6. Monitoring - Design Requirements
Once the basic requirements in terms of water type, matrix and determinand have been established, then other aspects of monitoring programmes can be designed. These monitoring requirements can be split into three aspects:
- analysis; and,
Barriers to harmonisation can be introduced at any one of these stages and can arise either because requirements differ (i.e. conflict) or because requirements are not clearly specified (i.e. are weak).
This Section analyses monitoring requirements with the aim of identifying both types of barrier. The specific aims are to:
- provide an indication of the degree to which monitoring requirements are specified;
- determine to what extent requirements are repeated in different commitments (data overlaps) or conflict;
- assess whether the most common requirements (standard requirements) meet the objectives, if not why not (i.e. data deficiencies);
- assess whether those requirements which differ from the standard are necessary to meet the specific objectives of that commitment or whether the standard requirements would be satisfactory;
- recommend areas for possible harmonisation or clarify the reasons where harmonisation is not possible; and,
- assess the extent to which directives do not specify adequate requirements to ensure harmonised implementation.
As for Section 5, summary tables of specific requirements described in this Section are given in Appendix E.
Sampling strategies may vary according to:
- sampling location;
- frequency; or,
- sample number, type and volume.
6.1.1. Sampling location
Sampling location will be considered when competent authorities design their own national monitoring networks. They will have to ensure that the networks not only meet their national and international obligations, statutory or otherwise, but also that they meet other needs and objectives. For example, general surveillance data from a larger proportion (compared to that required by international statutory requirements) of the total national water resource may be required. Operational data, often at a sub-catchment level, may also be needed, for example, to monitor the impact of specific discharges on water quality. There will be obvious benefit, where possible, in replicating the purpose of sampling points and also in usage of the information obtained. It is likely, therefore, that monitoring networks associated with directive and international obligations will not represent the total monitoring networks of individual nations. For surveillance purposes, sample sites may be located in relation to changes in water quality, perhaps associated with point discharges or tributaries. Where there are gradual rather than discrete changes of quality, for example along a river, the optimum number of sample locations needed to define overall quality would be quantified through an assessment of the spatial and temporal variability of the determinands of interest in that river.
The most specific requirements in terms of named water bodies and measuring stations are in the Exchange of Information Decisions. These rivers are nationally significant rivers and lakes and as such are quite likely to be sampled for other national and international obligations (e.g. Rhine and Elbe Conventions).
In other directives, sampling location is generic rather than specific and is directly related to the type of directive (purpose) (Table E.1). Use-related directives, and the monitoring requirements made in them, apply only to waters designated for the protected use. Therefore, the degree to which the requirements overlap in terms of sampling location depends on the extent to which the uses overlap. Since water uses are designated by Member States this varies from country to country. Similarly, industry sector and substance directives aim to control discharges, either those from a specific industry or containing certain dangerous substances, and sample locations relate to specific areas or zones impacted by discharges (e.g. the Titanium Dioxide Directive). The monitoring requirements made in these directives apply only to waters affected by discharges covered by the directives (the Titanium Dioxide Directive also refers to sampling in a neighbouring area unaffected by the discharge). The overlap in sample location requirements is, therefore, determined by the distribution of discharges and again varies from country to country.
Thus the scope for overlap of sampling locations between directives appears to be limited to areas where use and designation overlap, which is probably a limited circumstance, for example where drinking water is abstracted from a river which is also a designated freshwater fishery, which is also stipulated in the Exchange of Information Decisions and receives a discharge of a regulated dangerous substance. Locations are, however, cross referenced between directives. For example, the Nitrates and the Sampling and Analysis Directives both refer to locations specified in the Surface Waters Directive.
Less specific is one of the requirements of the Nitrates Directive where the general eutrophic states of surface waters have to be periodically reviewed by the national competent authorities. The requirements for the proposed Ecological Quality Directive are also less specific since monitoring will be required in a representative number of all surface waters, this will again be defined by Member States rather than be specified by the Commission. For many countries this will require an increased sampling load in terms of numbers of locations (and also numbers and types of determinands).
