Home 
Awakening effects of church bell noise: Geographical extrapolation of the results of a polysomnographic field study ^{1} Correspondence Address: Based on a previously published exposureeffect model of Electroencephalography (EEG)awakening reactions (AWR) due to nightly church bell noise events, as well as on geocoded building and population data, we estimated the total number of the church bell noise induced awakenings on the population of the Canton of Zurich, in Switzerland. The calculated mean number of EEG awakenings per person in the studied region, triggered by church bell ringing, varied between 0 and about 5.5 per night. The results suggest that up to 120150 m distance from churches, on average more than one additional EEG awakening occurs per night per person. An estimated 2.53.5 percent of the population in the Canton of Zurich experiences at least one additional awakening per night due to church bell noise. To provide a simple decision support tool for authorities that consider limiting bell ringing in the night in some form, we simulated different scenarios to estimate the effects of different sound attenuation measures at the belfry as well as the effects of different lengths and positions of nocturnal bell ringing suspension periods. The number of awakenings could be reduced by more than 99 percent by, for example, suspending church bell ringing between midnight and 06 h in the morning. A reduction of the number of AWRs of about 75 percent could be achieved by reducing the soundpressure levels of bells by 5 dB.
Introduction Many Swiss churches and chapels ring their bells at night to indicate the time  usually every quarter of an hour. The nightly ringing of the bells not only occasionally surprises foreign visitors, but is suspected to cause sleep disturbances in the population. As a matter of fact, local and governmental authorities are increasingly confronted with residents complaining about the nightly ringing of church bells. Consequently, church bells were brought into the focus of noise regulation policy. As yet, however, no specific regulation for this type of noise exists in the Swiss noise abatement ordinance. [1] To date, little has been published about the effects of nocturnal nontraffic related ambient noise sources, such as noise from neighbors, noises from the restaurant and bars, noise from animals, or bell ringing from churches. [2] Church bell sounds are acoustically characterized by both high tonality and impulsivity; thus, they may easily trigger awakening reactions (AWR) of sleepers. To determine the exposureresponse relationship between church bell ringing and Electroencephalography AWRs (henceforth referred to as "AWR" or "awakenings") we carried out an observational field study, [3] where 27 volunteers living in the vicinity of nine different churches were measured in their home setting for four consecutive nights with ambulatory polysomnography and concurrent acoustic recordings in and outside their bedrooms. An eventrelated analysis of the recorded data showed that church bell noise increases the probability of awakening in a similar way to that reported for traffic noise events. While, the aforementioned field study aimed to elucidate the exposureawakening relationship on the level of individuals, the objective of the present study was to quantify the impact of the church bell ringing on AWR in a select population; here, the Swiss administrative Canton of Zurich (Area 1729 km 2 , Population today 1.4 million). [4] For this Canton, the number of awakenings elicited by nightly church bell ringing was estimated by extrapolating the exposureeffect model from, [3] to its whole area. Finally, we estimated the effects of adherence to certain limitations of soundpressure of bells and frequency of ringing on the sleep of the population. Methods In the following, we use the term additional awakening probability to indicate an awakening probability triggered solely by church bells (a detailed discussion of awakening probability calculation is provided in, [5] ). In principle, we modeled the number of additional AWR experienced in each residential building of the Canton of Zurich by estimating the distribution of maximum sound pressure levels of churches in the canton, and extrapolating the individual awakening probabilities on the whole population, taking account of the proportion of people asleep in each hour in the night. Furthermore, we assessed the severity of the induced noise effect by calculating the estimated numbers of AWR above the preset night noise protection criterion suggested by Basner et al. [6] for the Leipzig/Halle airport. A report in German of most of the calculations used for the present analysis can be found in. [7] Calculation perimeter, sources and receiver points Geographic coordinates of churches were extracted with Twixtelsoftware (of Twix AG), [8] and verified by using the vector 25 digital landscape model of the federal office of topography swisstopo. [9] The heights of the church bell towers were determined by calculating the difference between the digital surface model and the digital terrain model from swisstopo. [9] We assumed the positions of the bell cages by the x and ycoordinates of the churches and by the height of their bell towers. The vertical position of the bell cage was assumed to be at 75% of the height of the bell towers. The nightly bellringing schedules of the churches were obtained by systematic