.g. exactly where to take a sample
• Heterogeneity and ambiguity combine and interact to contribute U in final measurement result. – e.g. analogy with green grains, then with le]uce, soil, bu]er
• Visual analogy of the effects of in situ heterogeneity – white grains contaminated with green grains – what is the concentraHon of green grains?
• expressed as % of all grains • what is the uncertainty of single measurements?
Pb heterogeneity by X-ray Micro-Probe = 51%
Analogy
Es9mate the concentra9on of ‘contaminated’ green grains in a batch
Sample area = 10% of popula9on area
Expect 4 green grains
= Batch = Popula6on
40 green grains
out of 400
True Conc. = 10%
Analogy to show U caused by heterogeneity in the sampling target
Sample #1
6 grains
out of 40
ConcentraHon
= 15% of grains
Single measurement
-‐ Posi9on of sample ambiguous
-‐ no idea of whether conc es9mate is reliable
-‐ no idea of uncertainty
Sample #2
3 grains
out of 40
Conc = 7.5% of grains
Two measurements
-‐ Can judge that conc es9mate is not very reliable
-‐ see uncertainty (U) is appreciable
-‐ What is U?
Sample #3
2 grains
out of 40
Conc = 5%
Sample #4
5 grains
out of 40
Conc = 12.5%
Es6ma6ng uncertainty(random component) Standard uncertainty (u = smeas) = 4.56% of grains
Expanded uncertainty (U, 2s for 95% conf) = 9.1% of grains
RelaHve expanded = 91% of measurement U first measurement = 15 +/-‐ 9.1 % of grains U = range within which the true value (10) lies = 5.9 to 24.1% -‐ Can be es9mated from just 2 observa9ons (from each of 8 targets) Only includes random component – not yet systema6c component, e.g. from sampling bias – e.g. grains not randomly distributed, so sample under-‐represents green grains (e.g. lower down in container)
c – U = 5.9%
• c = 15% of grains
c + U = 24.1%
Ð True value = 10%
U ' =100× 2smeasx
Obs Conc % of grains
1 15 2 7.5 3 5 4 12.5
10=x
Methods to es6mate uncertainty from sampling (& analysis)
• Various methods developed to esHmate measurement uncertainty arising from sampling
• Some empirical, some based on modeling, details in:-‐ • Ramsey M.H., and Ellison S. L. R.,(eds.) (2007) Eurachem/
EUROLAB/ CITAC/Nordtest/ AMC Guide: Measurement uncertainty arising from sampling: a guide to methods and approaches Eurachem ISBN 978 0 948926 26 6. h]p://www.eurachem.org/index.php/publicaHons/guides/musamp
• Empirical approach requires inclusion of 4 components:
Four empirical methods for es6ma6ng uncertainty including that from sampling
CTS = Collaborative Trial in Sampling ,
and SPT = Sampling Proficiency Test (e.g. Analyst, 2011, 136, 7, 1313-1321)
1 = Estimate analytical bias using CRM, 2 = Analytical bias partially or completely included where multiple labs involved
Simplest Empirical method is ‘Duplicate Method’ (#1)
1 m
The Duplicate Method • Define your sampling target and sampling protocol
• Use a balanced design
• Take a sample at the nominal sampling target. • Take a second sample displaced from original (in space or Hme) to reflect ambiguity in sampling protocol (& reflect analyte’s in situ heterogeneity) • Carry out duplicate analyses on both the sample duplicates • Take duplicate samples from 10% (≤ 8) of targets • EsHmate uncertainty components using ANOVA (Analysis of Variance) – usually Robust
Sampling Target
Analysis 1 Analysis 2 Analysis 1 Analysis 2
Sample 1 Sample 2
Sampling Target
Analysis 1 Analysis 2 Analysis 1 Analysis 2
Sample 1 Sample 2
Sampling Target
Analysis 1 Analysis 2 Analysis 1 Analysis 2
Sample 1 Sample 2
S1A1 S1A2 S2A1 S2A2
Case Study using duplicate method Nitrate Concentra6on in LeUuce
• Nitrate a potenHal risk to human health • EU threshold 4500 mg/kg for batch concentraHon • Current sampling protocol specifies taking 10 heads to make a single composite sample from each batch (in ‘W’ or ‘star’ design) European Direc9ve 79/700/EEC. OJ L 207, 15.8.1979, p26.
