Road traffic noise in European cities: Learnings from a European-wide study on the health impacts of road traffic noise.

Sasha Khomenko 1 Barcelona Institute for Global Health (ISGlobal) Carrer Dr. Aiguader, 88, 08003, Barcelona, Spain Marta Cirach 2 Barcelona Institute for Global Health (ISGlobal) Carrer Dr. Aiguader, 88, 08003, Barcelona, Spain Mark Nieuwenhuijsen 3 Barcelona Institute for Global Health (ISGlobal) Carrer Dr. Aiguader, 88, 08003, Barcelona, Spain

ABSTRACT

Environmental noise is a main environmental risk to health and wellbeing. In a previous study, we evaluated the health impacts of road traffic noise for cities in Europe and estimated that almost 60 million adults were exposed to noise levels harmful for health. In this paper, we describe the main methodological difficulties encountered in our European-wide assessment. Our findings indicate the need to increase the coverage and quality of strategic noise maps. We propose several solutions that could be implemented under the Environmental Noise Directive (END) legislation. These include mechanisms to increase data coverage and protocols to ensure the delivery of harmonized and good- quality data. The END can be leveraged as a main mechanism to provide strategic noise maps for research purposes. We hope that the proposed solutions can help the research community to more easily retrieve noise data and conduct research on the health effects of environmental noise.

1. INTRODUCTION

Environmental noise is a major environmental risk to health and wellbeing (1,2). Sources of environmental noise include transportation (road, rail and air traffic), industries and leisure activities (such as night leisure) (1). Among these, transportation is the primary source of environmental noise, and considered the second main environmental cause of detrimental health effects in western Europe, after particulate matter (3,4). In European cities, noise from road traffic is the most prevalent source of environmental noise (2).

Exposure to environmental noise has been related to multiple adverse health effects, highlighting the importance of this environmental stressor. The main health outcomes include sleep disturbance, annoyance, cardiovascular and metabolic disease, adverse birth outcomes, cognitive impairment and poor mental health and wellbeing (1,5–9). In the long-term, exposure to environmental noise might cause a sustained stress response, which leads to the activation of the sympathetic nervous system and endocrine system and results in the release of stress hormones (7). This response can lead to increases in the heart rate, blood pressure and vasoconstriction, eventually favouring the development of chronic diseases, such as cardiovascular and metabolic disease, depression and anxiety disorders (6,7). For road traffic noise, the World Health Organization (WHO) recommends to reduce exposure below 53 dB for the average 24-hour exposure and below 45 dB for nigh-time exposure to protect health (1).

1 sasha.khomenko@isglobal.org 2 marta.cirach@isglobal.org 3 mark.nieuwenhuijsen@isglobal.org

Currently, the main European legislative framework for environmental noise control is the Environmental Noise Directive (END, Directive 2002/49/EC) (2,10). The END requires the member states to produce strategic noise maps every five years, calculate the number of people exposed to noise from road, rail and air traffic inside and outside of urban areas and to develop plans to control and reduce exposure to environmental noise (2,10). The last round of environmental noise mapping under the END was conducted in 2017, while the next one is planned for 2022.

In a previous study we retrieved strategic noise maps for European cities delivered under the END in the 2017 round of noise mapping and estimated population exposure and health impacts due to road traffic noise (11). We estimated that almost 60 million adults were exposed to road traffic noise levels harmful for health, which equated to a median of 42% of the adult population across the analysed cities (11). Notably, the main challenges of our study were the availability and quality of strategic noise maps (11). The objective of this paper is to provide an in-depth description on the main methodological issues encountered in our previous study and to discuss plausible solutions that could be implemented under the END for future European-wide assessments of the health effects of environmental noise.

2. METHODS

This paper is based on a previous study on the health impacts of road traffic noise in European cities (11). We follow a descriptive approach to emphasize the main methodological issues encountered in this previous study. We focus on the availability and quality of data and methods to estimate population exposure to road traffic noise. The findings described here correspond to the round of noise mapping conducted in 2017. Data collection for this study was done in 2020. The scope of our study were 978 cities and 49 greater cities in 31 European countries, defined in the Urban Audit 2018 dataset (12). We focus on strategic noise maps for urban areas that show road traffic noise exposure and employ the 24 h day-evening-night noise level indicator (L den ) (11).

