Do something for our planet, print this page only if needed. Even a small action can make an enormous difference when millions of people do it!
This chapter summarises methods for obtaining a preliminary overview of the air quality situation in a zone by measurements, in the case that no previous assessment is available. These measurements are not intended to control compliance of limit values under Article 6 of the FWD, but rather are screening techniques meant to determine what compliance measurements and other assessments are needed to comply to EC legislation. The results of preliminary measurements can be complemented with an assessment of emission sources and modelling to obtain a full picture of the air quality in a zone, as measurements are inherently limited in their representativeness in space and time.
A measuring strategy depends on the objectives of the monitoring, and the pollutants to be assessed. For the relevant air quality parameters (concentration of pollutants and associated averaging time), we need to specify where, how, and how often measurements should be taken. The measuring effort will be dependent on:
It is possible to derive, in quantitative terms, a measuring strategy from this information. However, this is not always practicable, as the required accuracy is often not specified, and the variation of the pollutant in space and time may not be sufficiently well known. In this guidance report, the following approach is proposed:
| 1. | Estimate the levels and spatial variation of the quantity from existing measurements in the zone, or in similar situations elsewhere, or from emission inventories and model calculations. |
| Where measurements are being used as the basis of the assessment, | |
| 2. | Design a measuring strategy for assessing the spatial distribution of pollutants, the pollutant levels in areas with highest concentrations and in background locations. |
| 3. | Carry out measurements with appropriate methods, with a frequency and over a period of time as specified below. |
| 4. | Calculate from the measurement results estimates of the relevant air quality statistics. |
| 5. | Estimate the uncertainty of the measured estimates, taking into account the accuracy of the individual measurements, station representativeness, and variability of the pollutant in space and time. |
| 6. | Where available compare model estimates and measured estimates, and evaluate discrepancies. |
Steps 2 and 3 are covered in paragraph 3.4, 3.5 and 3.6, steps 4 and 5 in paragraph 3.7. Information on emissions and models is provided in chapters 4 and 5.
A measuring strategy for preliminary assessment will concentrate on the question in which areas pollution levels exceed limit values, or the associated assessment threshold levels, as set by the Directive. For those areas, more intensive assessment as described in article 6 of the FWD will be required. The preliminary assessment should not only identify the location of maximum concentrations, but also determine the extent and the limits of the area of exceedance.
Currently used air quality measurement techniques can be sub-divided in three main categories:
These are the simplest and cheapest measurement methods, usually based on a sampling procedure followed by chemical analysis (or gravimetric determination for suspended particulate matter). According to the implemented sampling procedure, different manual methods can be recognised: the use of bubblers for gaseous pollutants, the diffusive sampling method for gaseous pollutants, and the collection on filters for suspended particulate matter (Black Smoke and PM10) and heavy metals measurements.
These currently constitute the most widespread monitoring technique in the air quality monitoring networks. The analysis of the pollutants is based on physical principles and processed electronically. Automated analysers allow for the continuous, automated, on-line and time-resolved measurement of air pollutants. The major drawback of this technique resides with the high costs for purchase and maintenance of the analysers, often resulting, as a consequence, in low network density and low spatial resolution of the measurements
Mobile laboratories equipped with automated analysers constitute a useful application of this technique as a tool for measurement campaigns at locations of interest.
such as the Differential Optical Absorption Spectrometry (DOAS), allow for the simultaneous monitoring of various gaseous pollutants integrated over a distance of several hundreds of meters. As for the other automated methods, the long-path optical methods allow for the continuous, automated, on-line and time-resolved measurement of air pollutants.
Among these measurement techniques, screening techniques based on the use of a mobile laboratory and the diffusive sampling technique or other manual methods are of particular interest, because of their relatively low cost and their simple and fast operation, in comparison with fixed monitoring stations. Three different approaches are proposed in this chapter:
or a combination of one or more of the proposed approaches.
The low cost and easy operation of the diffusive sampling technique make it an ideal tool for large scale air pollution surveys with a high spatial resolution (De Saeger et al.,1991,1995). A diffusive sampler is a device capable of taking gas samples from the atmosphere at a rate controlled by molecular diffusion, and which does not require the active movement of air through the sampler. The diffusive sampler consists of a tube, one end containing a sorbent which fixes the pollutant. The pollutant is sampled onto the sorbent at a rate controlled by the molecular diffusion of the pollutant gas in the air, without requiring any pump or electrical power. After exposure of the samplers over periods varying from a few days to a few weeks, the tubes are closed and returned to the laboratory for analysis. According to the type of device and the measured pollutant, analysis can be performed using different techniques, such as colorimetry, ion chromatography and others. Maps of the pollutant concentrations over the area can be obtained by interpolation of the diffusive sampler measurements (see Fig. 1: Example of Madrid - Spatial distribution of NO2 levels determined by diffusive sampling) Diffusive samplers are today available for a large number of gaseous pollutants, such as SO2, NO2, O3, CO, Benzene.