Many of the directives allow the competent authority of each Member State to make decisions on such details as the exact sampling point, the distance from this point to the nearest point where pollutants are discharged and the depth at which the samples are to be taken. It is indicated that these sampling points should be fixed on the basis of local environmental and hydrological conditions. This is so for the Titanium Dioxide Directive where the depth of sampling and distance from the discharge is also decided by Member States but the same sites and depths should be used in all surveys, including in relation to physical and temporal conditions such as relativity to high tide and tidal coefficient. Once the fate and behaviour of the effluent is known and the effects have been established, and as long as there is no deterioration, then there is scope for Member States to use a lower sampling frequency than specified in the directive. This proviso is absent from the Dangerous Substances Daughter Directives.
The proposed Landfill Directive has a requirement for the monitoring of leachate and run-off water based on a 10 litre global sample representative of the average composition of the waste deposited. In addition surface water run-off composition should be monitored monthly during the operating phase and every six months in the after-care phase. This should be undertaken at not less than two points, one upstream and one downstream of the landfill. At least one monitoring point should also be set up in the groundwater inflow region (0 level) and two in the outflow region.
Only two directives make specific requirements for sample depth: the Bathing Water Directive and Titanium Dioxide Directive. The former specifies 30 cm as the sample depth for all determinands other than for mineral oil for which surface sampling is required. The Titanium Dioxide Directive indicates that for discharges to fresh surface waters samples should be taken at a depth of 50 cm below the surface if possible. The requirements are not so definitive for discharges to salt water where samples should be taken at the same depth on each sampling occasion. For the Shellfish Waters Directive sample depth should be fixed by the Member States and similarly for the Freshwater Fish Directive sample depth should be defined by the competent authority. Sample volume is not specified in any directive. Thus sample number (e.g. replicates at a site), depth and volume are generally determined at Member State level and are, therefore, open to interpretation.
As has already been stated, the choice of sampling location is for some directives related to areas designated by the Member States rather than by the European Commission. It is unlikely, therefore that, for some directives, a comparison of quality across Europe of these designated waters will give a complete picture of quality because the degree of comparability will depend on the interpretation of the designation rules and national differences of how these are implemented. For example, it is likely that many more water bodies will be able to sustain fisheries than are covered by the directives. In terms of the Bathing Water Directive there is generally a wide spatial coverage across Europe of quality in designated salt water areas. If, however, the microbiological quality of inland waters was to be compared across Europe, there would be some major spatial gaps as some Member States do not designate inland waters as bathing waters. Therefore, the degree of coverage that water quality data encompasses within each country will be determined by national designations and the prevalence of the industries that are required to be regulated. The proposed Ecological Directive should, in the future, mean that a far greater proportion of a nations water resources are monitored than is required at present. Better spatial comparison, at least for ecological quality, should therefore be possible in the future.
For international agreements sample location is generally related to the purpose of the agreement (e.g. monitoring transboundary water transfer) often being at designated or fixed sites. For example, designated sites are specified for the sampling of water, biota and sediment under the Helsinki Convention, and at fixed stations under the Rhine Convention, the Protocol for Technical Co-operation between Greece and Bulgaria, and the Treaty between Austria and Hungary on Water Economy. Other agreements are less specific about sampling location, perhaps being determined by the research or information needs of the signatories or research programme (e.g. the North Sea Task Force). For many agreements signatory states also have to decide upon exact locations, perhaps within guidelines provided by the relevant Commission, for example, as in the quantification of riverine loads for the Paris Commission.
6.1.2. Sampling frequency and period
The requirements for sampling frequency vary considerably from directive to directive (Tables E.2 and E.3). Only 12 directives recommend minimum sampling frequency. Most directives allow the relevant authorities in each Member State to establish monitoring requirements or to provide recommendations to ensure that samples are representative of the quality of the water and sufficient to show any changes in the aquatic environment, taking into account in particular natural variations in the hydrological regime. For those where frequency is stipulated the requirement is quite specific to the objective of the directive, for example the monitoring of the quality of bathing water during the bathing season. Those which do not specify any monitoring frequency include the Mercury Directive and other Dangerous Substances Daughter Directives.
Sampling period is not usually specified or, if it is, the interpretation of its definition can give rise to differences between countries (for example bathing season). These imprecise requirements can give rise to different interpretation.
