1
Objective: conducting monitoring programme in order to assess the water level
„How to attain the data confidence and accuracy levels as required by risk analysis, operation
programmes”
Introduction to the analysed issue
Monitoring is defined as a system of measurements, assessments and forecasts on an
environmental status, undertaken by authorised organizational units, focused on a diagnosis of such
environmental status targeted mainly at a proper definition of effective operations aimed at its
protection and possible improvements. Its purpose is to ease the origin (based upon preceding
characteristics of river basin and analysis of pressures and influences) of coherent and complete
image of ecological and chemical status within the river basin. Consequently, such an image will allow
for the creation of identical (or highly congenial) assessment of quality and level of water in Member
States, aimed at the initial preparation and subsequent introduction of reconstructive operations
undertaken in order to achieve, in the light of the main objective of Water Framework Directive, a so-
called “good status” (good potential) of surface and ground waters till 2015.
Water Framework Directive (WFD) has recommended introduction of a unified methodology of
water monitoring to the Member States, particularly defining:
a purpose of monitoring,
criteria of selection of monitoring points,
criteria of selection of monitored quality elements,
the frequency of proceeding a monitoring.
However, taking into account the differentiation of both natural physical and geological conditions
as well as anthropogenic pressures as existing between the Member States, the Directive has not
imposed strict conditions of proceeding a monitoring upon the Member States. It defines a flexible
hierarchical system adopted to the regional conditions, which shall, inter alia, embrace:
an assessment of deviation of the observed conditions in comparison with the ones that
normally would be observed in the reference state,
the natural and artificial physical variability of habitats,
the scope of natural variability and variability caused by an anthropogenic activity for all of
the quality elements, at all water bodies,
an interaction between surface and ground waters,
a full scope of potential interactions in order to enable the proper classification of ecological
status.
The incorporation of the above-mentioned issues to the assessment systems, as hitherto used
in the Member States, is definitely going to furnish the Commission with information on the water
ecological quality based on the proportions or fractions of the reference values. It is going to enable
the constant usage of existing domestic, national assessment system with simultaneous data passing
in accordance with the common European scale.
2
Monitoring, as assumed in the light of WFD’s provisions, is going to be a constant, cyclical
process embracing certain, constant and predefined elements, i.e. : data collection, reporting, data
analysis (including laboratory one), data processing, enhancement of monitoring network etc. A
method of data collection and subsequent processing of the latter plays a vital role as properly
conducted monitoring requires interrelation, involvement and co-operation of numerous bodies. A
definition of monitoring objective, reporting and usage of data is of equal importance to technical
issues, i.e. : metering network density, designation of examined factors, conduct method of
measurements and sampling due to the fact that mistaken identification of technical factors
whatsoever impinges on findings thereof resulting in duplication and accretion of errors.
Generally, the following are connected with the proper monitoring data confidence and accuracy level:
- the number and representation of monitoring points resulting from findings of analysis of an existing
water basin state,
- the frequency and sampling time within the monitoring points,
- the operations related to the sampling procedure,
- the operations related to the laboratory research and interpretation of findings.
Of necessity to the proper confidence and accuracy level is the ensurance of the adequate financial
means for research purposes, beginning from initial analyses and concluding with the laboratory
research. So-planned financial means constitute a preventive treatment to the mistaken identification
and an art of programming the proper, with regard to the scope and financial expenses, reconstructive
operations aimed at the fulfilment of Directive’s objectives.
