INTER-NOISE 2016
END noise mapping for a sufficiently accurate people exposure
estimation in epidemiological studies
Gaetano LICITRA1; Diego PALAZZUOLI2; Elena ASCARI3
1
ARPAT, Coordinator of Area Vasta Costa, Italy
2
ARPAT, Direzione Tecnica SIRA, Italy
3
IDASC-CNR, Italy
ABSTRACT
The information required to perform an accurate exposure estimation are several, such as: geographical and
topographical indications, population distribution, source properties. Years of research have improved and
validated different numerical methods, which provide accurate results in controlled conditions, i.e. when all
of the above-described information are available.
Since the computational speed is increasing by means of parallel systems, it will be possible to develop new
more accurate methods. They will be theoretically able to manage more inputs, obtaining more thorough
analysis.
Actually, it is not unlikely that some of the data required are missing or their accuracy is not suited. This leads
to approximations, increasing accordingly the level of uncertainties in the exposure evaluation.
Epidemiologists are accustomed to adapt their studies to the amount of available data, nonetheless their
results depend onto which extent the exposure evaluation is accurate.
In the paper limits and advantages of noise mapping due to END will be discussed in the perspective of the
epidemiological studies and their needs and compared also to other approaches used in specific
epidemiological studies. It will be also presented the new annex II of END, the work developed in
CNOSSOS-EU, as well as some ideas for possible future improvements.
Keywords: Noise exposure, Indicators, Epidemiological studies
Number(s): 68.3, 69.3, 89.
I-INCE
Classification
of
Subjects
1. INTRODUCTION
With the approval of the European Directive 2002/49/EC on Environmental Noise (END) Member
States (MS) have developed new policies for its implementation. So in Europe, different levels of
experience, approaches and funds on environmental policies carried to different results in terms of
number of maps elaborated and calculation methods used (not all Countries reported to the EU
commission data required after the first and second round of mapping – 2007,2012). Beside MS efforts,
exposure results provided are not comparable, essentially due to different calculation methods used
(national and interim methods both available) and to interpretation of END, which leaves some
definitions unclear to be put into practice. Efforts have been done by the European Commission to
guide them in providing maps: practical guide on modelling choices and relative accuracy has been
provided (The Good Practice Guide, GPG (1)). However, it must be considered that comparable data
could be obtained only when a good level of expertise is achieved by all technicians involved in
mapping projects. In fact, a full understanding of mapping parameters would help the estimation of
map accuracy and improve comparability which is in any case limited by the use of different methods.
Thus, a project to develop a common calculation method for all EU Countries has been developed in
the years 2008-2011: CNOSSOS-EU is a project to define common methods based upon previous
studies HARMONOISE and IMAGINE such that each MS could fill his own source database (car rying
1
2
3
g.licitra@arpat.toscana.it
d.palazzuoli@arpat.toscana.it
elena.ascari@gmail.com
5687
INTER-NOISE 2016
out specific measurements) and then a common propagation method is applied so that results would be
comparable all over Europe. Only with comparable levels and exposure results, policies to improve
noise quality could be developed at European level. For this reason, recent revision of END revision of
Annex II includes not only the new method CNOSSOS-EU, but also a specific method for distributing
population along building façades and evaluating their exposure according a common method as
highlighted in the paper. Some examples of different distribution methods will be provided in this
paper to stress the relevance of this issue not only for noise mapping and action planning but also for
annoyance estimate and so for health risk assessment. Indicators able to rate risk assessment and to
drive prioritized actions will be presented. Indicators will take into account the aim of estimating the
Environmental Burden of Disease (EBD) and so the health outcomes of harmful effects of noise
exposure. Different approaches will be suggested according the study extent. In fact, small and large
scale studies may follow a different approach maximising accuracy of estimation.
2. END REVISION: THE NEW ANNEX II
END has been in place for more than 10 years. Recently, the Commission identified it as one of the
regulations "to be evaluated with a focus on regulatory fitness" in the context of the Regulatory Fitness
and Performance initiative (REFIT) and the Better Regulation programme of the European
Commission. This evaluation covered all current provisions of the Directive, and addressed questions
relating to effectiveness, efficiency, coherence, relevance and EU added value.
