Studying the association between noise exposure, stress and character- istics of green spaces: protocol and pilot study Javier Dopico 1 Empa Swiss Federal Laboratories for Materials Science and Technology. Ueberlandstrasse 129, 8600, Dübendorf, Switzerland. Beat Schäffer Empa Swiss Federal Laboratories for Materials Science and Technology. Ueberlandstrasse 129, 8600, Dübendorf, Switzerland. María García Martín Swiss Federal Institute for Forest, Snow and Landscape Research. Zürcherstrasse 111, 8903, Birmensdorf, Switzerland. Natalia Kolecka Swiss Federal Institute for Forest, Snow and Landscape Research. Zürcherstrasse 111, 8903, Birmensdorf, Switzerland. Silvia Tobias Swiss Federal Institute for Forest, Snow and Landscape Research. Zürcherstrasse 111, 8903, Birmensdorf, Switzerland. Julia Schaupp Swiss Federal Institute for Forest, Snow and Landscape Research. Zürcherstrasse 111, 8903, Birmensdorf, Switzerland. Nicole Bauer Swiss Federal Institute for Forest, Snow and Landscape Research. Zürcherstrasse 111, 8903, Birmensdorf, Switzerland. Martin Röösli Swiss Tropical and Public Health Institute. Kreuzstrasse 2, 4123 Allschwil, Switzerland. Danielle Vienneau Swiss Tropical and Public Health Institute. Kreuzstrasse 2, 4123 Allschwil, Switzerland.

1 Javier.dopico@empa.ch

worm 2022

Mark Brink Federal Office for the Environmen. Worblentalstrasse 68, 3063, Ittigen, Switzerland. Jean-Marc Wunderli 2 Empa Swiss Federal Laboratories for Materials Science and Technology. Ueberlandstrasse 129, 8600, Dübendorf, Switzerland.

ABSTRACT Urban areas are constantly growing, and densification is a common strategy to limit settlement ex- pansion. However, this leads to loss of green spaces (GSs) and increasing noise pollution, which is detrimental to public health. Within the research project RESTORE (Restorative potential of green spaces in noise-polluted environments), an extended cross-sectional field study will be performed in the city of Zurich, Switzerland, to assess the relationships between noise annoyance and stress as well as their link to road traffic noise and GSs. In total, 5000 participants will be contacted in three waves, during which an online survey will be carried out followed by a visit to a subsample at home to collect hair cortisol probes. Participants of this study were selected according to the characteris- tics of their residences (stratified by accessibility to GSs and noise exposure). Further, traits of indi- viduals in addition to acoustic and non-acoustic attributes of GSs are accounted. Thus, the associa- tion of noise annoyance, perceived stress and physiological stress among residents exposed to differ- ent road traffic noise levels and with different grades of access to GSs are studied. In this contribu- tion, the study protocol and first results of a pilot study with 256 people will be shown. 1. INTRODUCTION

Urban areas experience a fast and continuous growth of population and mobility, which results in increasing noise exposure of the residents and a qualitative and quantitative decline of green spaces (GSs). Urban areas are characterized by permanent activities including road traffic that often remains at high intensities also during the night. In European urban areas of 2017, more than 125 million people were estimated to be exposed to potentially harmful environmental noise exceeding 55 dB L den [1]. Given that 68% of the world's population will live in cities by 2050 [2] a large number of persons will likely be exposed to increased noise levels and thus at risk to negative health effects. Negative health effects of noise on humans have been highlighted in a broad number of studies [3]. Among these, noise annoyance and sleep disturbance among the most widespread effects [4] but noise can also trigger physiological stress reactions [5] and lead to very severe effects such as mortality [6]. Such negative noise-induced effects may be alleviated by GSs. Literature suggests that environmental resources like the access to quiet spaces in the neighborhood and/or dwelling-related greenery can be beneficial. For example, Dzhambov, Markevych [7] and Schäffer, Brink [8] showed that noise an- noyance can be reduced by GSs. Therefore, GSs may become a crucial element to alleviate negative noise-related health effects and thus to help maintaining public health in urban areas in the future.