Essential sources of perturbations and errors in measurements
The essential sources of incoherencies (in statistical approach – uncertainties), contributing
diverse perturbations and errors to the drawn findings, are:
an incomplete definition of an examined variable and its deficient realisation,
a type of a sample – a matrix and shape of the analysed, sampled material may affect the
recycling and instrument readings. The assessment of recycling may be falsified by other
concentration ratio of the examined material and potential tramp elements,
sampling, transport, storage and treatment of samples. In general, the laboratories avoid
including the sampling process into the analysis of measurement uncertainty, however, it is
that stage, until the ultimate sampling has been put across, that is mostly susceptible to the
highest variability related to the representation, natural variability, sampling conditions and
subsequent transport, maintenance and storage of samplings. The used reagents play a vital
role as well as the concentrations of solutions, including model ones, are not past all
3
recognition as the degree of accuracy is boundless – there always remains an element of
uncertainty related to the dilution or testing,
research personnel (analyst’s influence) – the readout’s errors may always occur (analogue
scale, colour comparison), different interpretations of findings, differences in skills and
knowledge,
an incomplete knowledge of impact of environmental conditions on the measurement
procedure or deficient measurement of parameters characterising those conditions. The
typical perturbations are input by f. in. temperature (its fluctuations may influence, f. in. glass,
liquids, reaction rate, at every stage of the research procedure) and humidity (influences the
weighing and condition of hygroscopic materials),
measuring instruments – usually they contribute to the uncertainty in a very insignificant
degree, however, at all times the influence of the following factors should be taken into
consideration: calibration (corrections, bias), thermal stability, natural surroundings variations
(temperature, humidity, external radiation), resolution (too small resolution results in
systematic bias, too large in random error), correctness (models uncertainty), incoherencies in
data processing (rounding), quality of supplementary equipment (f. in. developing dish of
photometer) and discrimination threshold,
an incomplete knowledge of ascribed values to models and reference materials as well as
assumed reaction model (the possibilities of deviations of stoichiometry, incomplete course of
reaction and side reactions should be taken into account).
an incomplete knowledge of constant values and other parameters, derived from the external
sources and used in the meanwhile of the data processing procedure,
calculations and data processing, oversimplifying approximations and assumptions used in
measurements methods and procedures, including inter alia: the selection of a model of
calibration curve (straight line instead of the curvilinear model), shortening, rounding etc.
an uncertainty resulting from the interaction of findings taking into account the systematic
elements,
spreading of values of an examined variable, obtained in the observation process repeated in
the apparently identical conditions,
adventitious effects – they may always occur, they relate to every single stage of research
procedure.
Therefore, a permanent assessment of the scheduled monitoring programme, embracing all
stages thereof and aimed at a constant improvement of the monitoring network as well as technical
issues related to its conduct, is indispensable.
The grounds for achievement of proper measurements findings
The starting point for monitoring planning is an assessment of both the existing standings and
possessed data inventory in order to evaluate which confidence and accuracy level may be potentially
obtained upon usage of that inventory. The fluctuation trends in the examined standings have to be
taken into consideration too. Then, the changes of and corrections to monitoring programmes should
4
be gradually introduced in order to attain the confidence and accuracy levels enabling the proper
assessment and classification thereof.
The properly organised and conducted monitoring of water state and quality should constitute the
basic instruments used in the planning process of river basins due to the fact that its intended purpose
is to provide information that are necessary for evaluation whether the Directive’s objectives have
been accomplished. In particular, it is going to enable the identification of water bodies subjected to a
non-accomplishment of such objectives, what will provide a proper identification and implementation of
scheduled reconstructive operations.
The types of monitoring and basic guidelines regarding its planning with regard to the
potential uncertainty of measurements
Having unified the monitoring systems in the Member States the WFD has assumed an
establishment of the following types of monitoring:
surveillance – defines the present water state, providing with information necessary for:
- a supplementation and approval of the procedure of assessment of
interactions, as stipulated by the annex no. II to the WFD,
- an effective and efficient planning of the future monitoring programmes,
- an assessment of long-term changes in natural conditions,
- an assessment of long-term changes resulting from anthropogenic activity,
operational – aimed at:
- defining of the state of those water bodies in relation to which it has been
agreed that the environmental objectives established for them are not going to
be accomplished,
- an assessment of changes whatsoever of state of those water bodies that are
a consequence of realisation of reconstructive operations’ programmes,
investigative – conducted where:
- the causes of non-accomplishment of environmental objectives have not been
clarified,
- from the surveillance monitoring it is obvious that the scheduled objectives for
water bodies are probably not going to be accomplished and the operational
monitoring for said water bodies aimed at the clarification of the cause thereof
has not been initiated,
- an objective is to define the magnitude of influence of adventitious pollution,
being supplemented by the additional requirements for the protected sites.