• Usual ambiguity in the protocol – e.g. where to start and orienta9on
• What is the uncertainty in measurements? • Is sampling protocol valid (suitable for rouHne use)? • Full details in Eurachem Guide (Appendix A1) & Lyn et al., (2007) ACQUAL, 12, 67-‐74
EsHmaHng U with Duplicate Method using Balanced Design
Target
Sample 1
Analysis 1 Analysis 2
Sample 2
Analysis 1 Analysis 2
At 10% of Sampling targets in whole survey n ≥ 8
- aim to represents these targets in general
Nitrate conc. in Duplicate Samples
Most analyHcal duplicates agree well < x0.1 (approx)
Sampling duplicates agree only < x0.2 (approx)
Range of conc. between batches x1.6 (approx)
Is level of Uncertainty OK?
<4500?
>4500?
Reliable decisions whether batch is > 4500 mg/kg?
S1A1
S1A2
S2A1
S2A2
3898
4139
4466
4693
3910
3993
4201
4126
5708
5903
4061
3782
5028
4754
5450
5416
4640
4401
4248
4191
5182
5023
4662
4839
3028
3224
3023
2901
3966
4283
4131
3788
mg/kg
ANOVA output for Uncertainty es6ma6on
CLASSICAL ANOVA RESULTS
Mean = 4345.5625
Standard Deviation (Total) = 774.5296
Sums of Squares = 12577113 4471511 351320
Between-target Sampling Analysis
Standard Deviation 556.2804 518.16089 148.18063
Percentage Variance 51.583582 44.756204 3.6602174
ROBUST ANOVA RESULTS:
Mean = 4408.3237
Standard Deviation (Total) = 670.57617
Between-target Sampling Analysis Measurement
Standard Deviation 565.39868 319.04834 167.94308 360.5506
From ROBAN sokware, free download from: h]p://www.rsc.org/Membership/Networking/InterestGroups/AnalyHcal/AMC/Sokware/ROBAN.asp
Uncertainty es6mate for LeUuce • Using Analysis of Variance (ANOVA)
• on duplicate measurements • Robust staHsHcs to accommodate outlying values
• Gives repeatability precision smeas = u = 361 mg/kg • Expanded U’ = 16% • U from analyHcal bias (from CRM/ or spike)
• can be added – not staHsHcally significant in this case
• Does not include U from any sampling bias – Can be included using values from Sampling Proficiency Test (SPT) – with >8 organisaHons
U ' =100× 2smeasx
U ' =100× 2*3614408
U es6mates for range of foods/analytes
0
10
20
30
40
50
60
70
80
Infan
t milk
(Pb)
Milk (a
dded
wate
r)
Spread
able
fats (
fats)
Butter
(mois
ture)
Sausa
ges (
meat)
Glassh
ouse
lettu
ce (N
itrate)
Apples
(Prop
argite
)
Infan
t wet
meals
(Cd)
Strawbe
rries (
(Myc
lobuta
nil)
Pistac
hio nu
ts (A
flatox
in)
U'(%
)
Sources of U – sampling or analysis?
0
10
20
30
40
50
60
Infan
t milk
(Pb)
Milk (ad
ded w
ater)
Spread
able
fats (
fats)
Butter
(mois
ture)
Sausa
ges (
meat)
Glassh
ouse
lettu
ce (N
itrate)
Apples
(Prop
argite
)
Infan
t wet
meals
(Cd)
Strawbe
rries (
(Myc
lobuta
nil)
Pistac
hio nu
ts (A
flatox
in)
U %
(95%
con
f)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Ratio
of v
sam
p an
d vm
eas
Usamp%Uanal%Ratio
Sampling generally dominated, but someHmes chemical analysis (e.g. if conc near DL) Recent meta-‐analysis of 128 analyte/food combinaHons has show that: Median U’ = 16%, with 70% of the cases having sampling as main cause (FSA E01088)
Alterna6ve methods to es6mate U
Unbalanced design -‐ 33% less cost* Simplified design -‐ use external esHmates of analyHcal U • Include sampling bias in U using mulHple samplers/protocols (Method 4)….