However, knowledge on the effect chain of noise exposure, noise annoyance, perceived stress, accompanying physiological stress levels and stress reduction in GSs is still scarce. To shed light on these questions, the research project RESTORE (Restorative potential of green spaces in noise-pol- luted environments) assesses the effects of GSs as facilitators and noise as impediment to recover

2 Jean-marc.wunderli@empa.ch

worm 2022

from stress. The current study, as part of the RESTORE project, focusses on noise annoyance and long-term stress as affected by road traffic noise exposure and green spaces in a large field survey in the city of Zurich, Switzerland.

2 METHODOLOGY

In this section, the study protocol (concept and design) is presented. The study is to be conducted in the city of Zurich, Switzerland.

2.1. Study concept In a cross-sectional study, the association of accessibility to public GSs (in terms of free access and proximity, within a certain distance from home) and the properties of these GSs on noise annoyance and stress (self-reported and physiological) shall be investigated. The focus is thereby on participants with high levels of road traffic noise exposure at home to elucidate the role of noise as an environ- mental stressor. GSs are classified by size and noise exposure. The design includes the following seven study groups according to the exposure conditions: four groups with high noise exposure at home with access to GSs (quiet and small, quiet and large, loud and small as well as loud and large GSs) one group with low noise exposure at home with access to GSs , one with low noise exposure at home without GS access and one with high noise exposure at home without GS access . Data collection will be performed in two waves. Per wave, a two-step approach will be followed. First, potential participants are contacted by letter, based on a stratified sample of the seven groups. They are invited to participate in an online field survey focused on noise annoyance, stress and leisure activities. Thereby, a participatory geographical information system (PGIS) is used to identify and characterize the participant's most visited GSs. Second, a subsample of participants will be visited at home to collect hair cortisol probes to measure long-term physiological stress, take photos from the partici- pant's living room's window and ask additional questions about the perceived soundscape and the view from home.

In the following sections, the implementation of this study concept is presented as planned for the extended study. Further, a pilot study to test the design is presented which, however, only accounts for two of the seven study groups.

2.2. Noise exposure For the noise exposure classification of the study sites, the date of the Swiss noise database son- BASE for the year 2015 was used [9]. Data was available for road traffic and railway noise in a grid with a spatial resolution of 10 m × 10 m and for aircraft noise in with a resolution of 150 m × 150 m. This dataset was used to assess the noise exposure for each housing location. As no L den values were available, the daytime road traffic level L day , i.e., the yearly averaged A-weighted equivalent contin- uous sound pressure level over 16 hours, from 6 am to 10 pm, was taken from sonBASE, and its mean was determined for a buffer of 50 m around the corresponding coordinates. However, since L den is one of the most widespread metrics to establish exposure-response relationships (e.g., [10], [8]), this will be presumably used in the large field survey study to quantify the noise exposure, given that updated data is available by them.

For the classification of the noise exposure at home the threshold for " low road traffic noise expo- sure at home " was set to L day ≤ 53 dB and the threshold of L day ≥ 68 dB was defined as " high noise

worm 2022

exposure at home ". The former threshold for road traffic noise L den corresponds to the recommenda- tions by WHO [3] and the latter is chosen with reference to Brink, Schäffer [10], who derived for an L den of 68 dB an average percentage 25% of highly annoyed for the Swiss population.

Further, with the goal of not mixing different noise sources as these may disparately impact health, and primarily focusing on road traffic noise, situations with railway noise of L day > 54 dB and aircraft noise L day > 45 dB are excluded from the study areas.

These thresholds (as the one for road traffic noise) were chosen according to the recommendations by the WHO [3]. While the latter are defined for the L den instead of L day , the two metrics do not substantially differ in situations with dominating daily traffic, as is the case for the city of Zurich. (Note: Further details on the conversion of different noise metrics are given in Brink, Schäffer [11]).

In addition, we characterized the road traffic noise exposure of the GSs . For this, L day was also used (and will be kept also in the main study). This metric is more appropriate than the L den for GSs, as visitors primarily spend time in public GSs during the day, and hence the noise exposure at night is unimportant. Defining a threshold to discriminate quiet and loud GSs is a challenging task, espe- cially in large GSs, where road traffic noise levels vary substantially within the area. Further, scien- tific evidence on potential thresholds is scarce (specifically in relation to health outcomes). Using a GIS-based analysis for the GSs of Zürich, we derived the following definitions for quiet and loud GSs.