Sampling frequency requirements also vary within directives reflecting:
- water type (for example, the Titanium Dioxide Directive requires three samples per year for surface waters but only 1 per year for groundwater);
- matrix type (for example, the Dangerous Substances Directives require monthly sampling of the water column but only annual sampling of sediment);
- determinand (for example, under the Freshwater Fish Directive it is required that temperature be measured weekly whilst other determinands such as dissolved oxygen and ammonia are measured monthly);
- extent of water use (for example, the population supplied under the Surface Water Directive); and,
- quality of the water (several directives allow for a reduced sampling frequency for some determinands if the quality of waters is higher than prescribed in the directive, and where there is no pollution and no risk of deterioration e.g. the Bathing Water Directive and the Titanium Dioxide Directive).
The Sampling and Analysis Directive has a very complex specification for sample frequency. It is dependent upon the population the water supply is to serve, the category of the determinand (I, II or III) and the level of treatment required (A1, A2 and A3). Stipulated sampling frequency varies from a minimum of once a year for category I determinands in A1 waters serving a population of 10,000 to less than 30,000, to 12 times a year for A3 waters serving a population of >100,000 for category I determinands. For some categories of determinand, population and water treatment classification, it is the role of the competent national authority to specify the sampling frequency.
The Mercury, Cadmium and Hexachlorocyclohexane Directives do not set a standard sampling frequency but specify that this must be sufficient to show any changes in impact. For the Freshwater Fish Directive, sampling frequency is specified for four of the eight determinands (petroleum, pH, total ammonia and total zinc).
The Bathing Water Directive also has a multifactorial approach in determining sampling frequency and depth of sampling. For many determinands the sampling frequency is fortnightly (faecal coliforms, mineral oils and surfactants) during the bathing season. Others should be sampled if, once checked, it is shown that the substance may be present or the quality deteriorated. These determinands include faecal streptococci, salmonella and enteroviruses. Sampling frequency may also be reduced by a factor of two if good results are obtained the year before and when no new factor has likely to have lowered the quality of water. Determinands in this category include faecal coliforms and colour. Ammonia and Kjeldahl nitrogen are also included if there is a tendency towards eutrophication. Most of the specified determinands should be sampled 30 cm below the water surface, mineral oils are to be sampled at the surface.
If adopted the proposal for a revised Bathing Water Directive (COM(94)36) will make some important changes in the sampling frequency for some determinands. For example, faecal streptococci, dissolved oxygen and pH will have to be sampled fortnightly and enteroviruses monthly whereas in the present Directive all only have to be monitored when an inspection in the bathing areas shows that the substance may be present or that the quality of the water has deteriorated. In the case of enteroviruses there is a long descriptive qualifier in the proposal: This determinand must be measured once in the fortnight before the start of the bathing season. If during the two preceding bathing seasons the Bathing Water complied with the guideline value for Escherichia coli and the I value for Faecal streptococci, on the basis of Table 3 and 2 respectively, and the Bathing Water does not receive discharges of chemically treated sewage, then the determinand needs only to be measured once more. This measurement should be made in the middle of the bathing season. In the present directive, sampling is required to begin two weeks before the start of the bathing season. In the new proposal, the usual requirement will be for sampling to begin before the start of the bathing season.
The proposed Directive on the Ecological Quality of Surface Waters is again non-specific in defining sample frequency or period. Again this will be determined by the competent national authority and ideally would take account of the variability of the determinand being measured and the desired level of precision for the results. For many countries this potentially will require an increased sampling load in terms of numbers of locations (and also numbers and types of determinands).
It is not apparent from most of the published directives whether there have been any, or if so, what, statistical considerations when defining the required sampling frequencies or numbers. It is not clear, for example, whether the quoted different frequencies for some determinands are related to differences in the variability of those determinands in water.
Sources of error in the overall assessment of a determinand in a water body would include sampling error and analytical error. The former would include how representative the samples taken in time and space are of the underlying true quality/quantity. Clearly, except in a very stable body of water, it is likely that a single sample will give a very unrepresentative estimate of true quality/quantity. It is a common sense principal that the more samples taken the more precise (the smaller the margin of error) will the estimate be of the true underlying quality of a determinand in a water body. As a general rule, to secure double the precision requires four times the number of samples (Ellis 1989). As well as specifying the desired precision, the level of confidence associated with the precision must be stipulated. Again, the higher the desired confidence the greater the required number of samples. Sampling frequency and sample numbers, therefore, have a significant effect on the statistical precision and confidence of the monitoring data produced.