* Rostron P. and Ramsey M.H. (2012) Accreditation and Quality Assurance: J. for Quality, Comparability and Reliability in Chemical Measurement 17, 1, 7-14. http://link.springer.com/article/10.1007%2Fs00769-011-0846-2
• Modify duplicate method to reduce the cost – using either:-‐
Including Sampling Bias in Es6mate of Uncertainty using Sampling Proficiency TesHng
Sampling Proficiency Test. Ramsey et al., Analyst, 2011, 136, 1313-‐1321
U% on concentra6on from SPT on moisture in 20t fresh buUer:-‐
≈ Duplicate Method U% = 0.39 %
SPT Includes bias between-‐samplers (n=9) U% = 0.87% -‐ larger x 2.2
Two samplers had poten6ally non-‐proficient z-‐scores (z > 3)
Samplers; I (z = -‐3.2) & G (z = +3.5)
If only proficient samplers, U% = 0.69%
-‐ U s6ll larger x 1.8
one laboratory under repeatability conditions.15 The former
approach could be called more precisely a ‘measurement profi-
ciency test’ (MPT), as each participant undertakes the whole of
the measurement process. The latter approach eliminates the
effect of any analytical bias between the participants, helping the
performance score to reflect only the participant’s sampling
performance. However, the centralised analysis in the latter
approach will also tend to reduce the estimate of the total
measurement uncertainty to an unrealistic level, unless it is
dominated by the sampling component, or the analytical
component is added independently.
Specific experimental design of SPT on moisture in butter
The determination of moisture in butter is an important
parameter considered by the UK Rural Payments Agency for the
acceptability of butter, either for the intervention scheme26 or as
‘butter for manufacture’.27 A study of previously frozen butter28
showed that the measurement uncertainty from sampling was
low for moisture (Usamp ! 2.5% relative or 0.39% m/m), due
primarily to low levels of heterogeneity in the spatial distribution
of the moisture in the butter. However, the closeness of the
measured concentration (15.76% m/m in this case) to the legal
limit (16% m/m) makes the sampling an important source of
uncertainty. This is because it contributes 96% of the measure-
ment uncertainty, and can therefore lead to misclassification of
the butter against the legal limit.
The sampling target selected for this SPT was broadly defined
as a batch of 20.1 t of fresh butter made in a creamery in SW
England. The experimental design used is essentially the general
one described in Fig. 1, with 9 participant samplers (identified by
letters A–I), and full details are reported elsewhere.25 The butter
is packaged in the normal way, in around 804 cardboard boxes
each containing a 25 kg block wrapped in a blue polythene sheet,
and stored in a cold room chilled to 0–5 "C, on separate pallets to
allow access to all boxes. This SPT design generally meets all of
the selection criteria (Table 1), but with two partial exceptions.
The 100 g samples were taken with a metal ‘trier’, that was a half-
round tube around 30 cm long, with a 2 cm internal diameter and
a handle at one end. The trier was inserted into the selected 25 kg
block of butter at an angle of around 45" (variable between
samplers 0–90" observed), twisted and then removed. The top
and bottom portions of the resultant core (#20 cm, but variable)
were rejected and the rest of the core transferred into a plastic pot
(volume around 500 mL). Two cores were taken from each 25 kg
block, to give a sample mass of around 100 g in each pot. Each
sampler sampled six blocks, randomly selected from the whole
batch, that generated a sample mass of around 600 g. This
composite sample represented the 20.1 t of product, giving
a sample ratio of 0.003% m/m (600 g/20 100 000 g). Criterion 1
was therefore satisfied. Each sampler took around 75 min to
execute the sampling protocol, and the whole SPT took 5 hours
to complete (criterion 8), during which time there was no
detectable change in the concentration of the analyte (criterion
3). The variability in the position and angle at which the trier was
inserted, and in the length of the discarded portion of the core,
left room for improving the consistency of the sampler’s
performance (criterion 7).