enquiry from the parishes. Residential building data were obtained from the register of houses and dwellings from the Swiss Federal Statistical Office and matched with the population census of the year 2000. [4] We assumed the relevant receiver points to be the positions of the bedrooms and estimated the locations of these based on the coordinates of the residential building midpoints and a mean height of 4.6 m above the ground. For each residential building, the nearest church with permanent nightly ringing was identified using ArcGis Desktop 10 from Esri. All calculations were limited to the area and population of the Canton of Zurich. Time period According to representative Swiss survey data, [10] 99% of the Swiss population take their sleep within the time period from 21 h to 9 h. Hence, we decided to basically restrict all calculations to this period. On the same note, the probability of awakening changes throughout the night. Awakenings are more likely to occur at the end of the night than at the beginning. [11] In the model, this was considered in the parameter NEpochs (average number of epochs that passed since sleep onset). The number of epochs passed after sleep onset at each hour h was determined from data about sleeping habits from the aforementioned survey, [10] [cf. [Table 1]]. We calculated the mean number of epochs passed since sleep onset for the middle of each hour h. We assumed a sleep latency of 18 min (derived from the agerelated Figure 2, p. 1270; [12] as age statement we used the average age of 41 years of the Swiss population in 2009, [13] ). Furthermore, we calculated, based on the same survey data, the percentage of the people asleep (Pasleep ) during each hour within the period between 21 h and 9 h [cf. [Table 1]]. Pasleep is further used as a weighting factor of the number of AWR within each hour in the night (cf. Eq. 1).{Table 1} Basic calculation procedure The number of additional AWR for a sleeper in each residential building (here, receiver points rp) within the canton was estimated for the ringing events of the bells from the nearest church. We calculated the number of additional AWR in the season s (summer or winter) within hour h at each rp with the parameter awakening probability (PAWR, s, rp, e) according to the logistic model from [Table 5] that we published in. [3] We assumed all churches ring their bells every quarter of an hour throughout the night, i.e. 4 times perh. Eq. 1 gives us the number of additional AWR for a particular season s within hour h at rp. [INLINE:1] By multiplying the number of additional AWR at an rp with the population count and cumulating over all rp in the Canton of Zurich one obtains the number of additional AWR for a particular season s within hour h for the whole population as shown in Eq. 2 below. [INLINE:2] Function and parameters used for calculating additional awakening probability Additional awakening probability due to church bell noise events (P AWR, add, s, rp, e) was calculated according to the awakening probability model derived from our field study [Table 2] in, [3] . The model comprised the following parameters: the maximum indoor soundpressure level (L AF, max, indoors ), the background level in the 60 s preceding the noise event (L Aeq, Background ), the interaction of L AF, max, indoors with L Aeq, Background , the number of 30sepochs that passed since the onset of sleep (N Epochs ) and the sleep stage (S2, SWS, or (REM) (rapid eye movement)) preceding the noise event.{Table 2} We assumed that the ringing schedules and the bell sound characteristics as well as propagation characteristics in the field study were similar to that of all churches in the Canton of Zurich. To simulate LAF, max, indoors in all calculations, we estimated the maximum soundpressure level in a reference distance of 1 meter from the bells, based on the distribution of the maximum outdoor soundpressure level values (LAF, max, outdoors ), and recorded in our field study cp. [Figure 2] in, [3] . To obtain a distribution of maximum sound pressure values at 1 m distance from the bells, a simple distance model considering only the geometric attenuation for point sources according to ISO 96132, [14] was applied to the empirical field study data. In this distance model, the contributions from reflections, obstacles, and attenuation to the sound propagation were not considered as it was assumed that they will be compensated in the later back calculation. Differences in geographic altitude were taken into account to calculate the distance d between the estimated positions of the bell cages and the positions of the sleepers. Hence, any measured outdoor maximum sound pressure level could thus be translated to a corresponding value at 1 m distance, according to Eq. 5. [INLINE:3] The reason for calculating two different seasons is that the observable proportion of open versus closed bedroom windows in the population follows a seasonal trend, and hence does the window attenuation that is in effect. The difference between the outdoor and indoor sound pressure level (Parameter A in Eq. 6) is dependent from the window position. For the situation "open or tilted window," we assumed A to be 15 dB and for the situation "closed window" 25 dB. In a survey about aircraft noise around the airport of Zurich, [15] it was found that during summer, 93% of the respondents keep their bedroom windows open or tilted, and 75% do so during the winter. Therefore, the