Some indicators and quality elements in the given water bodies will be characterised by a
significant variability resulting from:
natural factors,
anthropogenic activity,
5
errors accompanying the sampling process.
Due to the lack in the wording of WFD of “fixed” guidelines on the methods of conducting of
monitoring, the Member States are obliged to ensure the monitoring of a sufficient number of each
type of water bodies and, to designate the number of standings in relation to every water body that is
considered as being enough for the purposes of proper classification of their ecological and chemical
state. The question of frequency of monitoring may as well be approached elastically due to the
different variability of indicators and quality elements, in particular in relation to the ground waters.
In the monitoring methodical guidelines the conceptional models plays a vital role, as they
should be used as a basis for preparation and updating of monitoring programmes. Their role is
particularly significant with regard to the planning of monitoring of ground waters, where the risk
assessment of non-accomplishment of objectives may be based upon the conceptional
model/understanding of ground waters system and mutual interactions between the pressures and
that system.
The conceptional model is a simplified method of presentation or working description of real
operation of hydrogeological system taking into accounts the pressures and their impacts. The said
model embraces the hydrogeologists’ assumptions with regard to ground waters and:
constitutes a set of working hypotheses and assumptions,
focuses on those characteristics of the system that are the most representative ones with
regard to the required expectations or initial estimations,
is based upon the facts,
constitutes a resemblance of the reality,
should be supplied with documentary evidence in such a way that it could be verifiable using
the existing or new data,
a level of model’s minuteness of details is directly proportional to impediments for estimations
or expectations and for the potential consequences of errors in those estimations.
The usage of conceptional models leads to the “understanding” of ground waters systems, including
their interactions with the surface waters and ecosystems dependent on ground waters, with regard of
present or expected pressures and their impacts. As a result of the monitoring operations and
subsequent findings analyses the model should be currently “revised”.
The selection process of water bodies and the monitoring standings should take into account
the statistical methods¹ and ensure that the water state review is characterised by acceptable level of
accuracy and precision. The statistical methods are regarded to be the most universal and cheapest
instrument enabling the obtainment of credible and unquestionable findings, based upon the least of
numbers of undertaken research, involving the least efforts and means. Inclusion of the mathematical
statistics in order to assess the obtained findings and selection of the proper analytical methods
ensures:
the credibility of findings with the observance of required level of precision and confidence,
6
the optimalisation of costs of conducted researches and analyses
therefore enabling the accomplishment of objectives with the optimum number of standings and
frequency of measurements.
It has to be taken into consideration that the used calculation methods are the simplest possible
ones, therefore being apprehendable to the practitioners and leading to the clear, unambiguous
conclusions.
1
range – it is a difference between the maximal and minimal observed value,
The knowledge of statistical methods among the persons involved in preparation of monitoring findings
should be broad enough in order to:
enable the proper assessment of measurements,
provide with the grounds for decision making process regarding which of the methods should
be used for a given analytical objective,
enable the detection of errors related to the conduct of analyses,
enable the preparation of unfailing system of quality control.
1
Basic statistical notions used in monitoring planning and assessment:
Arithmetic mean (in other words ideal arithmetic or unweighted mean) – it is the sum of sampled observed
values divided by their numbers.
Geometric mean – it is a principal root equivalent to the number of set’s elements from the product of all
set’s elements.
Median – it is a spatial average because it divides the set into two equal parts: half of the units are of value
that is smaller or equals the median, whereas the remaining half –values equalling or greater than the
median.
Poisson’s distribution – a proper distribution for microbiological research, it refers to the random events or
randomly distributed objects in voluminal or superficial samples; it is designated by one parameter - mean
value (variance s² in that distribution equals its mean). By means of that distribution, the distribution of the
aggregate number of a certain occurrence with little probability is described in practice, with the great number
of independent experiments.