The Commission has collected in March 2016 the views of citizens and associations of citizens, as
well as all other interested stakeholders, through an online public consultation. The consultation
questionnaire addresses specifically the relevance, effectiveness, efficiency and EU added value of the
Environmental Noise Directive.
Moreover, at the time of the adoption of the END, the legislators included in the Directive an
obligation for the European Commission to adapt its Annexes I, II and III to technical and scientific
progress, notably to establish common noise assessment methods (Annex II) and methods for
assessing harmful effects of noise by means of dose-effect relations (Annex III).
The work on the former has recently been finalised, while the work on the latter is still ongoing and
should be finalized in this year.
In the following paragraphs a brief explanation of the revised annex II is provided.
2.1 The new common assessment method: CNOSSOS-EU
In the period since the adoption of the END, the different approaches to noise mapping have been
one of the key implementation challenges recognised. In fact , besides the Commission provided
guidance as the GPG, the lack of comparable and common assessment methods has caused significant
inconsistencies in exposure estimates, between different countries, within a single country and across
the two main reporting rounds (2007-2012).
Several studies demonstrated the need of a common method to obtain comparable maps ( 2, 3, 4).
Different exposure distributions have been found to be reported according calculation method used,
pointing out national clusters.
A major step forward in this respect came about with the development of common noise assessment
methods in Europe, the methodological framework under the name of CNOSSOS -EU. The
CNOSSOS-EU has recently become part of the EU legislative framework in the form of the revised
Annex II of the END.
The new methodologies will ensure that noise in each Member State is assessed in a harmonised
way, thus providing a consistent and reliable picture of the acoustic situation in the EU. The use of
CNOSSOS-EU will be mandatory for all Member States after 31 December 2018, however Member
States will have the possibility to transpose and start using the revised Annex II even before this date.
The methods is based upon the “fit for purpose” concept, i.e. a different level of detail can be
achieved according to the aim (large scale mapping or small scale action evaluation), using more or
less data input (optional requirements). The innovation is to have a very detailed methodology,
customizable for each MS, which can be used in a simplified version for strategic maps and in the
detailed one for action planning. The CNOSSOS methodology will remain freeware, but an
implementation in commercial software is required to make users able to map large areas with
instruments already known. However, nowadays only a point to point calculation system is available
and the implementation in commercial software is still to be published due to the difficulties in
implementing the standard and the national adaptation coefficients in software. In fact, the standard
5688
INTER-NOISE 2016
common methodology, as it is described in annex II, do not specify some issues leaving them free for
software implementation. Therefore, some Countries like Austria are already working on databases to
test and adapt the new methodology. The Commission has already provided tables to convert existing
standard methods (as NMPB, Nord2000, CRTN, RLS90, SRMII).
Moreover, this new methodology is valid in a wider frequency range than actual interim methods. It
is based on 1/3 octave-bands from 25 Hz to 10 kHz. This aims a better evaluation of annoyance effects
which are mostly influenced by noise at low frequency bands. A methodology considering the wide
spectrum allows a correct evaluation of these effects and also a correct evaluation of possible solutions.
So common method according CNOSSOS project will allow policy makers to base their decisions on
more reliable data, based upon local databases, that are not only an overview of the noise pollution of
the city but a detailed analysis of the noise environment.
2.2 The method for assigning levels to population
In the same revision of Annex II, also the method of assigning noise values to population has been
changed. In fact, the END first version assessed the exposure of population assigning the maximum
level of a building (or dwelling depending on information available) to all the inhabitants living there.
In the actual revision, the VBEB method (German national standard, 5) has been adopted since it
allows a more reliable noise exposure estimation.
According to the new method, the inhabitants in a building are equally distributed over all the
receivers ring located at the façades according to the rules established in the German national method
and the incident sound level is calculated for each receiver on the ring; the number of receivers is
automatically set by VBEB rules, i.e. locate at least one receiver every 5 m on the same façade and
none on façades shorter than 2.5 m.