The two criteria that were not fully met included the sufficient
heterogeneity of the material (criterion 4). The heterogeneity at
around 0.08% RSD was low as expected, but the precision of the
gravimetric method (0.17% RSD) was not low enough to quan-
tify this heterogeneity accurately, and should ideally have been
less than 10% of that value (i.e. 0.008% RSD). The analytical
precision was low enough to quantify the between-sampler effect
(0.39% RSD, 0.78% at 95% confidence). This made the presence
of non-proficient samplers or bias between samplers detectable,
but potentially hard to quantify accurately (discussed below).
The independence of the samplers (criterion 10) was the second
criterion that was not fully met. Although the samplers were
mainly from different local authorities, they had received some
degree of common training, causing a possible lack of indepen-
dence. This possibility was supported by the observation that
many of the samplers adopted a very similar protocol that was
implemented in a very similar way, despite the ambiguity in the
written version of the protocol.29 The importance of the selection
of participant samplers with ‘required skill’ (criterion 6) was
investigated by including one non-professional and relatively
untrained sampler (identified by letter I). This sampler was
included to see if the SPT could detect this nominal lack of
sampling proficiency.
The routine sampling protocol normally employed in the
factory was slightly modified for all of the samplers, to enable the
more accurate evaluation of the uncertainty and the scoring of
the participants. Each sampler took two sets of six increments
from the whole batch, the second set constituting the duplicate
sample. This was different from the routine protocol in which
one set of six increments was taken, and then divided into two
sets of three increments that were combined physically to
represent the two halves of the whole 20 t target. Routinely, the
position of the split between the two halves of the batch was
made at a random point based upon the location of the third and
fourth increments. This latter routine practice caused some
ambiguity in the definition of the sampling target, which was
resolved in the data processing, as explained below. This illus-
trates that although methods for the evaluation of uncertainty
can generally be applied without changing the essence of the
sampling protocol, there are some situations where the routine
Fig. 1 Experimental design typically applied in sampling proficiency
tests (SPTs), adapted from35.
1316 | Analyst, 2011, 136, 1313–1321 This journal is ª The Royal Society of Chemistry 2011
Benefits of Knowing Uncertainty
Benefit #1:-‐Improving reliability of decisions – e.g. for poten9ally contaminated lekuce – Risk assessment:-‐
Nitrate concentra6ons (mg kg-‐1) for S1A1 (rouHne sample) with the associated measurement uncertainty (calculated from U = 16.4%).
e.g. Target F value of the measurand (or true value) between 4332 mg kg-‐1 and 6032 mg kg-‐1, = ‘Probably Contaminated’, compared with threshold (T) 4500 mg kg -‐1
ProbabilisHc classificaHon has only one batch definitely uncontaminated (G), whereas determinisHc classificaHon has 4 batches uncontaminated (A, B, G & H)
Only one batch (C) is Definitely Contaminated – posiHon taken by some regulators (e.g. EC)
Benefit #2 Judging fitness-‐for-‐purpose in valida6on
• Equivalent as asking ‘How much U is acceptable?’ • One opHon -‐use the opHmised uncertainty (OU) method* • Balance the cost of measurement
-‐ against the cost of making incorrect decisions -‐ Knowing sampling and analyHcal components -‐ judge whether either is not FFP -‐ therefore where improvements/ increased expenditure required
* Lyn, J.A., Ramsey, M.H., and Wood, R. (2002) Analyst, 127, 1252 – 1260
Acceptable level of Uncertainty?
0
200
400
600
800
1000
1200
1400
1600
0 200 400 600 800
Uncertainty mg/kg(smeas)
Cos
t (£)
(E
xpec
tatio
n of
Los
s)
Actual U
Optimal U
Cost of falsely discarded batches
Cost of lowering U on measurement
Actual U ~2x higher than OpHmal
Judging FFP for range of foods • Applied Duplicate Method to sampling 10 different foods/analytes, range of costs,
Ra6onal basis for alloca6on of finance, to:-‐ 1. Measurement as a whole, and 2. ApporHonment between sampling and analysis Allows achievement of opHmal uncertainty -‐ and fitness for purpose of whole measurement
method – e.g. lekuce
How can U be changed to achieve FFP?