Measure of dispersion – defines the dissimilarities between the given research findings and the arithmetic
mean value, the most commonly occurring measures of dispersion are:
variance (s²)– it is the mean value of squares of deviances of certain characteristic’s values from
the mean value of collectivity. It constitutes a basic differentiation measure, defined likewise as a
quadratic mean from deviances of arithmetic mean; it constitutes mainly a base for determination
of standard deviation.
standard deviation (s. d.) – defines the average differentiation between the given characteristic’s
values from the mean value. It is a positive quadratic mean from the variations present in a given
sample. The standard deviation is a measurement of precision, a definition of a adventitious error
magnitude.
relative deviation standard (RDS) – it is a ratio of standard deviation s in a given sample from the
mean value (in the sample), it is a relative measure of distribution of findings’ values in the sample
coefficient of variation % (CV%) – it is a ratio of standard deviation s from the mean value
expressed in percentage.
7
In the process of planning of monitoring programme, the basic issues have to be predefined, such as:
the monitoring objective (f. in. the definition of water bodies chemical state, research on the
current trends, hypotheses analysis etc.)
the place of conducting of research,
the time of conducting of research,
the frequency of conducting of research.
Only the identification of the above-mentioned issues will allow for a definition of required, in order to
accomplish certain objectives, levels of risk, accuracy and confidence of findings, what consequently
shall enable the description of optimum monitoring programme, ensuring that:
the required objectives are going to be accomplished,
the sufficient number of standings shall be subjected to the monitoring,
the specified frequency and monitoring timeframe shall ensure the required accuracy and
confidence of findings, consequently allowing for the elimination of a potential risk,
the implemented monitoring shall be of optimum value taking into account economic approach
and shall be based upon credible scientific grounds.
The essence of risk and error with regard to the water monitoring
The adopted level of acceptable risk influences the scale of monitoring required for the definition
of water bodies state. In general, the lower the risk level of mistaken identification is allowed, the
greater the scale of monitoring required for the assessment of certain water bodies state (and
therefore the greater costs in overall).
In the most primitive understanding, the risk is described as a chance of appearance of an
undesirable occurrence, comprising two aspects: the chance (the probability of occurrence) and the
occurrence, which potentially may appear (the consequence). In relation to the monitoring, the issue of
risk is firmly accompanied by its wrong planning phase and subsequent errors resulting from initially
produced uncertainty sources, what consequently leads to the mistaken classification of waters. As a
result, the proper surveillance or operational monitoring project shall be oriented at decreasing the
acceptable level risks of wrong assessments (and at the same time mistaken classification of waters)
with the simultaneous selection of costs of its conduct.
Confidence means a long-term probability (expressed in percentage) that the real value of a
given statistical parameter (f. in. the population average) de facto lies within the calculated and
produced limiting interval, located round the ultimate finding, obtained as a result of a given monitoring
programme (np. test average).
Accuracy means a statistical uncertainty measure, equal to the half of the C% confidence
interval’s width, embracing two aspects;
correctness – refers to the degree of conformity between the arithmetic mean of a great
number of single research findings and a real value or an adopted reference value, expressed
by means of various load compounds by using the conventional measurement method,
8
precision - refers to the degree of conformity between the findings of a research; it is defined
by the standard deviation of repetition or standard deviation of reproduction.
For every single monitoring exercise, the evaluation error is a difference between the individual
outcome (as obtained from the samples) and a true value of measured (real) quantity, error regards a
single outcome. Accuracy therefore is defined as a level of evaluation error, which is attained or
corrected with the specified (high) participation of C% occurrences.
In the given case the two types of errors – error of I type – the risk of a false detection of
change defined by (100-C)% and error of II type, which is a risk that the difference is de facto going
to occur will not be detected by the monitoring programme.