3. RESULTS

3.1. Data availability

Due to the large scope of our study, it was agreed that the best approach would be to retrieve strategic noise maps from a common European data source. Accordingly, our first step was to review the strategic noise maps delivered under the END and available through the Eionet Central Data Repository (13). For several countries in which we observed data gaps and had knowledge about the availability of strategic noise maps from local sources (such as research institutes and public institutions), we used this additional data to cover the data gaps (14–19).

Once data collection was completed, we explored the extent of data coverage. We observed that for 6 out of the 31 countries there were no strategic noise maps available, while only 5 countries had complete data coverage for all cities (see Table 1 ). For the remaining 20 countries data coverage varied from 7 to 90%, with 9 countries having data coverage of less than 50% (see Table 1 ). We generally observed lower data coverage in eastern and southern European countries (such as Croatia, Cyprus, Greece, Romania and Slovakia).

To increase data coverage, we implemented a previously described gap filling procedure to estimate noise exposure in cities without available data (20). We constructed prediction models based on the available strategic noise maps that and predictor variables which included total city population and population density, city size and the length of distinct road typologies (highways, primary, secondary and tertiary roads) (11). While model validation indicated that this could be an acceptable method to estimate noise exposure when strategic noise maps are not available, we observed that the

models tend to predict similar exposure values for cities within a country and were not able to detect outliers (e.g. cities with very low and very high exposure levels) (11).

The extent of data coverage and the need to apply a gap filling methodology pointed out the need to increase the availability of strategic noise maps that are delivered under the END.

Table 1. Road traffic noise data availability. Countries are ordered from highest to lowest data coverage.

Country Defined cities (n) Cities with data (n) % cities with data Data sources and year Luxembourg 1 1 100% END, 2017 Malta 1 1 100% END, 2017 Netherlands 47 47 100% RIVM, 2017 Slovenia 2 2 100% END, 2017 Switzerland 17 17 100% END, 2017 Finland 10 9 90% END, 2017 Austria 6 5 83% END, 2017 Lithuania 6 5 83% END, 2012, 2017 Belgium 11 8 73% END, MIRA, 2012,

2017, 2018 Estonia 3 2 67% END, 2017 Norway 6 4 67% END, 2017 United Kingdom 190 124 65% CNOSSOS-EU, END,

2017 Denmark 5 3 60% END, 2017 Germany 127 68 54% END, 2017 Czech Republic 18 9 50% END, 2017 Ireland 6 3 50% END, 2017 Hungary 19 7 37% END, 2017 Spain 143 48 34% Catalonian government,

END, Madrid City Council, 2012, 2016,

2017 Bulgaria 18 6 33% END, 2017 Italy 94 27 29% END, 2017 France 114 29 25% Bruitparif, END, 2017 Latvia 4 1 25% END, 2017 Poland 69 17 25% END, 2017 Portugal 27 2 7% END, 2017 Sweden 14 1 7% Stockholm City Council,

2017 Croatia 7 0 0% Cyprus 2 0 0% Greece 16 0 0% Iceland 1 0 0% Romania 35 0 0% Slovakia 8 0 0% * CNOSSOS: Common noise assessment methods; END: Environmental Noise Directive; RIVM: National Institute for Public Health and the Environment. ** MIRA refers to strategic noise maps produced by Geopunt Flanders.

3.2. Data quality

After exploring the extent of data availability, we studied the quality of the available strategic noise maps. We found considerable variability in data formats, exposure ranges and exposure categorization. The majority of the strategic noise maps were available in polygon format (72%), followed by raster (17%) and polyline (11%) formats (see Table 2 ). The raster maps showed continuous and highly resolved exposures generally in pixels of 10m resolution and the majority including exposure levels below 55 dB L den (see Table 2 ). The polygon maps generally represented

exposure as categorical 5-dB noise bands within the inner areas of isophones (see Table 2 ). The polyline maps depicted noise levels through street lines or isophones also generally categorizing exposure in 5-dB noise bands (see Table 2 ).

We considered the raster maps to be of higher quality for health impact assessment (HIA) as these showed highly resolved and continuous exposures, which allowed for a more accurate evaluation of population exposure to road traffic noise (11). Contrastingly, the polyline maps were considered of lower quality as these often showed exposure only at the street level (without noise propagation to the buildings) (11). The polygon maps were generally considered of moderate quality as these often categorized noise exposure in 5-dB noise bands, which is a wide range and less detailed compared to the continuous exposures shown in the raster maps (11) (see Table 2 ).