• Graph for le]uce shows that U is too high – need to reduce it • Need to know source of U
– from sampling or from chemical analysis? – Duplicate Method + ANOVA -‐ tells us sampling 78% of U
• We need to reduce the U by a factor of 2 (360→180)
• Sampling theory predicts (e.g. Gy’s) increasing sample mass by factor of 4 (= 22)
• ReducHon in U was achieved in pracHse → FFP – By taking composite sample with 40 heads instead of 10 – Make whole method valid (i.e. suitable for rouHne use) – Full details in Lyn et al., (2007) ACQUAL, 12, 67-‐74
Benefits #4 of Knowing Uncertainty
Provides tool for monitoring Quality of Sampling -‐ Be]er than assuming:-‐
-‐ ‘correct’ sampling protocol was ‘correctly’ implemented -‐ resulHng in a ‘representaHve’ sample
-‐ Gives quanHtaHve esHmate of sampling quality -‐ Bring sampling within similar QC/QA to analysis -‐ Other tools to improving quality:-‐
• When U’ is very large (>30%?), we need be]er ways of expressing it – If U range goes < 0, can’t be the ‘true value’
• New ways of expressing in situ heterogeneity1 use heterogeneity factors (HF): – e.g. HF = 2, gives UCL = c*2, LCL = c/2
• More accurately match the real frequency distribuHon – oken ~ lognormal for very heterogeneous targets
• Can be modelled as a funcHon of spaHal scale1
• – also express uncertainty as a factor (UF), not as U’%?
1 Ramsey M.H., Grace Solomon-‐Wisdom G.O. and Argyraki A. (in press) EvaluaHon of in situ heterogeneity of elements in solids: implicaHons for analyHcal geochemistry. Geostandards and GeoanalyHcal Research (GGR-‐0236.R1)
U in 6 Ex situ cont. land inves6ga6ons Site# Source of
Pollution End use Prime
contaminant Sampling method
U’% %Prop Samp
PropAnal
1 Mine Sn/Cu Housing Arsenic Trial Pit 65 94 6
2 Gasworks waste
Public access
Lead Trial Pit
51 93 7
3 Infill after WWII Bombing
Private gardens
Lead Window 25 99.9 0.1
4 Gasworks Commercial Dev.
Total PAH Trial Pit
186 <1 >99
5 Railway sidings
Public access
Copper Trial Pit
158 Simplified design - not separated
6 Ex-firing range
Housing Lead Hand auger 75 72 28
Boon K.A., Ramsey M.H. and Taylor P.D. (2007. Geostandards and GeoanalyHcal Research, 31, 237-‐249 (doi: 10.1111/j.1751-‐908X.2007.00854.x )
Median U’ is 70%. Sampling typically contributes > 90% of U
U of On-‐Site vs Ex-‐situ Measurements on Contaminated Land
Site Location
Key Contaminant
Anal Method
On site? In or Ex
situ?