The sampling frequency specified in other international agreements is also very variable within agreements and between agreements (Table E.4). For example, the Helsinki Convention requires a frequency of once per month (preferably in the morning) for microbiological determinands, seasonal for others (alkalinity, temperature, total nitrogen and oxygen), and up to once a year for specified synthetic organic compounds. Twelve samples a year is the most common sampling frequency required for three conventions (Danube (Bucharest Declaration), the Water Economy Agreement between Austria and Hungary 1956, and the Protocol for Technical Co-operation between Greece and Bulgaria 1991). The latter also requires continuous monitoring for some determinands (e.g. dissolved oxygen, temperature and turbidity) and very frequent for others (7 to 10 days e.g. for ammonia and some heavy metals).
6.1.3. Sampling number, type and volume
EU legislation and other international agreements rarely specify requirements for sampling number and volume (Table E.5). Most directives do not specify any requirements or provide recommendations to ensure that sufficient samples representative of the quality of the aquatic environment are taken.
6.2. Analytical methods
Analytical error is another component of the overall error in any estimate of quality or quantity. This Section summarises requirements for pre-treatment and analysis as defined by the methods (or media for microbiological analysis) and the performance (i.e. limits of detection, accuracy and precision). A separate comparison has been made for microbiological determinands (see section 6.2.3).
Although directives sometimes require pre-treatment for water column samples before analysis, methods are generally not specified. The exceptions to this are:
- extraction before analysis for mineral oils, arsenic, cadmium, chromium and mercury is specified in the Bathing Water Directive. Although several other directives also require analysis for these determinands in the water column, methods for pre-treatment are not specified;
- distillation, for nitrogen required in the Surface Waters Directive;
- filtration for salmonella and iron also under the Surface Waters Directive; and,
- organic cooling, an alternative method recommended for pre-treatment prior to analysis for dichloroethane under the Dichloroethane Directive.
These appears to be no uniformity in the requirements for pre-treatment of the water column.
For analysis of sediments and biota, the Daughter Directives to the Dangerous Substances Directive generally require analysis after appropriate preparation but do not define appropriate techniques. The exceptions to this are:
- the Hexachlorocyclohexane Directive recommends the use of extraction techniques; and
- the Mercury from the Chloro-alkali Sector Directive which requires pre-oxidation and successive reduction prior to analysis for mercury.
The Shellfish Waters Directive is also non-specific in its requirements for the pre-treatment of biota, recommending analysis after appropriate preparation. Only for organohalogenated substances does the Directive specifically recommend extraction. The Titanium Dioxide Directive is more specific. It recommends the use of wet/dry mineralisation and purification for ten metals, and extraction for hydrated oxides and oxides of iron.
The analytical requirements made in directives are generally very basic and defined in terms of performance criteria (i.e. limits of detection (LoD), precision and accuracy) and/or by the method. The degree of definition, however, varies greatly from directive to directive (see Tables E.6 and E.7). As for pre-treatment, many directives make very broad requirements to use appropriate methods.
Several directives recommend one or more analytical methods appropriate for selected determinands. The methodologies provided are very brief and since many of the directives were published in the 1970s, the majority of these methods are now out-of-date. However, the methods of analysis recommended are only guidelines and alternative methods can be used providing that they can achieve the same performance criteria. This flexibility is essential given the wide range of methods now available and because simple and cheap methods can often achieve the same performance as more expensive methods. For example, for the analysis of iron as required by the Titanium Dioxide Directive, a simple, cheap but sensitive molecular absorption method can achieve the same analytical performance requirements as an expensive atomic absorption spectrophotometric method. Insisting on using the latter would incur a much greater financial burden without giving appreciative analytical benefit.
The performance criteria are, therefore, the key requirements with regard to analysis. Despite this several directives fail to establish performance requirements for analysis. These include:
- the Bathing Waters Directive;
- the Freshwater Fish Directive;
- the Shellfish Waters Directive; and,
- the Titanium Dioxide Directive;
Directives which establish analytical performance criteria for precision and accuracy usually express requirements as a percentage of the standard. They therefore vary depending on the substance and the media type. For example, in the Surface Water Directive there are different criteria depending on whether the standard is a guideline or a mandatory value. Of more concern is the lack of standardisation in approach to determining these requirements. Thus, for example, within the Surface Water Directive, the required level of precision varies from 5 to 50% of the standard value.
Probably the most significant omission in requirements for analytical technologies is a requirement for analytical quality control (AQC). Increasingly AQC is being recognised as essential for data from monitoring programmes to be reliable and comparable.
