Citizens’ exposure to predominant noise sources in agglomerations

2.1 NSP and I-NSP maps

NSP and I-NSP maps were initially developed within the Maritime Interreg Project [40] to identify the best locations for monitoring port noise. They were then previously presented to assess the impact of port noise on the surrounding area [41]. Both maps are built around the predominance criterion inspired by the UNI 10855:1999 [34] for the sound-level evaluation of a specific source in the presence of non-negligible background noise. According to the UNI, the minimum criterion to be able to evaluate a source in those conditions is that the specific source level must be equal to or higher than the sum of the levels produced by all the other sources. The criterion is formalized by Equation (1):

where N is the number of the sources present in the environment, L i is the level of source under investigation, and L j are the source levels refer to the remaining ones.

The criterion is equivalent to require that the environmental noise should be at least 3 dB(A) higher than the background noise, and then, the source noise level can be extrapolated from environmental noise measurements. Citizens can easily perceive the source where a predominance exists. If the predominant source would be mitigated, the overall acoustic energy would be reduced by half or less. During the process of creating NSP and I-NSP maps, the criterion is evaluated in each point of the calculation grid and the predominance is attributed if present. Then, a symbol is attributed to each point according to the source legend in Figure 1. For the I-NSP map instead, the shape of the symbol is determined according to the type of predominant source, while the color represent the noise level and is assigned following the color scale in Figure 1 [42].

Figure 1

Legend for NSP map (left), and color scale for noise levels in the I-NSP map (right).

2.2 PE

In this work, predominance criterion is extended to the evaluation of the citizens’ exposure. Based on the provided data, the predominance was here assigned by the authors to population using the assignment methodology provided by the Directive 2015/996 [43], which asked to distribute inhabitants according to façades length. Receiver points are placed in front of building façades of residential buildings and levels are computed excluding reflections from the façade under investigation.

According to this method, the population of the building was evenly distributed to the receivers around all the building façades. Then, the predominance was evaluated for each receiver, and the predominance was associated with the corresponding number of inhabitants. Once the predominance was assigned, histograms representing the number of people corresponding to the PE of each source at different noise levels were produced. According to Annex VI of the END, noise exposure histograms must be rounded to the nearest hundred (e.g., 5,200 = between 5,150 and 5,249; 100 = between 50 and 149; 0 = less than 50). For each noise interval, the bars of each source indicate the number of inhabitants subjected to the predominance exposure of the source itself, with the relative percentage to the city’s total population above.

It must be highlighted that predominance is established at all points around the façades, which is a different approach compared to the one suggested by the last update of END [44]. In fact, when no information of the position of dwelling inside buildings and inhabitants is available, the directive proposes two different methods on a building-by-building basis for the estimation of the exposure to noise of the dwellings and inhabitants within the buildings. If dwellings in buildings are such that they have a single façade exposed to noise, the allocation of the number of dwellings and inhabitants to receiver points shall be weighted by the length of the represented façade. Otherwise, the set of associated receiver locations shall be split into a lower and upper half based on the median value of the calculated levels for each building and the number of dwellings and inhabitants shall be distributed equally for each receiver point on the upper half of the data set.

However, this building-by-building approach required by END is not applicable to the present work, and probably also by others in the near future, because the number of buildings whose interior distribution of dwellings is unknown was very high for all the investigated cities. Therefore, the building-by-building assessments are still very time-consuming at present. Furthermore, as a result of variations in noise-level distribution around the building, allocating population to the upper half of the dataset for each building can result in a non-uniform population distribution among different sources. This is because a varying number of inhabitants would be assigned to the same receiver point for each source, making impossible PE calculation due to ambiguous results.

2.3 Test areas

Noise maps data elaborated in this work were acquired from the strategic noise maps of the four Italian cities of Bolzano, Livorno, Padova, and Pisa. Pisa noise strategic map was not required by law but it is a relevant case study as it includes all the sources addressed by the END within a small city with less than 100,000 inhabitants, as sketched in Supplementary Material. The other cities are all agglomerations with more than 100,000 inhabitants, including road, railway and industrial noise, but without airport.

Unfortunately, even if for Livorno the port noise was mapped within the mentioned projects, the available map used in this work was produced in force of the current regulation; thus, the port area has been considered as mere industrial sources. All maps were provided to authors by relative authorities and consultants, and were previously developed with interim methods [7] before the last mapping round requiring the application of the CNOSSOS-EU Common noise assessment methods in Europe [8]. Thus, road noise maps are created with NMPB-Routes-96 [45] interim method, railway noise maps are realized with national RFI-INAC Italian model [46], industrial noise maps are made following ISO 9613-2 [47] interim method, and airport noise maps are built with ECAC29 interim methodology [48]. The noise maps, acting as input for the present work, were provided to the authors with a grid pitch length of 10 m, 10 m, 5 m, and 5 m for Bolzano, Livorno, Padova, and Pisa, respectively. Besides, exposure of citizens was originally evaluated assigning maximum levels to all inhabitants of each building unit; exposure evaluation within this article is carried out only if the level of each façade is available, i.e., if the noise assessed all around the buildings is available. Unfortunately, this was not the case of Bolzano, which is then not used for PE analysis. At the time of noise maps were computed, Livorno had 143,683 inhabitants, Padova 196,289, Bolzano 106,622, and Pisa 90,870.