estimations comprised for summer scenarios a distribution of 93% open or tilted (A = 15 dB) and 7% closed (A = 25 dB) windows, and for winter scenarios a distribution of 75% open or tilted (A = 15 dB) and 25% closed ( A = 25 dB) windows. To calculate P AWR, s, rp, e for each event e, the model comprised, besides LAF, max, indoors, s, the background level (L Aeq, Background ), the number of epochs that passed since sleep onset (N Epochs, h) and the binary dummy variables SWS stage in the preceding epoch (1 = S3 or S4, 0 = all other stages), and REM stage in the preceding epoch (1 = REM, 0 = all other stages). Background level (LAeq, Background ) was assumed to be 27.1 dB in all calculations. To account for the sleep stage effect (S2, SWS, or REM), a sleep stage was randomly drawn from the distributions of the sleep stages in each hour h as observed in. [3] This gives, for P AWR, add, s, rp, e: [INLINE:4] To increase repeatability in the calculation of the number of additional AWR, for each simulated ringing event, we drew 25 maximum sound pressure levels from the distribution in [Figure 1] and averaged the result. This procedure was used in all scenarios (season s, number of hours h with church bell ringing per night, soundpressure levels). Therefore, the estimated numbers of additional AWR are always average values. To determine the precision of our estimated number of additional AWR we calculated the 95% confidence interval (CI) was using the Student's tdistribution and the standard error of the mean considering the propagation of error for each scenario.{Figure 1} Application of a protection criterion In the wake of the discussions of a noise protection plan for Leipzig/Halle Airport in Germany before its expansion in 2007, the German Aerospace Centre Deutsches Zentrum für Luft und Raumfahrt (DLR) suggested a night noise protection policy [6] that defined a limit of no more than one additional AWR due to aircraft noise per night in order to prevent negative health consequences. This protection criterion corresponds to the highest risk category of the Night Noise Guidelines for Europe, [16] for the association between nocturnal noise exposure and health effects as put forward by Basner et al. [17] This risk category describes the health effects as follows: "The situation is considered increasingly dangerous for public health. Adverse health effects occur frequently, a sizeable proportion of the population is greatly annoyed and sleepdisturbed. There is evidence that the risk of cardiovascular disease increases." The DLR protection concept has been operative in Leipzig/Halle since 2004. [18] The people living in the area exceeding the awakening criterion (protection area) have received noise protection measures, e.g., soundproof windows. The concept is based, for preventive reasons, on rather sensible conditions (e.g. its formula always accounts for the time period 2206 h, sleep S2, 601 epochs passed since sleep onset and a malus of 1.4 dB for the L max from 02 h to 06 h). On the basis of Eq. 3 (with which the number of additional AWR for a particular time period at a rp was calculated), we determined the number of people experiencing more than one additional AWR per night due to church bells, but with slightly different conditions in our formula (time period 2109 h, sleep stage S2, REM or SWS, mean number of epochs passed since sleep onset for each hour and no malus for the L max for a certain time period). In the analogy of the aircraft noise protection measures around Leipzig/Halle airport, when exceeding this limit, this could mean that measures would be necessary to regulate nightly church bell ringing. Thus, we simulated the effect of feasible reduction measures (such as the suspension of ringing during certain hours of the night or reduction of the permitted LAF, max for church bell ringing) on the number of AWR elicited by church bells in the Canton of Zurich. Results Our enquiries yielded that there are 5,688 churches and chapels in Switzerland, of which 310 are situated in the Canton of Zurich. Of those, 55% (170) ring their bells every 15 min throughout the night, 7% (22) of them have reduced nightly bell ringing (for example the bells ring only every half hour or are quiet between midnight and 06 h in the morning) and 38% (118) do not ring their bells at all or at least not between 22 h and 06 h. Church bell ringing potentially affects a population of 1,247,906 residents (of the canton of Zurich) in a total of 196,299 residential buildings with known distance to the next church bell tower and known number of inhabitants. The results given in the following refer to only those 170 churches, which ring quarterhourly throughout the whole night. Number of additional AWRs in the population Church bell noise triggered, cumulated for all residents in the Canton of Zurich, in summer 286,520 (95% CI: 286,360, 286,926) additional AWR and in winter 239,021 (95% CI: 239,149, 239,310) additional AWR per night. The estimated cumulated numbers of additional AWR for each hour h of the night in summer and in winter NAWR, add, pop, s, h are shown in [Figure 2].