Taking into account the causes of origin, as statistically approached, the following errors may be
distinguished:
random (accidental) – originated as a result of the presence of multiple accidental factors (f.
in. at research: a slight inexactitude in sampling, fluctuation of environmental conditions,
analyst’s tiredness). Those influences change at a random basis, they may not be foreseen
or compensated, however, they may be reduced by an increase in the number of
observations. Random errors may be maintained in restrictive numbers by observance of
strictly defined conditions required at research and genuine conduct of every stage of
research,
systematic – originated in the course of presence of always the same, hardly detectable
causes. Their avoidance may be attained by the usage of certified reference materials and by
systematic improvement of used methods and means of personnel trainings,
grave /serious/ – originate mainly due to the nature of human mistakes, improper functioning
of a given instrument or errors in its readings. They result in the obtained outcome being not
credible and consequently it shall not be relied upon. Though the grave mistakes are not
always so obvious, however, they are easy to detect, because the findings saddled with them
differ distinctively from the others (so-called outliers). If the sufficiently big number of
repetitions have been performed then, in order to assess whether there are findings saddled
with the grave error in a given set of findings, the statistical tests on elimination of outliers’
values have to be performed.
An adequate level of confidence and accuracy is partially going to influence the consequences of
conduct of erroneous water bodies assessment (f. in. an erroneous classification of water bodies) what
eventually will impinge on the improper method of designation of reconstructive operations, having
specified financial consequences.
The data derived from the monitoring standings will enable the conduct of an analysis on the risk
evaluation of the non-accomplishment of the WFD’s stipulated objectives for the certain water bodies.
The differentiation between the good and the moderate status, understood as an isolation of waters
that do not fulfil certain quality norms, should be established upon the comparison of the “consistence
of confidence level” with the relevant norm or threshold value. The assessment of a non-fulfilment of
objectives will have to take into account the allowed errors of I and II type. The error of I type occurs
when the water bodies, which in reality possess the parameters of good (satisfactory) state, are going
9
to be categorised as the worse than good (unsatisfactory) water state. The error of II type embraces
directly opposite circumstances – the waters of an unsatisfactory state have been classified as falling
into the category of satisfactory ones. The exemplary situation presented at the slide “The approach to
the risk of non-accomplishment of the objective” shows, upon assumption that the value of the
examined parameter amounts to 90-percentile, four possible variants: A variant - when the interval of
confidence is above the threshold value; B and C variants - when the interval of confidence and
threshold values “overlap” one another; and D variant – when the interval of confidence falls below the
threshold value. For the A and D variants the assessment is an easy one, whereas the complications
occur with regard to B and C variants. It is then possible to analyse the issue of risk of non-fulfilment of
the objectives, based upon:
- benefits-of-the-doubt – a given standing (water body) is positively assessed even if the examined
value marginally fails to fulfil the criteria of the positive assessment, however, provided that a part of
the confidence interval is classified within the good state,
- fail-safe – a reverse standing (water body) is negatively assessed even if the examined value
partially fulfils the criteria of the positive assessment, but a part of the confidence interval is classified
within the unsatisfactory state,
- face-value – a sampling error is being ignored and a positive (negative) assessment is entirely
contingent on a position of the examined value.
How to attain the proper level of confidence and accuracy?
The assumption of the proper confidence and accuracy level of monitoring will be associated with
the precise recognition of the existing conditions at a drainage area, including its characteristics and
present pressures and interactions, f. in.:
in a subdrainage area not subjected to any significant pressures, the narrow scope of
monitoring information enables the credible classification,
in a subdrainage subjected to significant pressures, however, little environmental sensitivity,
as above,
in a subdrainage subjected to various pressures and different environmental sensitivity to
them, only the broad scope of monitoring operations enables the proper classification.
Of course, the above-mentioned issue has been presented with a high degree of simplification
because the expenditure of monitoring operations is also determined by a difficulty of definition of the
consequences of significant pressures exerted on the water environment.
Should the monitoring objective be associated with the quality characterization (f. in. for
determination of the state of water bodies), the statistical objective is defined by reference to:
a parameter, which is to be evaluated (f. in. mean or percentile),
the requested accuracy (f. in. 0,5 mg/l, 20%),
the requested confidence level (f. in. 90%, 99%).