3.1 NSP and I-NSP

The obtained NSP and I-NSP maps for the L DEN metric of the whole municipalities have been computed following the previously described method and are reported in supplementary materials for a better visualization. Considering the size of agglomerations and the typical pitch size of the calculation grid (5–20 m), it is impossible to distinguish individual symbols at full view. Thus, the predominance areas in an NSP map appear made of solid colors, confirming the faster identification of each source predominance over the others introduced for port areas [33]. At full view, I-NSP map looks like a standard strategic noise map, but the embedded information about predominance to be used for in-depth analysis of the results emerges with a closer view. For this purpose, the present paragraph reports zoom areas of the investigated cities as examples.

As expected, all the agglomerations are mainly characterized by road predominance due to the capillary diffusion of road infrastructure over the territory. Railways and industrial predominance are more restricted and located close to the source, except for some particular situation that NSP maps helped to emerge. NSP emphasized the significance of buildings in the propagation of sound in urban areas. Screening or reflection effects can cause unexpected areas where road noise dominates over railway noise and vice versa. For instance, in the vicinity of the railway reported on the left of Figure 2, the predominance of road traffic can be due to unexpectedly high flows, whereas on the side of buildings opposite the railway, the road noise might be shielded and rail noise might unpredictably become the predominant sound once again. This information is valuable for studying mitigation strategies for sources that exceed limits at a building as it enables the avoidance of unnecessary actions by identifying the responsible source for a tuned effort.

Figure 2

Top: Zoom of NSP map of Padova for the L DEN metric. Bottom: Zoom of I-NSP map of Padova for the L DEN metric.

If the strategic noise map had taken into account the reported dissimilarity between real and simulated railway noise [49], the level of exposure to railway noise could have been increased and described more precisely. In fact, railway noise models only account for ordinary train transits and neglect “unconventional noises” such as maneuverings, loadings, truck movements, braking, squeals, and whistles. In the authors’ case study, the inclusion of these supplementary sounds resulted in an average increase of 5.5 dB(A) in noise levels, which exhibited a varying distribution across the area depending on the type of sound. For instance, simulations overestimate noise at stationary locations such as marshalling yards and railway stations, but underestimate it around curves, entrances to stations, and washing areas.

In Bolzano, Livorno, and Padova, hard edges have appeared along the railway lines, as shown on the left of Figure 2 for Padova. They can be attributed to two different reasons: (1) the absence of obstacles in the propagation, which causes the railway noise level to fall below the road traffic background level at a certain distance from the railway, and then suddenly changes the predominance assessment; (2) an error in the preparation of the overall strategic noise map, which is prepared by technicians by aggregating the specific map provided by the railway operator with the road map prepared on behalf of the municipality. In these cases, the most likely hypothesis is that the maps of the railway infrastructure have only been made within a certain distance from the center of the tracks, leading to an incorrect calculation of the predominance beyond this certain distance, because no noise level is available.

The solution came from the I-NSP map reported on the right of Figure 2, which shows how hard edges are present also in terms of levels. Such an evaluation was possible using the standard strategic noise map; however, the NSP map provided clearer results and enabled the identification of the fault’s source. In this regard, NSP maps prove to be efficient also for the inspection bodies designed to verify the quality of the submitted strategy maps.

In general, for all the agglomerations, industrial predominance occurred almost only around industrial areas, or the port area in the case of Livorno. In Bolzano, some industrial predominance emerged even far from the source, an unexpected result from the NSP map that showed how slopes in mountain territories play a key role in sound propagation. In such orography, then, some sources can disturb even in areas where they are not predicted, as reported in Figure 3.

Figure 3

Zoom of NSP map of Bolzano for the L DEN metric.

Among the investigated agglomerations, airport source was only present in Pisa. Even in the presence of all other END sources, the NSP map reports large predominance of airport noise in all the territory, even very far from the airfield itself. Predominance is well evident and expected on the west side of the agglomeration, in a wide quite area consisting of a natural park without any ground sources. As reported in Figure 4, surprisingly, airport predominance is very frequent also in the city center, especially when the other sources have obstructed sound propagation paths. This result confirms the near-impossibility of shielding receivers from the overhead source and the high intrusiveness of airport noise in the urban setting, where routes pass overhead during take-off despite past mitigations and studies [50].

Figure 4

Airport noise predominance in Pisa city center.