Fig.1 Spatial distribution of NO2 levels determined by diffusive sampling
(example of Madrid)
The technique is particularly suited to determine the pollutant distribution over a large area, and to assess integrated concentration levels over longer periods of time (long-term limit values). Short-term limit values can be derived from statistical data, by comparison with extended and time resolved measurement series from similar measurement locations. The proposed methodology can be used to determine areas of maximum concentration and combined with the use of a mobile laboratory as described in 3.5. In addition it may support the optimisation of monitoring networks and assessments supporting generalisation.
When applying this methodology in the case of a preliminary assessment, the following steps are proposed:
It should be noted that diffusive samplers are very cost effective, but that their implementation on a large scale may be labour intensive and hence costly. They can however also be a useful tool when used less intensively in conjunction with other assessment methods.
Other manual measurement techniques, such as bubblers (total acidity, Thorin and TCM method for SO2, Saltzmann method for NO2), can be used as an alternative to the diffusive sampling technique, in particular when the number of samplers to be implemented is low. The methodology proposed for the diffusive sampling technique applies in that case also for bubblers.
3.4.2 Data quality requirements
When performing diffusive sampling campaigns, the following data quality requirements are proposed. These data quality objectives are only indicative, and may be strengthened where possible.
In the case of those agglomerations for which an intensive measurement campaign is undertaken, it is proposed to install a number of samplers equal to 15 times the initial number of measurement stations required for mandatory measurements (Ni). This would result in a number of 30 samplers for agglomerations with a population of 250.000 inhabitants, of 60 samplers for agglomerations with a population number of 1.000.000 inhabitants and of 150 samplers for an agglomeration of 6.000.000 inhabitants (see Daughter Directive proposal for SO2, PM10, NO2 and Pb). The sampler density may vary in function of the emission sources configuration, and it is good practice to increase the sampler density in city centers with respect to the outskirts (see Fig. 2: Example of Athens - Construction of the measurement grid). Additional samplers would be installed at a representative sample of hot spots, such as along busy roads and crossings, as well as in the surroundings of industrial pollution sources, in particular if they are likely to affect local pollution levels. A limited number of samplers should be installed at the periphery of the area under investigation, in order to assess the impact of adjacent areas.
Fig.2 Grid construction for diffusive sampling measurements in an urban area (example of Athens)
In other cases (industrial zones, rural background), the number of stations should be sufficient to determine the extent of pollution and exposure.
Diffusive samplers for ambient air measurements have been developed for various pollutants. The Palmes diffusion tube for the measurement of SO2 and NO2 is certainly the best known, but several other types are widely used today (Palmes, 1973; 1976). Figure 3 gives an example of currently available diffusive samplers. Other kinds of samplers are today available, covering almost all the gaseous pollutants foreseen by the Framework Directive (Brown,1993). A guide for the selection and the application of the diffusive sampling technique is currently being prepared by CEN - Technical Committee 264 - Working Group 11 (CEN,1996).
Fig.3 Example of currently available diffusive samplers
The following table gives a review of the existing diffusive samplers and of their analytical principles:
| Pollutant | Analytical principle |
| SO2 | Chemical absorption + colorimetry / ion chromatography |
| NO2 | idem |
| NOX | idem (same as for NO2 + oxidation layer) |
| O3 | Chemical absorption + colorimetry |
| CO | Chemical absorption + colorimetry |
| BENZENE | Chromatographic adsorbent + GC analysis |
It should be noted that the principle of molecular diffusion does not adapt to particulate matter, and that the diffuse sampling technique is therefore not applicable for PM10 or heavy metals (Plants have been used as passive samplers for some of these substances, but this is a surrogate for deposition rather than for concentration measurements).
In urban areas, the spatial variation for primary pollutants such as NO, CO, Pb, PAH's and benzene is mainly determined by their emissions from automotive traffic. As a result of this, one single pollutant representative of the emissions from automotive traffic may be used as indicator for the other pollutants, when determining areas of maximum concentrations. This "indicator approach" is however valid only if large industrial sources with low level emission heights are not present in the area. Particularly for Pb, PAH or benzene, this cannot be taken for granted. Nor is this approach acceptable for secondary pollutants such as NO2.
The methodology allows evaluation of the maximum concentration levels in a zone over a period that is representative of the reference period(s) of the limit value. It is used as a preliminary assessment method in order to verify whether a zone is in exceedance or near-exceedance of the limit values, and determine the ongoing assessment regime that will be required under Article 6 of the Framework Directive.