Moving to the new assignment method, the higher exposure reported by all the European cities
compared to German cities will disappear in the next round implementation and comparable results
will be published since the noise levels assigning method is a major factor in influencing reported
exposure.
So, even if it is not possible to foresee the distributions of next rounds it is clear that an
improvement of exposure will be reported but it will not be due to mitigations but to the change of
calculation methods. Not only there will be changes in the reported exposure but also in the health
effects estimations and so the risk assessment. Next paragraph will clarify these facets.
3. NEEDS ANALYSIS OF EPIDEMIOLOGICAL STUDIES
The process of risk assessment of environmental noise requires knowing:
 the nature of the health effects of noise;
 the levels of exposure at which health effects begin to occur and how the extent of the effect
changes with increasing noise levels;
 the number of people exposed to these hazardous levels of noise.
Quantitative risk assessments based on Environmental Burden of Disease (EBD) methodology
have been developed and used by WHO (6) to help the Member States quantify several
environment-related health problems. The EBD is usually expressed as the number of deaths and the
metric disability-adjusted life year (DALY), which combines the concepts of potential years of life lost
due to premature death and equivalent years of “healthy” life lost by virtue of being in a state of poor
health or disability.
So people exposure estimation is the first step of risk assessment and it is itself a complex issue that
needs several steps. Each step include uncertainties and different possible met hodological choices.
Assessment of exposure to noise requires considering many factors, including:
 choice of noise indicator;
 measured exposure or calculated/predicted exposure;
 population distribution/levels assignment;
 time-activity patterns of the exposed population;
 combined exposures to multiple sources of noise.
In the following details will be provided regarding measurement/calculation methods and
population distribution and levels assignment.
Then next paragraph will be focused on indicators and multiple sources effect.
5689
INTER-NOISE 2016
3.1 Noise levels accurate estimation
The exposure of the population to the noise source can be obtained by measurement or by using
models that calculate noise exposure based on information about the source and on information about
sound propagation conditions from source to receiver. Such calculation models can also be used to
predict levels of noise exposure for some time in the future based on estimated changes in noise
sources. Best-practice methods should be adopted for measurement or for calculation in the
assessment of exposure, with a full understanding of the assumptions, limitations and potential errors
associated with any approach to measurement or estimation. This objective will be fulfilled with the
adoption of the common assessment method, in fact noise exposure mapping is a commonly adopted
step in the process of estimating the noise exposure of a population. However, noise maps realized till
now applied a lot of different methods leading to results that differs a lot wit hin each other. The
influence of noise mapping methods in exposure results have been largely demonstrated in literature
(2,4). Accuracy of existing methods has been evaluated in the GPG but a new version is expected with
the new CNOSSOSS method.
Moreover, there are evidences that a different exposure estimation leads also to different
dose-effects relationships according implemented mapping methods. Lercher in (2) demonstrated that
highly annoyed percentages (%HA) varies a lot and that the difference between methods is not
constant varying the context (urban / sub-urban / rural), the type of source (motorway, main road,
railway) or even the noise indicator (L den , L night ). In particular, the comparison between standard
implementation of ISO9613 and NMPB-96 interim method was performed, coupling their results with
annoyance surveys. The derived dose-effects relationships were compared to the standard ones
demonstrating how they vary according the noise predicted level of exposure. In Figure 1 difference
was reported for L night indicator and motorway source.
Figure 1 – Dose-effect relationship: highly annoyed by motorway exposure during night by two modeling
strategies and compared with the Standard annoyance relationship (from (2))
Another approach in estimating noise levels is frequently used in epidemiological studies: the Land
Use Regression (LUR) method. The LUR is mostly used to derive air pollution exposure data since air
pollutants propagation is simpler than noise one. However, when a large territory exposure estimation
is needed, a first fast raw approximation can be obtained with LUR. In fact, in (7) it has been used to
estimate exposure to road traffic noise in Montreal and to calculate %HA and %A. Input data of LUR
models are measurements and few propagation rules. Of course, LUR are more accurate next to the
sources and generally are not suited to assess small scale variations. Therefore, LUR models could be
useful to estimate noise exposure at large/medium scale for a first attempt to demonstrate health effects
without using noise mapping methods. In fact the greater limitation of this method is the cost of
measurements needed to elaborate LUR. In the study (7), LAeq24h levels of Montreal were computed
basing the LUR model on 2-min noise measurements from 204 sites.