Also, non-predominance areas can provide valuable information. The close zoom to an area in Livorno reported in Figure 5 is analyzed in detail as an example. L DEN NSP map is on the left side, and L DEN I-NSP is on the right side for combining the visualization of the same region. The selected zoom is a boundary of multiple areas of predominance: green pentagons in the upper-left side indicate industrial noise predominance, i.e., port noise and light blue rhombuses on the right side are railway noise predominance from major railways, and orange circles denote road noise predominance due to urban traffic. The middle is characterized by white circles, indicating no predominance. This area is like a buffer area where no source emerges over the others. The comparison of L DEN maps with L Night maps illustrates the evolution of non-predominant areas from the average period to the night period due to differing source working hours. At a specific point on the boundary, a predominance may occur during nighttime, but not during daytime, and vice versa. This comparison provides a straightforward approach to interpreting the map and the relative differences among the noise levels generated by different sources. In this specific case, non-predominance area moves toward the south, and the road noise predominance area decreases and yields ground to industrial noise predominance. However, this day-night dynamic in non-predominance areas can be expected around any local roads, where traffic flow is generally significantly reduced at night, confining with industrial/port having 24-h activities. A similar situation is observed when considering the border between areas dominated by industry/port and railways. For the L Night metric, there is a shift toward the railways, indicating a higher reduction of noise from railways than from industry/port during the night. This type of observation is more feasible in areas with near-free-field propagation, where few to no obstacles exist. However, similar evaluations can still be conducted in areas with dense urban structures. The reported example showed the important role played by I-NSP map, which is able to represent both the predominance and the value of the noise metric, thus allowing us to determine the responsibilities.

Figure 5

Non-predominance area reported in the NSP map (top-left) and I-NSP map (top-right) of Livorno for the L DEN metric, and NSP map (bottom-left) and I-NSP map (bottom-right) for the L Night . Non-predominance area is between three sources: industrial, railway (depicted with a red and white dashed line), and road.

3.2 PE

This section reports the histograms for the PE for each city’s L DEN metric, calculated with the methodology described in Paragraph 2.2. The same histograms were also produced for the L Night metric, but they are not shown for reasons of space and ease of reading. However, comparing the PEs for the two metrics is useful to study how the temporal evolution of different sources impacts the population.

PE of Livorno with L DEN metric is reported in Figure 6 (left). In accordance with the noise maps, road noise resulted to be the most impacting source on population, with road PE clearly evident in all bands from below 40 dB(A) up to 75 dB(A) and with maximum in 55–60 dB(A) band. Industrial noise PE resulted to be confined between 40 and 55 dB(A), with no more than 3,000 inhabitants suffering from its predominance. This low level of exposure can be partly attributed to the sufficient distance between residential and industrial/port areas. However, the main contributing factor is the underestimation of noise sources from the port in the current official strategic noise map. Railway noise predominates significantly in the city due to the presence of a busy railway branch that stretches throughout its entire length. This results in a large number of exposed residents, and the potential effects are critical enough to warrant attention from managers.

Figure 6

PE of the agglomerations with L DEN metric. Livorno (top), Padova (center), and Pisa (bottom).

Specifically, it is necessary to investigate the elevated levels of railway dominance within the 65–70 dB(A) and > 70 dB(A) ranges. These levels pose a potential source of compliance and health-related issues. The PE facilitated their discovery, and if utilized in combination with a thorough map analysis, they can be located.

If the predominance at high levels is concentrated in a singular area, the tools can provide valuable support in prioritizing noise action plans. In this particular instance, this area can be readily resolved through a targeted intervention.

In Figure 6 (center), PE distribution for the agglomeration of Padova is reported. Road noise confirms to be the most impact on PE, followed by railways. Industrial noise was highlighted by NSP to generate a small predominance area but, on the contrary, it does not generate significant PE. A high emitter that does not affect a notable number of citizens is indicative of effective noise source localization and management by the agglomeration. It is remarkable how the PE provides significant information when compared to NSP by indicating whether the predominance is into inhabited areas or not.

In Figure 6 (right), PE of the city of Pisa is reported for the L DEN metric. Road traffic noise is significantly dominating over all other sources, especially at high levels. This is probably due to the presence of a highway and a high-flow road passing close to the suburbs, but also to the conformations of narrow streets with very close buildings typical of old town centers. Industries appear to be well located far from residential areas, although the few inhabitants who are exposed to their influence may potentially complain. Railways seem to have the same behavior of the other agglomerations. PE evaluation of the airport indicates that it does not have a significant presence in the highest level bins. This might lead to believe that airport noise is not a concern. However, exposure to 55–60 dB(A) of airport noise is sufficient to require attention to avoid disturbance.

The examples given demonstrate that with PE, it is easier to make objective observations such as “predominances at high levels impact the population, or not” or “there is no industrial predominance above level x in this agglomeration,” or to verify the presence or absence of all sources in the agglomeration in an easy way. These are all useful responses for administrations to better manage the territory and aim mitigation efforts more effectively.

Finally, the PE, expressed as a percentage, facilitates an immediate comparison of the impact of different sources among various agglomerates, as illustrated in Supplementary material and Table 1 for road traffic noise, Table 2 for railway noise, and Table 3 for industrial noise. It is clear that Padova, among the cities, has the highest exposure to road and railway noise levels above 60 dB(A). Livorno, on the other hand, has serious industrial noise predominance. Although the exposure levels are all below 60 dB(A), it is apparent that the port’s presence significantly impacts the city, necessitating specific and thorough research.