We assessed the overall quality of the strategic noise maps by considering the noise map format and the range and categorization of exposure levels (11). In addition, we evaluated the plausibility of exposure levels (based on the lack of quiet areas) and noise propagation from the street level (11). As detailed above, strategic noise maps were considered of higher quality for HIA when these showed continuous exposure levels and were in raster format. The strategic noise maps were considered of lower quality when these did not have quiet areas and did not display noise propagation from the street level, which generally applied to polyline maps (11). Overall, 17% of strategic noise maps were considered of high quality, while 43% and 40% of the maps were considered of moderate and low quality, respectively (see Table 2 ) (11). These findings highlighted the need to improve the quality of the available strategic noise maps. Table 2. Road traffic noise data quality.

Noise map format

City count

% of cities

Noise exposure range *

City count (n)

% of cities

Noise exposure categorization

City count (n)

% of cities

Noise map quality

City count (n)

% of cities

(n)

Raster 76 17% < 35 dB 75 99% Continuous 76 100% High 75 99% 40-50 dB 0 0% Categorized in 5-dB noise bands

0 0% Moderate 1 1%

> 55 dB 1 1% Low 0 0% Polygon 321 72% < 35 dB 18 6% Continuous 123 38% High 0 0% 40-50 dB 193 60% Categorized in 5-dB noise bands

198 62% Moderate 188 59%

> 55 dB 110 34% Low 133 41% Polyline 49 11% < 35 dB 9 18% Continuous 13 27% High 0 0% 40-50 dB 4 8% Categorized in 5-dB noise bands

36 73% Moderate 1 2%

> 55 dB 36 74% Low 48 98% * The noise exposure range indicates the lowest noise level shown in the strategic noise maps.

3.3. Methods to estimate population exposure

Due to the variability in data formats, exposure ranges and exposure categorization, a harmonized method to estimate population exposure to road traffic noise had to be developed for each format of the strategic noise maps (i.e. raster, polygon and polyline) (11). Common European data sources were employed for all cities. We outline our approach below.

Given that the strategic noise maps generally categorized noise exposure in 5-dB noise bands and often only showed exposure levels starting at 55 dB (see Table 2 ), we estimated the population distribution in 5-dB noise bands: < 55 dB, 55–59 dB, 60–64 dB, 65–69 dB, 70–74 dB and ≥ 75 dB

L den for all strategic noise maps. We combined population data from the Global Human Settlement Layer (GHSL) (21), data on residential built-up area from the European Settlement Map (ESM) (22,23) and the strategic noise maps (11). Our initial step was to distribute the population data from the GHSL among the residential building units from the ESM weighting by their area. Afterwards, the method to estimate exposure varied for each format of the strategic noise maps. For raster maps, the population exposure in each building unit was directly proportional to the pixels corresponding to each noise level. For polygon maps, the exposure in each building unit corresponded to the median noise levels of the intersecting noise bands. Finally, for polyline maps, the exposure in each building unit corresponded to the nearest street or isophone noise value (11).

We applied the methodology to estimate population exposure to road traffic noise for each strategic noise map and compared the estimates for each noise map format. We observed that the exposure estimates from raster and polygon maps were more comparable than the ones obtained from polyline maps, which were generally higher (see Figure 1 ). We estimated a median of 39% and 45% of population exposed above 55 dB L den for raster and polygon maps, respectively, compared to a median of 84% for polyline maps (see Figure 1 ). Given the state of data availability it was not possible to evaluate whether the higher exposure for polyline maps was due to the method or due to a “true” higher exposure in cities with polyline maps. Nevertheless, we believe that the method for polyline maps was prone to overestimations by assuming that the noise value from the nearest street or isophone was the same as at the building façade. These findings indicated that the use of polyline maps can lead to potential overestimations of population exposure and that potentially more accurate estimates can be obtained from raster and polygon maps.

Uep7 @P 9g < uonendod % 25° ‘polygons

Figure 1 . Comparison of population exposure to road traffic noise estimates by noise map format. Figure adopted from Khomenko et al., 2022 (11).

4. DISCUSSION

In this paper we outline the main methodological issues encountered in our previous study on the health impacts of road traffic noise for cities in Europe. We highlight limitations regarding the availability and quality of data and methods to estimate noise exposure. Our findings indicate the need to increase the availability and quality of strategic noise maps. We believe this will allow to

obtain more accurate estimates on population exposure and health effects of road traffic noise. Below we propose several solutions that could be implemented under the END to tackle these issues.