10
Then, taking into account the variability of an examined parameter in a given water body, the required
number of samples may be calculated. It has be noted that the same exact level of confidence may be
obtained by a change of the number of monitoring standings (what is related to a change of the
number of samples) with the observance of various levels of accuracy. For instance, the evaluation of
the average level of phosphate concentration in two types of rivers in Great Britain and Wales can be
presented for the confidence level equivalent to 90% and different types of accuracy involved, and,
respectively, the number of samples must equal to:
Accuracy little mountain little lowland rivers Confidence
rivers (number of (number of level 90%
samples) samples)
50% 13 39
25% 40 124
10% 214 675
Whilst conduct of a monitoring it has to be noted that the readings are never accurate. Without
the proper acknowledgement of that fact, there is no possibility of understanding of the measurement
process and accompanying occurrences, influencing an outcome. The readings, even subjected to
corrections, are always posing a high degree of uncertainty. The simplest example for statement of
uncertainty may be a multiple measurement of the known and invariable quantity, f. in. the weight of a
field-rock. The usage of a high-resolution measurement device may result in obtainment of divergent
findings. Therefore, it has to be assumed that the uncertainty is a parameter always associated with
the outcome, characterising the spreading of values, which may be in a justifiable manner attributed to
the examined quantity. It has been accepted in statistics that in order to characterise the uncertainty of
a measurement (u) the following are alternatively used:
standard deviation s or its multiplicity f. in. 2s, 3s…,
the width of the designated interval of confidence at a certain level of probability, generally
amounting to 95%.
The term of uncertainty expresses the scope of values that may be in a justifiable manner attributed to
the value of the examined quantity and refers to the general connotation of doubt, i.e. limited
knowledge of a certain quantity.
The uncertainty of a measurement does not, at any point, suggest doubts with regard to the
credibility of a measurement, but increases the confidence related to the credibility of an outcome..
In the analytical measurements the numerical values shall be ascribed to the identified
sources of uncertainty, they may be grouped – then their joint influence is solely put through an
examination. The evaluation of uncertainty by means of various methods is allowed. The evaluation
based upon the exact statistical calculations and the one using the alternative methods, including the
use of all available sources of information (including experts’ opinions) are considered to be
equivalent.
In the microbiological researches the exact statistical and metrological approach to the evaluation
and expression of the uncertainty of an outcome is however impossible. In the qualitative analysis the
11
outcome of research is an answer to the present/absent concentrations at a certain level (in a given
volume or mass).
PN-ISO 7218/1998 norm recommends that the uncertainty in microbiological research is defined
by designation of an interval of confidence in order to assess the validity of a given outcome and to
avoid its interpretation being too rigorous. The Poisson’s distribution shall be used for the purposes of
designation of borders of confidence intervals upon an assumption that microorganisms in the
examined samples are randomly distributed. That interval, however, will define the distribution of
microorganisms in a sample from the statistical point of view and will be designated for a limited
number of colonies, therefore the widths of that intervals are going to be identical for all laboratories –
the norms stipulate the ready-made tables presenting the confidence intervals for a given number of
colonies. However, there will always be other sources of variabilities, whose range will be dependent
on the precision of findings obtained in a particular laboratory. The cause for such a differentiation may
be deviations from the research techniques, in particular the errors at sampling and conduct of
dilutions. Therefore, it is recommended that the laboratories designate the real confidence intervals
taking into account their own, respective precision of researches.
Every findings of research on monitoring is subjected to the uncertainty, which can be illustrated
as an interval embracing a real value. The magnitude of that interval is dependent on many factors,
which, in case of monitoring, will be conditioned by:
1. the selection of the monitoring standing; that factor is of extreme significance due to the
spatial variability of water bodies, meaning f. in. the habitation characteristic for biological
quality elements,
2. the selection of adequate indicators of biological quality elements and interactions with the
physico-chemical indicators,
3. the overall number of samples dependent on the number of monitoring standings, frequency
of sampling (of particular importance with regard to the temporal variability of water state) and
a number of repetition of a given indication,
4. the method of sampling,
5. laboratorial analysis.