1) If more than 50% of the area of a GS lay below 45 dBA L day , the GS is considered as quiet. 2) If more than 50% of the area of a GS lay above 58 dBA L day , the GS is considered as loud.

2.3. Study sites The study takes place in the municipality of Zurich, Switzerland. In the framework of this study, GSs encompasses traditional urban parks, but also other spaces such as community gardens or urban forestry in the vicinity of Zurich. Thus, all areas with vegetation in public, open and accessible spaces are considered as recreational green areas, referred to “green spaces (GSs)” . To identify the study sites, the main characteristics of interest, i.e., noise exposure and greenness, were quantified in an explorative spatial analysis. More specifically, the distribution of road traffic noise (again, L day ), res- idential green (satellite-based indicator of greenness Normalized Difference Vegetation Index, NDVI [12]) and public GSs as identified by land use classification data (see below) were spatially analyzed in Esri ArcGIS (version 10.8.1). Average values of these three characteristics were derived for dif- ferent buffer sizes around the buildings of the study groups (similar approach as in Schäffer, Brink [8] and Vienneau, de Hoogh [13]). Buildings which were in the vicinity of more than one GS within a 300 m circular buffer were excluded from the analysis to be able to assign participants to a single nearest GS and hence to facilitate the interpretation of results.

The GSs were identified and selected using land use classification data of the Federal Swiss Office of Topography (swisstopo) (for details see, Schäffer, Brink [8]). All GSs with restricted access (e.g., private/household gardens) or with access requiring payment, namely the zoo of Zurich, open air swimming pools, camping grounds, playgrounds of schools and sport fields (e.g., for soccer and golf) were excluded. Accordingly, accessibility was addressed in terms of free access and proximity, the latter by applying a circular Euclidean buffer with a radius of 300 m around each polygon that repre- sents a building. This allowed identifying areas without access to GSs (i.e., without GSs within 300 m) and with access to GSs from residents (with GSs within 300 m). Hence, a total number of 124 GSs in the city of Zürich were initially included in the dataset and divided into large (≥10'000 m 2 ) and small (<10'000 m 2 ) GSs. Of these 124, 23 GS were assigned to the loud group and 25 to the quiet

worm 2022

group according to the above definition. The remaining 76, which matched neither of the two defini- tions (45 dBA < L day < 58 dBA), were excluded from the study. With regard to greenness, NDVI was used to assess residential green ("How green are residential areas?"), as NDVI was found to be a crucial parameter for alleviating noise-induced health effects (see e.g. Schäffer, Brink [8]). No thresh- olds were for this variable. Also the NDVI data will be updated in the large field survey study.

2.4. Field survey Inhabitants of the study sites are considered as suited for participation if they lived for at least one year at the same location and were at least 18 years old. Participants will be reached by post with an invitation letter that contained both a link and a QR code to access the online questionnaire. As a motivation to participate, a lottery among all participants will be conducted. A reminder letter will be sent to all contacted persons that did not fill in the questionnaire within the first two weeks after having sent the invitation letters. The online questionnaire covers six different sections, namely per- sonal information (age, sex) and living situation, noise annoyance and sleep disturbance assessed through the 11-point numerical ICBEN scale [14], noise sensitivity (13-item NoiSeQ-R instrument, Griefahn, Marks [15]), personality and health, leisure time and employment situation, and occupation. The entire survey will be conducted in German. From the noise annoyance question, the binary vari- able Highly Annoyance (HA) is defined as 1 when one of the three top scale points on the 11-point scale is marked, i.e., 8, 9, or 10 (which corresponds to 28% of the scale length), and else as 0. Stress is addressed asking the ability to cope with stress (a 5-point numerical scale where 1 means "I cope badly with stress" and 5 "I cope well with stress"), as well as stress at home (a 5-point numerical scale where 1 means "no stress at all" and 5 "very high stress").

2.5 Pilot study A pilot study was conducted to test the study protocol and to get an estimate for possible response rates. In the pilot study only people from two study groups were included, namely (i) H igh noise exposure at home with N o access to GSs ( HN ) and (ii) L ow noise exposure at home with A ccess to GSs ( LA ). From the 256 persons contacted with an invitation letter, 26 participants completely filled out the online questionnaire. A subsample of 49 non-respondents were reached by phone in order to increase the sample size or find out the reason why they did not participate in the study. Unfortunately, none of these participants was willing or able to subsequently fill out the survey. A reminder letter was sent after six weeks to another subsample of 180 people that had neither filled in the online question- naire nor been contacted via telephone yet. 15 recipients of this reminder letter finally completed the survey, raising the number of participants to 41. This corresponds to an overall response rate of 16% in this pilot study.