There are several national and international texts that provide detailed methods of analysis for numerous determinands, in a variety of matrices (e.g. SCA Blue Book Methods, CEN, ISO, DIN, AFNOR). These, coupled with the propensity of analytical reference materials (e.g. BCR) and check samples (e.g. Aquacheck), that are now available on a pan-European basis, means that laboratories using any relevant analytical method should achieve the target LoDs, accuracy and precision stated in the directives. By participating in inter-laboratory assessments and analytical quality control schemes, any small bias that may occur between different methods (but smaller than the +/- 50% tolerance allowed in the directives for organics, for example) would soon be identified. If, and only if, a significant bias were to be measured, then careful consideration would have to be given to whether the method is suitable in the long term.
It is therefore recommended that directives should refer the reader to currently available texts describing recently developed methods that have been proven to be capable of performing satisfactorily. Consideration should also be given to including a statement that ensures that the laboratory undertaking the analysis uses appropriate internal quality control samples (e.g. as specified in Gardner 1989), participates in an accepted inter-laboratory AQC scheme and possibly has accreditation (e.g. NAMAS, ISO Guide 25, EN45001). In addition, the laboratories concerned with applying the directives should be free to use appropriate methods providing they satisfy performance criteria.
The analytical requirements made in the international agreements are also generally very basic and are not defined in terms of performance criteria (i.e. limits of detection (LoD), precision and accuracy) (see Tables E.8 and E.9). The Baltic Sea monitoring programme generally gives alternative analytical methods for the required determinands but there is apparently no stipulation of performance criteria the methods should achieve, though for some determinands, such as salinity, dissolved oxygen and temperature, details are given for what type of quality control procedures should be adopted. For example, salinity measurements should be calibrated with standard salinity water. By comparison, many monitoring programmes, such as for the North Sea Task Force, do not appear to specify either analytical methods or performance criteria.
6.2.3. Microbiological methods
In total, there are six directives which require microbiological analysis, the Bathing Water Directive and its proposed revision, the Surface Water Directive, the Drinking Water Directive, and, the Shellfish Waters Directive, for which the microbiological requirements have been superseded by the Shellfish Hygiene Directive. A summary of the analytical requirements made under these directives and international agreements is given in Table E.10.
The microbiological and virological determinands in EU directives for water quality all rely upon enumeration of viable (or culturable) units. The culturability is greatly influenced by exposure in the natural environment to stresses, such as starvation, salinity and long wavelength ultraviolet radiation in sunlight or to disinfecting agents such as chlorine, all of which reduce viability and culturability and ultimately bring about cell death. The culturability of bacteriological media is greatly influenced by their composition and by the temperature of incubation. In improving the specificity of culture media for the desired target organisms, specific microbial inhibitors may be incorporated to suppress unwanted species. These are not without effect on the target organisms. Similarly, control of incubation temperature can critically affect the specificity of the medium, and if too high, suppress stressed target organisms.
In the case of enteroviruses, numbers are usually so few that specific concentration and filtration procedures must be used. Inevitably, recovery is less than complete and is dependent upon conditions.
These factors indicate that microbiological methods must be standardised for efficiency of recovery and performance (accuracy, specificity, precision) and, for securing harmonisation of results between Member States, single reference procedures must be agreed for each determinand.
Some work has been undertaken in some Member States to develop the most specific methods given highest recoveries of stressed organisms, such as the guideline documents produced in the UK for bathing waters (National Rivers Authority 1992, and The Microbiology of Water 1994). The current work programmes of CEN/TC 230 (Water Analysis) and ISO/TC 147/SC4 (Water Quality/Microbiological Methods) are proceeding in parallel to produce standard reference procedures for microbiological examination of water, in particular, complete revisions of the current international standards for coliform organisms and Escherichia coli (ISO 9308-1, 2) and faecal streptococci (ISO 7899 - 1,2).
The microbiological determinands specified in international agreements for water quality are also given in Table E.10 together with a summary of the analytical requirements made under these agreements. Apart from the Helsinki and the Elbe Conventions, the analytical methods and performance criteria are in most cases not specified.