Finally, the most accurate method to estimate noise exposure is obviously measurements. This is
feasible only for very small scale studies in which accurate exposure levels influence the outcomes. In
5690
INTER-NOISE 2016
fact, taking measurements of every inhabitants exposed to noise is costly and you need to get in touch
with people willing to participate to the epidemiological study. In this sense, this approach is suitable
when surveys are coupled to noise estimations on a relative small population sample.
It must be considered that measurements and mapping methods accuracy can be estimated
respectively using GUM (8) and GPG (1) so that final results of epidemiological study can be
evaluated in terms of uncertainties. On the other hand, LUR methods for noise estimation have not
been much used till now, so no accuracy guide is available, nor a standard approach.
3.2 Noise levels accurate assignment
Many methods to estimate the noise exposure of population have been proposed in the li terature.
Some of them are also standardized at national level and implemented in almost all noise prediction
software for their automatic calculation.
The methods can be grouped into two categories, that is:
• those where the incident sound levels are calculated at the receivers’ positions; this computation
is suitable for detailed estimates on micro-scale areas;
• those where the sound level at a grid point is interpolated or assumed to be representative of the
incident sound level at the receiver position; this computation is suitable for estimates on
macro-scale areas.
A further distinction is between:
• methods requiring the design of a receivers ring around the building on which distributing the
inhabitants in the building;
• methods assigning all the inhabitants in the building to a single value of incident sound level.
No guide is available to establish accuracy of assignment method but literature studies demonstrate
that END initial approach is for sure not the best one and now it has been changed accordingly. In fact,
the past common approach to assessing the exposure of people to transport noise was to use, as a proxy,
the exposure of the most exposed side of the dwelling in which they live This may not always be a good
approximation, however, because the rooms in which people spend most time may not be on the most
exposed side of the dwelling. Moreover, when dwellings are not available the same approach used the
main building façade, regardless the position of the dwelling.
One of the main problems of noise assignment is that the living environment is known with high
uncertainty: sometimes you can have dwellings details but mostly you have only buildings; sometimes
you have address points and you should connect address points to buildings even when address p oints
are placed at fences or property borders. Therefore, depending on assigning levels to dwellings,
buildings or address points, risk assessment varies and dose effects relationships may differs a lot ( 9).
In particular, predicted and observed %HA due to road traffic noise have been estimated and
compared according assignment methods as in Figure 2 in (9).
Figure 2 - Observed and predicted proportions of %HA residents as a function of road traffic noise exposure,
according to method and location of assessment. (from (9)).
5691
INTER-NOISE 2016
In addition to people location uncertainty, estimation of people exposure is often influenced a lso by
the way levels are assigned to the (uncertain) position. In fact, sometimes levels are calculated at
receivers (LUR models, façade calculations, receivers’ calculation) but often levels are derived from
noise maps, simply interpolating from grid points. In (10) a comparison was performed between
interpolated data and direct calculations, showing big differences especially at local level.
Interpolation could be a good solution for large scale studies to assess the average risk of a population,
but if you want to correlate health effects found in population to their relative exposure, you should
calculate noise directly.
Moreover, according the distribution method, in (10) it was demonstrated that first reported
exposure class (55-60 L DEN) is the most influenced by the method change. In that study, a test in Pisa
municipality demonstrated the difference in people distribution according to the assignment method
(see Figure 3).