Basing upon the experiences of monitoring of drinkable water quality it has to be stated that the
uncertainty regarding the laboratorial analysis itself is generally of little importance in comparison to
the ones ensuing from the remaining above-mentioned factors. However, in order to maximally delimit
its value, and simultaneously fulfil the Directive’s requirements, the analyses shall be conducted by the
accredited laboratories for the conformity with the EN ISO 17025:2005 norm. Good laboratories must
be sure that the research process is reliable and carried out under proper supervision what ensures,
inter alia, the accreditation confirming that the laboratories:
are located within the proper premises, where the microbiologic air and surface purity is under
constant supervision,
12
ensure measure consistency, what manifests the usage of a proper class reagents as well as
a verified and calibrated equipment,
possess competent personnel,
conduct quality control, which allows for the detection of significant and adventitious errors,
participate in the interlaboratorial comparative researches, allowing for the detection of
systematic errors of a given laboratory (intercalibration),
use the validated methods, whose uncertainty is familiar thereto,
2
use control cards f. in. Shewhart’s one, for the purposes of research quality control
It is important that the laboratories that undertake to conduct monitoring researches possess the
accreditation for the designated indicators. The essential problem lies in the fact that currently there
are no established norms for designation of given indicators within the biological elements. Such
norms are nevertheless in the process of a preparation, but, at present, there is no information
whether the methods embraces thereby are going to be properly characterised by the definition of
limits of detectability, determinability, sensitivity, selectivity, specificity, linearity etc. The presentation
of such a characterization is crucial for the laboratories to assess whether they use a given method in
the proper way.
In addition to that, it is necessary to establish the interlaboratorial comparative researches or
certified reference materials, which will enable to assess the precision of a given laboratory.
A given research laboratory should always have and apply the procedures for evaluation of
uncertainties of readings. Even if the nature of a given research method hinders the exact metrological
and statistically justified calculation of the uncertainty, a laboratory in question should still at least
attempt to identify all possible sources of uncertainties, conduct their rational evaluation and ensure
that the used method of presentation of findings does not result in an erroneous impression regarding
the uncertainties. The above-described rational evaluation should be based upon the knowledge on
method’s possibilities together with the measurement scope and shall make use of f. in. earlier
experiences and validation data.
Accredited laboratories should conduct evaluation and expression method of uncertainties whilst
researches for the analytical measurement in conformity with the recommended guide, i.e.
EURACHEMIA/EA GUIDE 04/10 (being a supplementary to the PN/EN ISO/IEC 17025/2001 norm for
the accredited laboratories). Whereas in microbiological researches, where it is impossible to define
an exact statistical approach to evaluation and expression of uncertainties of readings, it is
recommended to identify and present the individual sources of uncertainties (f. in. a reagent, an
interpretation of readings undertaken by an analyst), in order to allow the latter to be controlled and
taken into consideration in the overall uncertainty of research. It is generally accepted that the
2
control card is a basic graphic statistical method used for the purposes of control of the quality of findings, it may be used
to control the stability of precision of a given method in time and to assess the stability of precision of particular analysts
13
evaluation of uncertainty of reading be based upon the analysis of repeatability and ability of
reconstruction of findings by multiple conduct of analyses. Ideally, the said evaluations should
comprise the systematic error (bias) assessed upon the findings obtained in the laboratories
specialised in the analysis of dexterity. Some uncertainty sources regarding pipetting, weighing and
influence of dilution may easily be measured and assessed – their influence is generally of little
significance in the overall uncertainty of research. Other uncertainty sources, as f. in. stability of
samples and their preparation, may not be directly measured and it is statistically impossible to assess
their influence on the uncertainty, however, their importance for the variability of findings need to be as
well taken into consideration. In addition to that, the accredited research laboratories should be able to
recognize the distribution of microorganisms within the examined matrixes (samples) and should take
in into account when conducting the subsamples in a laboratory. It is recommended that said factor be
included in evaluation of uncertainty because the uncertainty resulting from the decomposition of
microorganisms inside the matrix may not be treated as a determinant of a given laboratory’s dexterity
whereas it may be the findings of a research on a particular, usually scarce samples. A laboratory
should be aware of the frequency of occurrence of false, positive and negative findings of quality
research methods, used therein.