6.3. Reporting requirements
Summaries of reporting requirements for each directive in terms of reporting frequency and periods covered, and in terms of the contents of the reports are given in Tables E.11 and E.12 respectively. The reporting format has been standardised for most of the directives covering water in the Reporting Decision. Each Member State is required to periodically complete a questionnaire on aspects such as implementation, compliance details of any designation, emission standards, number of authorisations, quality objects, monitoring stations and action programmes. It does not influence the type of monitoring required under each directive.
Generally the requirement for reporting applies to each directive as a whole and covers the designated authority to whom the data are reported, the frequency of reporting and if there are any compliance rules specified.
The reporting frequency and the reporting period covered varies from directive to directive. In most cases, the information required covers monitoring results but also the number of monitoring stations and methods used. The reporting frequencies are:
- every five years for the Dangerous Substances and Daughter Directives;
- every four years for the Nitrates Directive;
- every three years for the Sampling and Analysis Directive and for the proposed Landfill and Ecological Quality Directives;
- every 12 months for the Surface Water Directive, Exchange of Information Decision and proposed Bathing Water Directive;
- at regular intervals in the present Bathing Water Directive, the Freshwater Fish and Shellfish Directives; and,
- on request (by the Commission) for the Groundwater and Urban Waste Water Treatment Directives.
There is no specification for the Drinking Water or Titanium Dioxide Directives.
The 13 directives (including 7 Dangerous Substances Daughter Directives) which require the reporting of the details and results of monitoring of surface waters are listed in Table E.12.
This report has not assessed how well Member States comply with the reporting requirements and hence how well the Commission can make temporal and spatial comparisons. Inadequate or infrequent reporting is another potential barrier to harmonisation of monitoring data and information across Europe.
Many international agreements also have reporting requirements, these are summarised in Table E.13. Annual reporting is usual.
6.4. Compliance assessment
Another important aspect of directives, particularly when a comparison of quality across Member States is expressed as a comparison of compliance against limits and standard values (e.g. as in the Bathing Waters Directive), is how the compliance requirements in the directives are expressed, calculated and interpreted. Differences in interpretation of these requirements is another significant barrier to harmonisation of monitoring and implementation of directives across Europe.
A compliance monitoring scheme can be defined as:
Any procedure for making a pass/fail decision about the quality of some defined population (e.g. effluent; river; potable water) on the evidence of the analysis of a number of randomly selected samples from that population (Ellis 1989).
Rules for establishing compliance would contain three key items:
- the number of samples;
- the statistic by which quality is characterised (e.g. in the case of Bathing Waters the number of samples exceeding the faecal coliform standard of 2000 per 100 ml); and,
- the pass/fail criterion to be applied to the statistic (e.g. 0 or 1 high samples = pass 2 or more = fail).
These three items constitute the practical instructions that enable the scheme to be applied and a wide variety of compliance monitoring schemes can similarly be defined by these three basic components.
Any rule to test compliance - however non statistical in origin - can be characterised by two key quantities: a true underlying level of quality so good that the rule will almost always be passed, and a level so bad that the rule will always almost fail. The greater the number of samples the closer these two numbers will be and so the greater the discrimination provided by the scheme. The statistical capabilities of any percentile type scheme can be determined non parametrically by look up tables or charts. Whichever compliance assessment rule is adopted, attention needs to be given to frequencies of sampling. The required frequency should be derived with reference to the quantified risks that some waters will be misclassified (against compliance criteria). Furthermore for fairness of comparison the frequency of sampling should be uniform throughout Europe.
There are specific compliance assessment rules in four directives (the Bathing Water, Shellfish, Freshwater Fish and Surface Water Directives) and more general rules for the Dangerous Substances and its Daughter Directives. There are no compliance criteria stipulated in the Nitrates, Groundwater, Drinking Water (initial examination), and proposed Ecological Quality and Landfill Directives. However, for the latter, observations must be evaluated by means of control charts with established control rules and levels for each downgradient well. Compliance is generally required to be demonstrated in a percentage of the number of samples taken (95%, 90%, 80% and 75%). There are also examples of 100% sample compliance being required e.g. in the Shellfish Waters Directive.
The different percentages of sample compliance is generally related to the determinand being measured. For example, in the Shellfish Waters Directive three compliance rules are given for the three broad categories of determinands:
- 100% of samples for organohalogenated substances and metals (based on a minimum half yearly sampling frequency);
- 95% of samples for salinity and dissolved oxygen (based on a minimum monthly sampling frequency); and,
- 75% of samples for other determinands (based on a minimum quarterly sampling frequency).