10
10
9
8
END method
7
% Population exposed
% Population exposed
9
Interpolated
receiver calculation
6
5
4
3
8
Interpolated
receiver calculation
6
5
4
3
2
2
1
1
0
VBEB method
7
0
30
35
40
45
50
55
60
65
70
75
80
30
35
Lden dB(A)
40
45
50
55
60
65
70
75
80
Lden dB(A)
(a)
(b)
Figure 3 - Distributions of the population exposure to road traffic noise over the entire territory of Pisa’s
Municipality calculated by END (a) and VBEB (b) methods and noise levels at receivers or interpolated from
grid levels (from (10))
Finally, it has to be considered that the case of health surveys is very different from analysing the
whole population. When considering the exposure of the whole population the unknown factor is h ow
many people live in that dwelling and how much noise is associated. On the other hand , in health
surveys you know exactly how many people’s exposure you are evaluating, but you don’t know exactly
where is the dwelling, usually because address points are placed at fences or in front of main door of
large buildings. In this case, using the VBEB data (if available due to noise maps) it is possible to
estimate dwelling position in the large building using probability. In fact, the surveyed person might be
associated with the same probability to each point around the building, so its exposure is the average
one, meaning the energetic average of noise levels.
This method here briefly outlined is completely different from the association of the whole
population to the energetic average, which is still an overestimation of the population exposure.
4. INDICATORS FOR EPIDEMIOLOGICAL STUDIES AT DIFFERENT SCALES
Several indicators are available to assess risks. The choice is driven by several factors:
- Data availability (Frequency estimation, number of events, Leq/h…)
- Type of source (especially time pattern)
- Presence of multi-source context
- Health effect studied (acute or chronic, daytime or nighttime)
- Policy interests in the results (mitigation planning, communication iss ues)
- Extent of area involved
and maybe more…
So, not only the noise level indicators used may differ, but also the way of aggregating exposure
results may be different according the aim of the study. For example, it is well known ( 6) that noise
5692
INTER-NOISE 2016
equivalent levels suggested by END (L DEN and L night) are suitable for long term effects like annoyance
or generic sleep disturbance but cannot be used to foresee specific symptoms like arousals that can be
estimated for example with the number of events above a fixed threshold of noise. Similarly,
aggregating methods may vary according national and local legislations when indicators are compared
to limits to plan mitigations or if they are compared to health thresholds to evaluate risks zo nes.
In this paper we focus only on epidemiological studies and risk assessment so aggregations based
on limits comparison will not be mentioned. In any case, annoyance dose-effects relationship could be
used in different ways to assess the health risks of different areas. Some example of indicators will be
provided according the scale of the studied area. In fact, some aggregation methods are suited for large
scale contexts (regional or country comparisons) and others are feasible only at small scale (cit ies,
cohorts).
The DALY has been established by WHO to calculate the Environmental Burden of Disease (EBD),
it can be used to set priorities between different pollutants since it allows a comparative evaluation of
the health risks. The use of DALYs may also improve risk communication to population as their
number can be expressed as a fraction of total burden of disease, and thus compared to other sources of
health loss. In practice, given a health effect, citizens can know which percentage of cases are due to
noise and which one to other factors (other pollutants, habits, genetic factors etc...).
Moreover, DALYs can be applied to economic evaluation of environmental policies to compare the
health benefits of different options and establish the best deal in risk management, given limited
resources; or at least to analyse whether the actions are worth the money invested.
The DALY combines in one measure the years lived with disability (YLD) and the years lost due to
premature mortality (YLL) in the general population:
(1)
DALY  YLD  YLL
The YLL corresponds to the number of deaths multiplied by the standard life expectancy at the age
at which death occurs. The YLD is obtained multiplying the number of incident cases, a disability
weight (DW) and an average duration of disability in years. DW is estimated by literature studies for
each outcome.
The approach to estimate total disease burden can be summarized in the following steps:
 estimating the exposure distribution in a population;
 selecting one or more appropriate relative risk estimates from the literature,
 estimating the population-attributable fraction.
This approach requires the measurement or calculation of:
 the distribution of the exposure to noise within the population (prevalence of noise
exposure, estimates are required of the distribution of the exposure in the population of
interest using the chosen noise metric);
 the exposure–response relationship usually obtained from epidemiological studies for the
particular outcome;
 a population-based estimate of the incidence or prevalence of the outcome from surveys or
routinely reported statistics;
 a value of DW for each health outcome.
The exposure–response relationship may be reported as a regression formula or as a relative risk
measure for a given change in noise. They are established for each outcomes and published in various
literature studies.