Mobile laboratories or transportable measurement stations used for stationary measurements at fixed sites, usually combine the advantages of automated measurement methods (continuous, time-resolved measurements) with mobility or flexibility. For pollutants for which automated measurement methods are not available, mobile laboratories may also be equipped to perform non-automated measurements (PM10, heavy metals, PAH's). The duration, the periods and the frequency of the campaigns or measuring periods will have to be established so as to be representative of the reference period of the limit value (1 hour, 24 hours, 1 year).
The location of maximum concentration levels in a zone will be chosen taking into account the source distribution, local meteorological conditions and orography. The types of sources present in an area are very important when choosing a measuring site. Impact from elevated point sources is often difficult to measure at one point at ground level because both wind direction and wind speed, and their variation with height is important for the location of the maximum ground level impact. For monitoring the pollution from roads, the impact will decrease with the distance from the road, and the level of pollution will on average be proportional to the volume of traffic. Time-series of hourly concentrations should reflect the pattern of traffic intensity. The highest concentrations for 24-hour periods should be expected to be located in areas where the road runs parallel to the most frequent wind-directions, or where the curvature of the road allows impact from several wind-directions. For monitoring pollution mainly from area-sources the location should be chosen close to the centre of the area, and avoid direct impact from "super local" sources in the vicinity (example: small incinerators or petrol stations). In complex situations resulting in a high variability of the pollutant distribution (sources of different origins, complex terrain and meteorology), it is advisable to perform the measurements in different representative locations.
When applying this methodology the following steps are proposed:
When performing the measurements with a mobile laboratory, the following data quality requirements are proposed. These data quality objectives are only indicative, and may be strengthened where possible.
Better estimates of minimum time coverage can be calculated from stochastic sampling from an existing series of monitoring data in a similar situation. This exercise has been carried out in the EC Working Groups on particulates and on benzene (see position papers, to be published)
The mobile laboratory would be equipped with one analyser for each of the pollutants under consideration. The selected measurement method shall comply with the reference method of the respective Directives as well as with the quality objectives. The following table gives an example of possible measurement methods to be used for the assessment.
| Pollutant | Measurement method |
| Sulphur dioxide | UV fluorescence |
| Nitrogen dioxide | Chemiluminescence |
| PM10 | Sampling on filter + gravimetry (manual method), beta attenuation, oscillating micro-balance |
| Lead | Sampling on filter + atomic absorption spectrometry, ICP, X-ray fluorescence (manual method) |
A mobile laboratory can easily combine measurements of various pollutants, and may constitute a unique screening tool for pollutants for which cost effective measuring techniques are not available (PM10, heavy metals).
Further to the assessment of pollution levels in areas of maximum concentrations, a mobile laboratory can also be used to assess the pollutant spatial distribution over a larger area. Grid monitoring is performed by dividing the particular area of interest into a grid of squares, and by measuring the pollution levels in each grid cell. The measurements are made during short periods of time at each intersection of the grid lines, and repeated over the course of a year. The dates and hours for the measurements are chosen randomly but in such a way that they are evenly distributed over the months, the days of the weeks and the hours of the day. The measuring schedule is laid out so that no neighbouring intersections are measured on the same day. The single values measured at the four corners of each grid are used to calculate the mean concentration value for each grid cell and the pollutant concentrations isopleths over the area. Percentile values can be estimated statistically from the accumulated frequency distribution.
Note that the method is not suitable to characterise air quality hot spots, for which additional sampling should be carried out.
A typical example of this methodology is illustrated by a study made in Karlsruhe for NO2, see fig. 4: Example of Karlsruhe - Grid monitoring of NO2 levels (UMEG,1996). In this particular case, the area was divided into a grid with a density of 1 x 1 km, and half-hourly measurements were repeated 26 times over the year, resulting in 104 half-hourly measurements for each grid square. The measurements were performed between 6 a.m. and 10 p.m., resulting in an overestimation of the concentration levels, but measurements during the night may be considered when a higher accuracy is necessary.
Fig.4 Use of a mobile laboratory for grid monitoring (example of Karlsruhe)
The proposed methodology is of particular interest in cases where a limited number of measurements are required (small agglomerations), or when other kinds of screening techniques are not available (SPM).
When applying this methodology in the case of a preliminary assessment, the following steps are proposed:
When performing the measurements, the following data quality requirements are proposed. These data quality objectives are only indicative, and may be strengthened where possible:
See 3.5.3
When the measurements have been carried out, the relevant statistical quantities (annual average, percentile values) for which limit values are defined, are to be estimated. This is particularly relevant for preliminary assessments. Particularly for higher percentiles, and for short measuring periods, this introduces major uncertainties. To our knowledge, no generally accepted methodology is available; we introduce here some important aspects only.