Mechanisms to increase data coverage . Under the END, member states are required to produce strategic noise maps for urban agglomerations of more than 100,000 inhabitants (10). Nevertheless, at the time of data retrieval, we observed that a significant proportion of countries still lacked or had very low data coverages. It must be noted that the scope of our study were cities of more than 50,000 inhabitants defined in the Urban Audit database (12). In this way, the described data gap could be partially due to the lack of obligation to deliver strategic noise maps for cities of 50,000 to 100,000 residents. To increase data coverage, we propose to extend the obligation to deliver noise data to smaller size cities of less that 100,000 inhabitants. Despite lower populations, exposure to road traffic noise could still be an important environmental factor in these urban areas. The delivery of strategic noise maps for smaller size cities will allow to study any specificities on noise exposure and health effects in these areas.

Protocols to standardize the strategic noise maps . We observed great variability in the available strategic noise maps (11). The main sources of variability were the data formats, the distinct exposure ranges and exposure categorization. This variability was a limitation for our study as it required big time investments for data harmonization and led to distinct approaches to estimate population exposure levels based on the data formats, limiting the comparability between the analysed cities. A protocol to ensure that all maps are delivered in a standardized form would help to address this limitation. Our proposal is to ensure that all maps are delivered in the same data format and show the same exposure ranges and categorization. Given our quality considerations and methods to estimate population exposure, we propose that noise maps are delivered in raster format, show continuous exposure levels and include exposures below 55 dB L den . We believe this will allow for a more accurate evaluation of population exposure and will additionally allow to study the health effect of road traffic noise below the current WHO recommendation of 53 dB L den .

Mechanisms to ensure the delivery of the best-available maps . Based on our quality criteria, only 17% of the strategic noise maps were considered of high quality for HIA (11). We downgraded the quality of the strategic noise maps when these categorized exposure levels, only showed exposure levels starting at 55 dB L den or did not display noise propagation from the street level (which generally applied to maps in polygon and polyline formats). While local administrations might produce good- quality strategic noise maps, it is possible that simplified versions are delivered under the END, which represent less detailed exposure levels. We believe it is important to ensure the delivery of the best- available noise maps and outline quality criteria to address this limitation. Our proposal is to align the quality criteria with the protocols to standardize the strategic noise maps. Based on our experience, we propose that quality criteria include the delivery of data in raster format, the mapping of continuous exposures which show noise propagation from the streets to the buildings and the inclusion of wider exposure ranges that go below 55 dB L den .

Common and public European database for researchers . While the END is a useful mechanism to ensure the reporting of noise data in Europe, it can also be leveraged as a main mechanism to provide noise data for research purposes. Driving from the proposals detailed above, the END has the potential to improve data coverage and promote the delivery of harmonized and good-quality strategic noise maps. At the time of data retrieval, strategic noise maps were available through the Eionet Central Data Repository (13), while the data for the 2022 round of noise mapping will be available through the new Reportnet 3.0 system (24). We propose the creation of a centralized European database for environmental noise in which the most recent and best available strategic noise maps are uploaded for each country and city, following a similar approach as for air pollution (25). We believe this will help the research community to more easily retrieve harmonized and good-quality noise data and evaluate more accurately the health effects of environmental noise.

5. CONCLUSION

In this paper we present the main methodological difficulties encountered in our previous European-wide study on the health impacts of road traffic noise. Our findings indicate the need to increase the coverage and quality of strategic noise maps. We propose solutions that can be implemented under the END legislation. These include mechanisms to increase data coverage and protocols to ensure the delivery of harmonized and good-quality data. The END can be leveraged as a main mechanism to provide strategic noise maps for research purposes. We hope the proposed solutions can help the research community to more easily retrieve noise data and to conduct research on the health effects of environmental noise.

6. REFERENCES

1. WHO. WHO environmental noise guidelines for the European Region [Internet]. 2018. Available from: https://www.euro.who.int/en/health-topics/environment-and-health/noise/environmental-noise- guidelines-for-the-european-region

2. EEA. Environmental noise in Europe - 2020 [Internet]. European Environment Agency. 2020. Available from: https://www.eea.europa.eu/publications/environmental-noise-in-europe

3. Hänninen O, Knol AB, Jantunen M, Lim TA, Conrad A, Rappolder M, et al. Environmental burden of disease in Europe: Assessing nine risk factors in six countries. Environ Health Perspect. 2014;122(5):439–46.

4. WHO, JRC. Burden of Disease from Environmental Noise - Quantification of healthy life years lost in Europe [Internet]. 2011. Available from: https://apps.who.int/iris/handle/10665/326424

5. Basner M, McGuire S. WHO environmental noise guidelines for the european region: A systematic review on environmental noise and effects on sleep. Int J Environ Res Public Health. 2018;(15):519.