The incidence or prevalence of the outcomes should be collected as for example, deaths from
cardiovascular disease in a population per year are usually available from national statistics. Other
outcomes, may not be available and in these cases prevalence or incidence may need to be determined
by surveys.
Disability weight quantifies time lived in various health states to be valued and quantified o n a
scale that takes account of societal preferences, a scale of 0–1, where 1 represents death and 0
represents ideal health. WHO has a reasonably comprehensive list of DWs and these are recommended
for use.
Finally, calculation of the attributable fraction is needed. It is the proportion of disease in the
population that is estimated to be caused by noise.
The methods described above are the most common approach used in health risk assessments
because the methodology has been established and accepted in comparative risk analysis of WHO’s
EBD projects.
However, data collection for these methods is not easy. If data availability is poor also the DALYs
5693
INTER-NOISE 2016
calculation would not be accurate. So other indicators might be used starting from easier formula
providing that they are correlated with annoyance. In this sense, indicators GDEN and Gnight for risk
assessment has been estimated in (11) based simply on L DEN and L night indicators. Their correlation
with %HA has been verified and their use for comparison of health risk has been tested at different
scales.
Moreover, a number of projects have defined different indicators able to rate priorities of different
mitigation sites. For example, an indicator was developed within the Qcity pro ject which is based on
annoyance dose effects relationships.
All the cited methods take into account the annoyance response to each different noise source (road,
railway, aircraft and industrial) and are therefore suited to multi exposure contexts. In fact , DALY and
Qcity methods refers directly to dose effects curves that are established for single sources. GDEN and
Gnight could be used without using dose-effects curves in single sources contexts but they can be
adapted to multi exposure contexts using the LDEN /Lnight road equivalent indicators in which noise
from sources other than road traffic are modified to express equally annoying values.
The formulation of GDEN is the following:
GDEN norm
 1
 10  log10 
N
 tot
 ni 10
i
LDEN i
10




dBA
(2)
where N tot is the total agglomeration population, n i is the population exposed to the i-th class of
exposure and L DEN,i is the representative value of i-th class of exposure or i-th inhabitants unit
depending on study extent, expressed as road equivalent noise level.
Cumulative GDEN for different sources can be obtained using LDEN weighted proposed by Miedema
and Borst (12); LDEN (and similarly Lnight ) values for rail(R) and aircraft(A) noise are transformed into
equally annoyed road traffic levels before calculating GDEN (11).
In the next paragraphs simple applications of these indicators are shown.
4.1 Indicators for large scale studies
GDEN calculation was performed in (11) to obtain a global rate and compare cities at EU level.
Equally spaced values have been assigned to reported exposure classes to compute L DENi
(52.5-57.5-62.5-67.5-72.5-77.5). This approximation obviously introduced errors which are larger for
those cities that have a huge number of people in extreme classes (<55 dBA and >75 dBA).
Thus, GDEN has been calculated for all cities and so calculated values are plotted in Figure
4Erreur ! Source du renvoi introuvable..
Figure 4 - Classes of GDEN values in European cities (from (11))
Similarly, values can be computed at Country level whenever data would be complete. In fact,
5694
INTER-NOISE 2016
actually the European Environmental Agency is establishing a method to fill the gap of missing
reported data. Methods have been proposed by a consultation panel under the consortium ETC/ACM
(European Topic Centre on Air Pollution and Climate Change Mitigation) to derive data based on most
similar data available: for example, if agglomeration data are missing and no previous mapping rounds
data are available, data from the same Country of similar extent cities are used, otherwise Country
average data are used or finally average European data. With the standardization of these methods it
would be possible to carry out large scale epidemiological analysis even when not all data have been
provided.
It can be observed that a large part of higher values are in South and Eastern Europe. In following
years these countries could provide improvements to raise awareness on noise & health issues. In fact,
only few countries have already a national legislation including limits for traffic noise, especially for
old infrastructures.