In principle, the problem may be approached by assuming a certain frequency distribution of the concentration, and determining the basic parameters for this distribution. An obvious choice is the log-normal distribution, which allows for estimation of the median and logarithmic standard deviation, from which all percentiles can be calculated. Clearly estimates of higher percentiles become progressively more uncertain in this procedure.
However when estimating these concentration statistics from the measurements, the following aspects should be considered:
The variability of the meteorological parameters (wind, temperature, atmospheric stability, precipitation, ...) constitute an important factor affecting the frequency distribution of the measured concentrations. When calculating the concentration statistics from the measurements, the meteorological conditions during the measurement period must be checked to be representative of the normal meteorological conditions, and, if not, the measurements should be corrected accordingly. The measurements are to be classified according to meteorological classes, and their weight/frequency adjusted in accordance with the frequency of occurrence of meteorological classes over a long (10-30 years) period. This method is recommended if the measurement period is at least one year.
An alternative possibility is:
From literature, (semi)-empirical relations may be found between one hour average concentrations and their longer-term averages (day, month, year), and relations between those averages and percentiles. However, temporal variations of concentrations depend a.o. on the source environment, and, consequently, these relations can be quite different for situations with pollutant emissions from different source categories (point, line, area). For continuous elevated point-sources, the impact at one point at ground level will be more dependent upon variability in dispersion conditions than for line or area sources. The peak-to-mean ratio for a point source is generally much higher than the ratios for line and area sources. A location where the pollution level is dominated by contributions from area-sources reflects the variation in area-source strength with time more than the variations in dispersion conditions.
Estimating concentration statistics from a limited number of measurements can therefore be a difficult task. The proposed methodology is best indicated in the case of area- and line-sources, i.e. in urban situations where the air pollution is diffuse and mainly dominated by automotive and domestic heating emissions. In the case of point sources, it is generally not possible to assess the relevant quantities from supplementary measurements. For assessment in these cases, the combination of emission data and model calculations should be preferred.
The impact of measurement uncertainty on concentration statistics is discussed by Van der Wiel et al. (1988) The variations related to measuring imprecision introduce upward shifts in the percentiles, which become increasingly important for higher percentiles. The reliability of this procedure is dependent on the size of the data base; also, in some cases with rarely occurring high concentrations, as for example around point sources, the correction may result in even larger errors.
The uncertainty of measurements in field conditions depends on:
Standard testing procedures have been established to determine the uncertainty of the measurements by a given instrument in laboratory and field conditions, in particular for type approval of the instrumentation. It is however far more difficult to estimate the measurement uncertainty in routine field conditions. This estimation can only be performed experimentally by submitting the measurement system to quality control in field conditions over a longer period of time (typically 3 months).
Concerning the spatial representativeness of the measurement location, a wrong location will undoubtedly lead to an underestimation of the true concentration, as the measurement strategy is aimed at the measurement of the highest concentrations. In practice however, it will be impossible to estimate the extent of the associated uncertainty. It is therefore proposed not to consider this element in the uncertainty estimation, but to require a detailed report on the siting criteria.
On the basis of experience obtained from the QA/QC programmes organised by the Joint Research Centre for SO2 and NO2 measurements, an estimation of the uncertainty for measurements made by a mobile laboratory and by the diffusive sampling technique is given in the following table (De Saeger, 1996). In both cases, the assumption was made that the laboratory was implementing a recognised quality system (ISO 9000, EN 45001) and using up to date instrumentation.
Estimated uncertainty SO2 and NO2 measurements (95% confidence interval)
Source of errors |
Mobile laboratory (hourly mean value) |
Diffusive samplers (2-week average) |
| Sampling Sampling efficiency Losses in the sampling line |
-5% |
±15% |
| Calibration chain Primary calibration Transfer standard Calibration check in station |
±10% | ±5% |
| Instrument response Precision, stability, linearity Selectivity Environmental variables (T, P, RH %) Maintenance |
±10 % | ±25 % |
| Propagated uncertainty | ±15% | ±30% |
The regular organisation of quality controls during the measurement period constitute therefore the only way to contain the uncertainty of the measurements within the established limits and to document the quality of the determinations.
For references, please go to https://www.eea.europa.eu/publications/TEC11a/page005.html or scan the QR code.
PDF generated on 19 Apr 2024, 02:24 AM
Engineered by: EEA Web Team
Software updated on 26 September 2023 08:13 from version 23.8.18
Software version: EEA Plone KGS 23.9.14
Document Actions
Share with others