Similarly it is also indicated in the Freshwater Fish Directive that 95% of samples for pH, BOD, ammonia, (total and non-ionised), nitrite, residual chlorine, total zinc and dissolved copper should comply with requirements. When the sampling frequency is lower than one sample per month all samples have to comply. For temperature and dissolved oxygen the percentages listed in Annex I of the directive should comply and average concentrations are set for suspended solids.
The requirements made in the Freshwater Fish Directive also demonstrate the use of secondary rules for when samples do not comply. This is also demonstrated in the Surface Water Directive where water is assumed to conform to the relevant determinands if it complies with the parametric values for the water quality in question, in the case of:
- 95% of the samples for determinands conforming to those specified in the I (mandatory) columns in Annex II;
- 90% of the samples in all other cases; and,
- if in the case of the 5 or 10% of the samples which do not comply:
- the water does not deviate from the parametric values in question by more than 50% expect for temperature, pH, dissolved oxygen and microbiological determinands;
- there can be no resultant danger to public health; and,
- consecutive water samples taken at statistically suitable intervals do not deviate from the parametric values.
Similar secondary rules on how to treat non-complying samples are also used in the present Bathing Water Directive which stipulates that 95% of samples must comply for mandatory determinands, 90% of samples for the remaining determinands, except for total coliforms and faecal coliforms where the percentage may be 80% (for the guideline value). Compliance will still be achieved if in the case of the 5%, 10% and 20% of the samples which do not comply; the water does not deviate from the parametric values in question by more than 50%; except for microbiological determinands, pH and dissolved oxygen; and consecutive water samples taken at statistically suitable intervals do not deviate from the relevant parametric value.
However the Bathing Water Directive only gives a brief explanation of the required method of compliance assessment. Even though it is relatively clear that compliance was to be judged by the percentage of samples that do not exceed the threshold values specified in the Annex, it is less clear how this rule should work if this specified percentage (e.g. 95%) did not correspond to a whole number of samples (Lacey et al 1995). It is also not explicit whether compliance was to be judged for each determinand in turn (determinand compliance) or across all determinands on the same sample (sample compliance).
The proposed Directive on Bathing Waters is clearer. First it is clear that determinand compliance is the meaning intended. Second the numbers of permitted exceedences is set out in look up tables of the Annex and completely specify the method of compliance assessment. In the look up tables the number of samples which need not conform with the guideline and mandatory values are given. The test statistic is the number of exceedences of sample values above the specified percentile value, and that number is referred to the look-up table to determine whether the frequency of exceedences is acceptable. The Directive, therefore, offers some clarification of the method of compliance assessment and also revises the quality requirements in particular by adding an I value for faecal streptococci. Another important feature is the footnote (1) to Table 1 in Annex 1 which allows re-sampling if an abnormal value for faecal streptococci is recorded. No definition of abnormal is given but since the footnote allows for the abnormal value to be replaced by the re-test value, this could have a very important effect on the operation of the compliance rule.
In the case where the compliance statistic is expressed as a maximum there are four principal statistical deficiencies: no unbiased method exists for estimating the population maximum (the sample maximum will always be an under-estimate of the true maximum - except where analytical error is appreciable in relation to the size of natural variability); the benefit-of-doubt approach to assessing compliance cannot be used (people will strive to take the very minimum number of samples so as to lessen their risk of an unlucky failure); the fail-safe approach cannot be used and so a face-value approach is in fact the only possible stance when monitoring against an absolute standard; and analytical error causes a special complication making in marginal cases making an important difference (Ellis 1989).
The Dangerous Substances Daughter Directives for Mercury, Cadmium, Hexachlorocyclohexane, and Carbon Tetrachloride indicate that all determinand concentrations relate to the arithmetic mean of the results obtained over 1 year. This requirement is not explicit in the other daughter directives. There are, however, no specific compliance rules for the determinands but competent national authorities must check that there is no deterioration in the environment concerned.
No information is available on how compliance is addressed in other international agreements.
For references, please go to www.eea.europa.eu/soer or scan the QR code.
This briefing is part of the EEA's report The European Environment - State and Outlook 2015. The EEA is an official agency of the EU, tasked with providing information on Europe's environment.
PDF generated on 28 Oct 2016, 10:54 PM