Moreover, national clusters are also evident: this means that maybe it is still diffic ult to consider
GDEN as a reliable indicator to compare all cities in EU because of national methods, different
methodologies to define traffic flows, different accuracies in estimating values but this will change
with the next implementation round where CNOSSOS –EU will be used by all countries. Therefore, at
this stage, GDEN could be useful at least for each Member State that can evaluate differences between
its cities and decides about funds/fee policies based on relative rating (especially in those coun tries
where a lot of cities have been already mapped).
Also DALYs can be used to compare cities relative risk, provided that DALYs values are
normalized by population. In Tuscany, exposure data collected within the provisions of END were
used to estimate DALY for road, railway and aircraft noise (13). DALYs were calculated according
exposure-response relationship for annoyance and sleep disturbance (6), considering people exposed
to values of L DEN above 55 dBA and to L night above 50 dBA in the three agglomerations with more than
100000 inhabitants and in Pisa, used as test case for noise strategic. Full people distribution in noise
exposure classes have been derived and the dose-effects relationships have been associated with the
central noise bands level. DALYs for each agglomeration, impact source and outcome were elaborated.
Figure 5 summarizes the results for road and railway noise in the four cities: DALYs values are
normalized by the number of inhabitants.
Figure 5 - Annoyance and Sleep disturbance DALYs, normalized to population, in Tuscany due to road and
railway noise
From the figure is clear that at Tuscany level:
 Livorno and Prato have a critical sleep disturbance due to road traffic noise;
 Road traffic annoyance is critical in all the large cities;
 Railway annoyance is critical in Pisa.
Therefore, DALY could be used to compare different agglomeration and this will became true also
at EU level with the implementation of the common method.
5695
INTER-NOISE 2016
4.2 Indicators for medium and small scale studies
In order to assess risk at medium scale, several opportunities are available. For example , to estimate
relative risks of cities areas, indicators can be calculated on cities sub-unit and mapped on GIS to
visualize results. Units might differ across Countries, regions depending on policies but a solution can
be census units or postal codes. Considering census units, different values of GDEN and DALYs have
been calculated in Pisa and Florence respectively (see Figure 6). In this comparison graduated colored
scales immediately highlight the most risky zones, pointing out mitigation urgency.
Figure 6 - GDEN in Pisa and Annoyance DALYs visualization in Florence
Other methods like the Qcity one are able to obtain finer estimations of critical areas, basing
calculation on grid step resolution. This would obviously provide a more detailed information to drive
mitigation actions but will cost more in terms of calculation time. In Figure 7 a test of the Qcity method
is reported for one of the risky census units found in Pisa with GDEN indicator. The method provide
further details pointing out the single buildings having priority (the U shaped buildings block is the
one that needs mitigation within the focused area). Qcity values are not in the same unit of levels so a
scale based on variations is provided.
Figure 7 – Qcity method applied to a critical area identified with GDEN method
Medium scale studies could also be based on LUR and mentioned indicators might be based on its
results, however being that LUR are not suitable to assess small scale variations, it is expected that
priorities rates would be more raw and similar within units.
As mentioned in previous paragraphs, small scale studies could also develop indicators based on
5696
INTER-NOISE 2016
measurements. In case of needing exposure of single citizens, noise indicators report ing the number of
events above a given threshold could be associated to health effect. For example N70 for aircraft noise
reports the number of events (aircraft overflights) having an L Max greater than 70 dBA. In the
research project of Pisa University on sleep disorders it was calculated to be correlated to arousals.
Here dwellings are reported in Figure 8 based on average measured values of N70 in the relaxing
period between 8 p.m. and 8 a.m.. Values ranges from 3 to 11 N70.
Figure 8 – N70 values for dwellings next to Pisa airport
5. CONCLUSIONS
The need of common methodologies for mapping and assigning levels have been demonstrated by
several studies and the work done within the revision of the END leads to new perspectives of
exposure analysis. The next implementation round, should provide large scale reliable data, finally
comparable. Accuracy of these data would also be evaluated and their suitability for epidemiological
studies should be questioned before trying to derive dose-effects relationships. Accuracy is expected
to be improved thanks to a better expertize of maps developers and to improved computation
possibilities. In fact, dose-effects relationships varies according different mapping approache s, and
their variability depends on accuracy of noise levels estimation. As presented in the paper, several
methods could be used at smaller scale to estimate noise levels: LUR methods and measurements could
be a solution to avoid mapping. LUR methods could be useful for epidemiological analysis
considering not only noise but also other pollutants, whose propagation could be derived similarly.