3. RESULTS

Figure 1 shows the study area, i.e., the city of Zurich, with the two areas where participants were selected from (LA, HN). Table 1 gives a characterization of the noise exposure and greenness as- signed to the housing address of the participants, separated for the two study groups HN and LA.

Table 1: Average exposure to noise and greenness at home of the participants (150 m buffer) of the groups " Low noise exposure at home with Access to GSs (LA)" and "High noise exposure at home with No access to GSs (HN)".

worm 2022

Study group LA HN Mean road traffic L day (dBA) 40.9 56.2 Mean greenness (NDVI) 0.65 0.41

Of the total of 41 participants participating in this pilot study, 24 belong to the study group LA (58%) and 17 to HN (42%).

worm 2022

Figure 1: Study area with all GSs included in the study and the two study groups " High noise exposure at home with No access to GSs (HN)" and "Low noise exposure at home with access to GSs (LA)". Just approximate locations of participants shown for data privacy purposes.

Figure 2 shows noise annoyance as well as sleep disturbance for three transportation noise sources, namely road traffic (cars and trucks), railway and aircraft. The group HN is more annoyed by road traffic and railway than LA , while the opposite is observed for aircraft noise. The same observation was also made by [8]. In addition, reported sleep disturbance shows the same trends as reported noise annoyance.

Figure 2: Mean (dots) and with standard error bars of noise annoyance and sleep disturbance for namely road traffic (cars and trucks) (left), railway (middle) and aircraft noise (right), by study group, "High noise exposure at home with No access to GSs (HN)" and "Low noise exposure at home with access to GSs (LA)".

Fisher’s exact test was used to determine statistically significant relationships between the number of highly annoyed persons and the explanatory variables. Also, an F-test was performed for continu- ous explanatory variables and a Kruskal-Wallis test was used for non-parametric variables. As shown in Table 2, significant differences in the number of persons highly annoyed by road traffic noise were found for the study group HN and basic education (within highest education level category). Further, significant differences between highly annoyed and non-highly-annoyed persons were found for L day and NDVI (Table 3). No significant relationships were found for age, gender and stress (both general and private life).

Table 2: Differences between the number of non-highly (No) and highly (Yes) annoyed persons by road traffic noise, and p -values.

No Yes p

>65 9 (24.3) 1 (25.0) 18-40 14 (37.8) 2 (50.0) 1.000 41-65 14 (37.8) 1 (25.0)

Age range

Gender Female 17 (45.9) 2 (50.0) 1.000 Male 20 (54.1) 2 (50.0)

Study group HN 13 (35.1) 4 (100.0) 0.024 LA 24 (64.9) 0

Basic education 11 (32.4) 0 0.013 Primary school 1 (2.9) 1 (33.3) High school 0 1 (33.3) University 22 (64.7) 1 (33.3) Table 3: Differences in stress and exposure characteristics between non-highly (No) and highly (Yes) a nnoyed people by road traffic noise, and p -values.

Highest education level

No Yes p Ability to cope with stress Mean (SD) 3.0 (1.0) 3.8 (1.0) 0.181 Stress at home Mean (SD) 2.6 (1.1) 2.8 (2.1) 0.842 L day road traffic (dBA) Mean (SD) 46.2 (7.5) 57.5 (0.7) 0.005 NDVI (150 m around home) Mean (SD) 0.6 (0.1) 0.4 (0.0) 0.018 In conclusion, the results of this pilot study with statistical significances for study group, L day and NDVI, support the study design and show the feasibility of the proposed methodology.