All available data, obtained during the monitoring conducted in conformity with the WFD’s
guidelines should reach the acceptable level of quality in order to ensure the comparable level of data,
of equivalent scientific value in all Member States. Therefore, all elements of the system of biological
and physico-chemical assessment shall correspond to the relevant international norms and
procedures. Obviously, the omnipresence of errors is inevitable, both in the phase of designation of
monitoring standings as well as in the sampling and samples’ analysis process, however, the ultimate
goal of a respective procedure is to ensure the quality and constant supervision over such errors. The
quality ensurance procedures may adopt a form of normalization of sampling and samples’ analysis
methods, repetition of analyses, double-checking the balance of samples’ ions and laboratories’
accreditations programmes. Consequently, f. in. for the samples to be taken at such a time that
reflects the water quality and its variability, the norms of ISO 5667 series have been adopted to clarify
the principles on sampling. On the other hand, for the laboratories the programme of quality
ensurance/quality guidance EN ISO 17025 has been introduced, in order to ensure comparability.
In the WFD the term “risk” is present with reference to two issues: a non-accomplishment of
environmental objectives and mistaken classification of waters. The proper level of confidence and
accuracy may partially influence the consequences of conduct of wrong assessment f. in. the mistaken
classification of water will result in the mistaken evaluation of costs involved in the realization of WFD’s
assumptions – achievement of a good ecological status (potential). The selection of the confidence
and accuracy level will delimit the line of tolerance of uncertainty (resulting from the natural and
anthropogenic variability) of the monitoring programmes’ findings. For the monitoring purposes, the
assessment of water state had to be undertaken in the first place, with a particular emphasis attributed
to the identification of those water bodies that are potentially subjected to non-attainement of the good
ecological status (or good potential). The initial state has been estimated by taking into account the
available data at that time, including the sampling process. Such estimation always differs from the
14
real values, i.e. the state that would have been estimated, should all the water bodies been subjected
to a constant, detailed monitoring and sampling procedure, taking into consideration all quality
elements. The WFD does not define the levels of confidence and accuracy of monitoring that should
be expected from monitoring programmes and water state assessments. Consequently, the definition
of too rigorous requirements would eventually lead to an increase in the level of monitoring and thus
the costs of the latter. It could have resulted in the outlays related to the conduct of monitoring being
utterly inadequate to the outlays necessary for the water state improvement, and the WFD stipulates
that the environmental objectives should be accomplished by means of the most costs-effective
operations.
The Member States are individually obliged to define the real level of confidence and accuracy in
the water basins management plans, to enable the conduct of a proper water state assessment and
their correct classification. It is going to be associated with an assessment and control exercised by
the other Member States, what will allow for the indication of real deficiencies and weaknesses,
leading to a correction of such monitoring programmes.
Summary
In order to produce an answer to the question „How to attain the data confidence and
accuracy levels as required by risk analysis, operation programmes” a targeted data confidence and
accuracy level has to be predetermined. Due to the fact that the Directive itself does not provide an
expressly stated answer to the posed question, a significance of magnitude of both confidence level
and data accuracy has to be taken into consideration. Firstly, the appellations herein stated –
“confidence level”, “accuracy” refer to the findings of monitoring, which is aimed at the collection of
information that subsequently constitute a ground for water status classification, which itself is a
ground for designation of operation programmes. In conclusion, the findings of monitoring shall be
precise to such a degree that would prevent the mistaken water status classification, and at the same
time, the mistaken operation programmes resulting in the unnecessary involvement of financial
means.
Generally, it may be pointed out that in order to prepare the appropriate operation
programmes the reduction of uncertainty of monitoring findings shall remain an objective, as it allows
for determination of magnitude embracing the real value with a higher level of accuracy. On the other
hand, however, the lesser uncertainty is equivalent to the greater financial means being shelled out for
proceeding the monitoring, therefore it has to be analysed, in what particular situation an involvement
of such greater financial means is justified. The answer seems easy – with regard to waters
categorised on the borderline between moderate and good status, i.e. those, which are endangered
with the risks of failure to attain Directives’ objectives.