6. Clark C, Paunovic K. WHO environmental noise guidelines for the European region: A systematic review on environmental noise and quality of life, wellbeing and mental health. Int J Environ Res Public Health. 2018;(15):2400.

7. Eriksson C, Pershagen G, Nilsson M. Biological mechanisms related to cardiovascular and metabolic effects by environmental noise [Internet]. European Commission Joint Research Centre. 2018. Available from: https://www.euro.who.int/__data/assets/pdf_file/0008/383921/noise-guidelines- eng.pdf

8. Guski R, Schreckenberg D, Schuemer R. WHO environmental noise guidelines for the European region: A systematic review on environmental noise and annoyance. Int J Environ Res Public Health. 2017;14(12):1–39.

9. van Kempen E, Casas M, Pershagen G, Foraster M. WHO environmental noise guidelines for the European region: A systematic review on environmental noise and cardiovascular and metabolic effects: A summary. Int J Environ Res Public Health. 2018;15(2):1–59.

10. European Commission. Environmental Noise Directive [Internet]. 2002. Available from: https://ec.europa.eu/environment/noise/directive_en.htm

11. Khomenko S, Cirach M, Barrera-Gómez J, Pereira-Barboza E, Iungman T, Mueller N, et al. Impact of road traffic noise on annoyance and preventable mortality in European cities: A health impact assessment. Environ Int. 2022;162:107160.

12. Eurostat. Urban Audit [Internet]. 2018. Available from: https://ec.europa.eu/eurostat/web/gisco/geodata/reference-data/administrative-units-statistical- units/urban-audit

13. EEA. EIONET - Central Data Repository [Internet]. 2020. Available from: https://cdr.eionet.europa.eu/

14. RIVM. Environmental Health Atlas [Internet]. 2017. Available from: https://www.atlasleefomgeving.nl/en/node/1591

15. Geopunt Vlaanderen. MIRA Road Traffic Noise Map, Status 2018 [Internet]. 2018. Available from: https://www.geopunt.be/catalogus/datasetfolder/687d2575-a53a-4530-94a5-be942ee01e00

16. Madrid City Council. Mapa estratégico de ruido [Internet]. 2016. Available from: https://transparencia.madrid.es/portales/transparencia/es/Transparencia-por-sectores/Medio- ambiente/Ruido/Mapa-estrategico-de- ruido/?vgnextfmt=default&vgnextoid=7151c15d15d32610VgnVCM2000001f4a900aRCRD&vgnext channel=27d9508929a56510VgnVCM1000008a4a900aRCR

17. Gencat. Mapas estratégicos de aglomeraciones [Internet]. 2017. Available from: http://mediambient.gencat.cat/es/05_ambits_dactuacio/atmosfera/contaminacio_acustica/gestio_ambi ental_del_soroll/mapes_de_soroll/mapes_estrategics_de_soroll/mapes_estrategics_d_aglomeracions/

18. Bruitparif. Cartes de Bruit [Internet]. 2017. Available from: https://www.bruitparif.fr/cartes-de-bruit/

19. Stockholms stad. Stockholms bullerkarta [Internet]. 2017. Available from: https://miljobarometern.stockholm.se/bostad-och-halsa/buller/stockholms-bullerkarta/

20. Houthuijs D, Fons J, Sáinz de la Maza M, Swart W, van Beek A. Method to estimate the road traffic noise exposure distribution in Europe [Internet]. 2018. Available from: https://www.eionet.europa.eu/etcs/etc-atni/products/etc-atni- reports/etcacm_tp_2017_10_traffic_noise_exposure_distribution_europe

21. European Commission. Global Human Settlement [Internet]. 2019. Available from: https://ghsl.jrc.ec.europa.eu/data.php

22. European Commission. European Settlement Map (ESM) 2015 - R2019 [Internet]. 2019. Available from: https://ghsl.jrc.ec.europa.eu/esm_R2019.php

23. Copernicus. European Settlement Map (ESM) 2012 - Release 2017 [Internet]. 2017. Available from: https://land.copernicus.eu/pan-european/GHSL/european-settlement-map/esm-2012-release-2017- urban-green?tab=mapview

24. EEA. Reportnet [Internet]. 2022. Available from: https://reportnet.europa.eu/

25. EEA. Air Quality e-Reporting (AQ e-Reporting) [Internet]. 2019. Available from: https://www.eea.europa.eu/data-and-maps/data/aqereporting-8#tab-figures-produced