4. CONCLUSION

In this paper, a study protocol for a cross-sectional study on noise annoyance and long-term stress as affected by road traffic noise exposure and green spaces in a large field survey, and a classification scheme for different residential areas regarding greenness and noise exposure is presented. The fea- sibility of the methodological approach was confirmed in a pilot study. The latter provides first in- formative insights about noise annoyance and sleep disturbance among people living under different exposure to noise and greenness. Noise annoyance and sleep disturbance is higher among residents living in areas with less greenness and increased road traffic noise exposure at home than those in rather quiet and green residential areas. The approach of this study may complement existing litera- ture on the topic in addition to providing relevant knowledge in the association of noise exposure,

worm 2022

stress and characteristics of GSs. Characterizing the spatial living environment among different real- world conditions as well as including personal traits could provide further relevant insights on this topic. Thus, the upcoming main study within RESTORE should be able to provide helpful infor- mation for urban planners in the future.

5. ACKNOWLEDGEMENTS

The authors are grateful to the participants of the field survey. Further, they would like to thank the city of Zurich for providing the addresses of the study participants to contact them. The study is funded by the Swiss National Science Foundation (project RESTORE, grant number CRSII5_193847 / 1).

6. REFERENCES

1. European Environmental Agency: Environmental noise in Europe. Denmark, 2020. 2. United Nations: World Urbanization Prospects: The 2018 Revision. United Nations, New York,

2019. 3. World Health Organization: Environmental noise guidelines for the European region, 2018. 4. World Health Organization: Burden of disease from environmental noise: Quantification of

healthy life years lost in Europe. World Health Organization. Regional Office for Europe, 2011. 5. T. Münzel, M. Sørensen, A. Daiber: Transportation noise pollution and cardiovascular disease .

Nature Reviews Cardiology 2021) 1-18. 6. D. Vienneau, A. Saucy, B. Schäffer, B. Flückiger, L. Tangermann, M. Stafoggia, J. M. Wunderli,

M. Röösli, S. s. group: Transportation noise exposure and cardiovascular mortality: 15-years of follow-up in a nationwide prospective cohort in Switzerland . Environment international 158 (2022) 106974. 7. A. M. Dzhambov, I. Markevych, P. Lercher: Greenspace seems protective of both high and low

blood pressure among residents of an Alpine valley . Environment international 121 (2018) 443- 452. 8. B. Schäffer, M. Brink, F. Schlatter, D. Vienneau, J. M. Wunderli: Residential green is associated

with reduced annoyance to road traffic and railway noise but increased annoyance to aircraft noise exposure . Environment international 143 (2020) 105885. 9. FOEN: Lärmbelastung durch Strassenverkehr in der Schweiz. Zweite nationale Lärmberechnung, Stand 2012 . 2015 [cited 2020; Available from: https://www.bafu.admin.ch/bafu/en/home/topics/noise/state/gis-laermdatenbank-sonbase.html . 10. M. Brink, B. Schäffer, D. Vienneau, M. Foraster, R. Pieren, I. C. Eze, C. Cajochen, N. Probst-

Hensch, M. Röösli, J.-M. Wunderli: A survey on exposure-response relationships for road, rail, and aircraft noise annoyance: Differences between continuous and intermittent noise . Environment international 125 (2019) 277-290. 11. M. Brink, B. Schäffer, R. Pieren, J. M. Wunderli: Conversion between noise exposure indicators

Leq24h, LDay, LEvening, LNight, Ldn and Lden: Principles and practical guidance . International Journal of Hygiene and Environmental Health 221 (2018) 54-63. 12. J. Weier, D. Herring: Measuring Vegetation (NDVI & EVI) . 2000 [cited 2021 27th May ];

Available from: http://earthobservatory.nasa.gov/Features/MeasuringVegetation/ 13. D. Vienneau, K. de Hoogh, D. Faeh, M. Kaufmann, J. M. Wunderli, M. Röösli, S. S. Group:

More than clean air and tranquillity: residential green is independently associated with decreasing mortality . Environment international 108 (2017) 176-184.

worm 2022

14. J. Fields, R. De Jong, T. Gjestland, I. Flindell, R. Job, S. Kurra, P. Lercher, M. Vallet, T. Yano,

R. T. A. R. UNIVERSITY: Standardized general-purpose noise reaction questions for community noise surveys: Research and a recommendation . Journal of sound and vibration 242 (2001) 641-679. 15. B. Griefahn, A. Marks, T. Gjestland, A. Preis: Annoyance and noise sensitivity in urban areas .

ICA: Madrid, Spain 2007).

worm 2022