Umeas (%) random
Usamp (%) Prop of U from
sampling
NW England Arsenic PRXF On site 179 178 99
NW England Arsenic HG-‐AAS Ex situ 110 110 100
S. Coast TPH Site-‐LAB On-‐site 42 31 54
S. Coast TPH GC-‐FID Ex situ 82 62 57
NW England Arsenic PXRF In situ 154 149 93 NW England Arsenic HG-‐AAS Ex situ 110 110 100
S.Wales Pb PXRF On-‐site 62 57 85
S.Wales Pb ICP-‐AES Ex situ 54 39 53
S.Wales Zn PXRF On-‐site 49 45 87
S.Wales Zn ICP-‐AES Ex situ 30 29 91
S.Wales TPH Site-‐LAB On-‐site 46 34 53
S.Wales TPH GC-‐FID Ex situ 97 91 89
Median On-‐site 49 45 85 Median Ex-‐situ 89 77 90
Median Both 62 57 87
Duplicate method equally applicable to on-‐site* and in situ** measurements U of on-‐site measurements are similarly dominated (~85%) by sampling -‐ Even more so for in situ measurements (93%) *Boon K.A. and Ramsey M.H. (2012) Science of the Total Environment h]p://dx.doi.org/10.1016/j.scitotenv.2011.12.001 **Ramsey M.H. and Boon K.A. (2011) Applied Geochemistry h]p://dx.doi.org/10.1016/j.apgeochem.2011.05.022
U from sampling other media
• Duplicate method applicable to many different media & analytes • Levels of U’ vary greatly from <4% to >72% • ContribuHon from sampling oken dominates (median 92%), but not
always in situ γ-‐ray spec of Cs-‐137 at nuclear site – integrates heterogeneity of 5 ton sample
• Ground-‐Water – heterogeneity sHll dominant • U esHmates ignoring sampling will be too low, and not improved by
lab acHon 1. Squire, S and Ramsey, M.H. (2001). J. Environmental Monitoring, 3, 3, 288 -‐ 294 2. Ramsey M.H., and Ellison S. L. R.,(eds.) (2007) Eurachem/EUROLAB/ CITAC/Nordtest/ AMC Guide: Measurement uncertainty arising
from sampling: a guide to methods and approaches. Eurachem, p59 3. Rostron, P., Heathcote, J.A., Ramsey, M.H. Journal of Environmental RadioacHvity (submi]ed).
Sampling Target Analyte Mode U' Prop Samp Ref Landfill Gas
Methane in situ 3.6 92.0 1 Oxygen in situ 4.8 96.0
Groundwater Fe-‐dissolved ex situ 10.1 97.0 2 Nuclear Decom. Site
Background Cs-‐137
in situ 42.6 0.0 3 ex situ 47.4 84.6
Moderate Cs-‐137
in situ 12.6 65.5 ex situ 72.6 99.7
FFP for the 6 Cont. Land Inves6ga6ons
• Can judge FFP by closeness to unity of raHo of actual/opHmal Umeas • Measurements are ~FFP in one case (Site 2), and be]er than FFP at Site 3 • Most measurements not FFP (have more U than is opHmal) -‐ by factor of 2.4 • Because sampling dominates U, improvement factor for sampling similar 2.8 • Usamp typically 62%, but opHmal (by OU/OCLI method) = 24% (range 16-‐54%)
• SituaHon presents a golden opportunity for measurement scienHst to broaden their role.
• Sampling personnel are usually poorly trained to tackle these measurement issues, but
• AnalyHcal chemists could usefully extent their exisHng experHse to consider the whole measurement process.
• Also applies to in situ and on-‐site measurements • Broaden analyHcal science to include design of primary sampling
Conclusions 1. Sampling is part of the measurement process 2. Uncertainty of measurement (U) needs to include the contribuHon from
primary sampling and sample prep – both random & systemaHc components 3. We now have methods to esHmate U (from all sources including sampling)
– Applicable to any measurement (& sampling) protocol, in any medium for any analyte 4. EsHmated values of U are:
– Inevitable, and usually larger than Uanal expressed by the lab – Usually dominated by the contribuHon from sampling (e.g. ~ 80%) – Beneficial to the users of measurements for more reliable interpretaHon
5. The fitness-‐for-‐purpose of measurements can be: – Judged by comparison with the opHmal level of U (= minimal overall cost) – Only judged realisHcally when the contribuHon for sampling (& sample prep) included – Oken achieved most cost-‐effecHvely by improving the sampling (rather than analysis)
• e.g. by increasing the number of increments in a composite sample – A be]er way of judging data quality requirements than arbitrary AQC targets
6. The scope of analyHcal chemistry (and role of measurement scienHsts) should be widened to include the process of primary sampling….
7. Enables unified approach to ensuring measurement quality (based on U)
Acknowledgements Co-‐workers Ariadne Argyraki Jenny Lyn Paul Taylor Katy Boon Peter Rostron Sharon Squires Jong Chun Lee Jacqui Thomas Grace Solomon-‐Wisdom Ilaria Palestra
Collaborators Mike Thompson Steve Ellison Roger Wood Andrew Damant Phil Po]s Mike Gardner Anna Whi]aker John Heathcote Members of Eurachem & RSC/AMC Working Groups on U from Sampling