Measurements are obviously suited for high detailed studies especially on health effects needing
higher detail than equivalent noise levels. In fact, the risk assessment of health outcomes should follow
different approaches according the type of outcome considered in order to estimate the most suitable
indicator. Indicators have been identified in literature and several assessing experience have been
carried out.
Beside the DALY is the indicator identified by WHO for EBD estimation, several other indicators
may contribute to identify areas or citizen exposed to harmful effects of noise and there are ones that
are more suitable for driving mitigation choices. Different issues are in fact raised according not only
the extent of the studied area, but also the aim of the study, i.e. if the goal is mitigation priority or
health outcomes identification. The paper has shown some experiences in this sense and the efficacy of
a new indicator (GDEN) in highlighting risk areas at large scale. Moreover, DALY calculation examples
have been given as method not only for comparing risk factors but also as method for identifying risk
areas and to drive priorities across Countries.
REFERENCES
1. European Commission Working Group. Assessment of exposure to noise. Good practice guide for
strategic noise mapping and the production of associated data on noise exposure, Version 2. August
2007.
5697
INTER-NOISE 2016
2. Lercher P, de Greve B, Botteldooren D, Baulac M, Defrance J, Rüdisser J. The effect on annoyance
estimation of noise modeling procedures. Proc. 9th International Congress on Noise as a Public Health
Problem (ICBEN) 2008, Foxwoods, CT.
3. Kephalopoulos S; Paviotti M. Comparison between implementations of European noise mapping
methods, Proc. of the 16th International Congress on Sound and Vibration vol. 1 p. 1-8, Krakow, Poland
2009
4. Ascari E, Licitra G, Teti L, Cerchiai M, Low frequency noise impact from road traffic according to
different noise prediction methods, Science of The Total Environment, Volume 505, 1 February 2015,
Pages 658-669, ISSN 0048-9697.
5. Vorläufige Berechnungsmethode zur Ermittlung der Belastetenzahlen durch Umgebungslärm – VBEB,
Federal German Gazette, 20/04/2007.
6. Burden of disease from environmental noise Quantification of healthy life years lost in Europe, WHO
Regional Office for Europe, 2011
7. Ragettli M.S.; Goudreau S.; Plante C.; Perron S.; Fournier M.; Smargiassi A. Annoyance from Road
Traffic, Trains, Airplanes and from Total Environmental Noise Levels. Int. J. Environ. Res. Public
Health 2016, 13, 90.
8. JCGM 100:2008, GUM 1995 with minor corrections: Evaluation of measurement data - Guide to the
expression of uncertainty in measurement, September 2008
9. Eriksson C, Nilsson ME, Stenkvist D, Bellander T, Pershagen G. Residential traffic noise exposure
assessment: application and evaluation of European Environmental Noise Directive maps. J Expo Sci
Environ Epidemiol. 2013 Sep-Oct;23(5):531-8. doi: 10.1038/jes.2012.60.
10. Licitra G, Ascari E, Brambilla G. Comparative Analysis of Methods to Estimate Urban Noise Exposure
of Inhabitants, Acta Acustica united with Acustica, Volume 98, Number 4, July/ August 2012 , pp.
659-666(8).
11. Licitra G, Ascari E, GDEN: An indicator for European noise maps comparison and to support action plans,
Science of The Total Environment, Volumes 482–483, 1 June 2014, Pages 411-419, ISSN 0048-9697
12. Miedema HME, Borst HC, 2007. Rating environmental noise on the basis of noise maps, Quiet City
Transport Project D 1.5
13. Palazzuoli D, Licitra G, Chiari C, Le esternalità dei trasporti: l'impiego dell'indicatore DALY per la
valutazione degli effetti sanitari dell’esposizione al rumore nei principali agglomerati in Toscana, Atti
della conferenza SIET 2014.
5698