{Figure 2} The estimated number of additional AWR individuals experience per night varies between 0 and 5.6 (summer and winter) and its distribution is shown in [Table 2]. The number of additional AWR is an estimated mean value per night for the respective resident. The majority of persons, about 812,000 (65% of the total population) in summer and 851,000 (70%) in winter are estimated to awake at maximum 0.2 times per night due to church bell ringing [first interval in [Table 2]]. In the following intervals, the number of people is rapidly declining. The sum of classes for example from [1; 1.2] to [5.8; 6] corresponds to the number of people experiencing more than one additional AWR per night. It is 44,055 (≈3.5%) (95% CI: 36,741; 52,454) people in summer and 31,142 (≈2.5%) (95% CI: 25,879; 37,659) people in winter. The corresponding values for people experiencing more than two additional AWR per night (sum of classes from [2; 2.2] to [5.8; 6]) are 6,364 (≈0.5%) (95% CI: 5,049; 7,972) people in summer and 3,697 (≈0.2%) (95% CI: 2,954; 4,670) in winter. These example values and further numbers of additional AWR per night estimated for the canton of Zurich are shown in [Figure 3].{Figure 3} The confidence limits in [Figure 3] and in the later following [Table 3],[Table 4] and [Table 5] are based on the 95% CI of the AWR add for each residential building of the canton of Zurich and a grouping criterion e.g. "AWR add > 1" (e.g. if the upper 95% CI value of the AWR add of one particular building is 1.07, the mean AWR add value is 0.99, and the lower 95% CI AWR add value is 0.91, then the upper 95% CI value fulfils the grouping criterion "AWR add > 1", whereas the mean value and the lower 95 CI do not). Since, the 95% CI of the AWR add of the individual residential buildings are considerably larger than the 95% CI of the mean AWR add of all residential buildings of the canton Zurich [cf. [Figure 2]], the calculated confidence limits of the cumulated number of people fulfilling the grouping criterion "AWR add > 1" are correspondingly large.{Table 3}{Table 4}{Table 5} [Figure 4] shows the scale of the expectable number of AWR due to church bell noise within different distance categories (concentric bands of 10 m width) from the nearest church. For each member of the population living in buildings within these distance categories, we estimated the number of additional AWR and then calculated the mean number of AWR per night within each distance category. At distances closer than about 140150 m in the summer scenario, and 110120 m in the winter scenario respectively, more than one additional AWR per night were calculated. At this point, it seems noteworthy to mention that the given numbers are averages per night, over longer time periods. Individual distributions of numbers of AWRs per night may range between no awakening at all, to several awakenings per night (corresponding distributions could, e.g. be derived with MonteCarlo simulations).{Figure 4} The validity of the estimation of the number of awakenings for people living up to about 20 m from the nearest church may be limited as the flat of the person living nearest from a church from our field study is situated at a distance of 23 m. Efficiency of the church bell noise reduction measures The DLR protection criterion defines a maximum limit of one additional AWR per night per resident due to (aircraft) noise. [6] As mentioned in their recommendation, this limit should not be exceeded in order to prevent negative health consequences. Since, the above extrapolation exercise revealed that a considerable number of residents in the Canton of Zurich are exceeding this limit, we simulated the effects of suspending nightly bell ringing for different time periods [Table 3] as well as the effect of reductions of the sound pressure level of ringing [Table 4] or a combination of both [Table 5] on the number of additional AWR. [Table 4] impressively demonstrates that the detrimental effects of the church bell ringing on awakenings could be reduced by lowering sound pressure levels, e.g. through dampening measures at the belfry or bell frame or by using the smaller bells etc., The effects of such measures were simulated by reducing the values of the L AF, max, 1m distribution by 1, 2.5, 5, 10, and 15 dB respectively, and again calculating the numbers of additional AWR per night. The outcomes of these scenarios are tabulated in [Table 4]. [Table 5] shows combinations of a bell sound attenuation by 1 dB respectively 2.5 dB and three different suspension periods of 2 h. Comparison with aircraftnoise induced awakenings To demonstrate the scale of the nightly effect of the church bell noise in the Canton of Zurich, we compared our populationbased extrapolation with estimations of awakenings caused by aircraft noise at the Leipzig/Halle airport, which is the thirdbiggest cargo airport in Germany, [18] and with cargo flight operations predominantly at night. At Leipzig/Halle, approximately 30,000 people are living within the "protection area" defined by the DLR night protection criterion. [19] Based on the assumptions set forth in the DLR protection concept (awakening from S2 sleep, 601 epochs after sleep onset), we reckon 83,617 people (95% CI: 70,528; 99,595) experience more than one additional AWR per night in the canton of Zurich. This number would correspond to the noise impact to nearly three airports the size of Leipzig/Halle. Discussion The present extrapolation estimates the mean number of awakenings due to church bell noise in the Canton of Zurich to be about 287,000 additional AWR per night in summer and about 239,000 additional AWR per night in winter. On the individual level, the number of additional AWR per night varies for the individual residents between 0 and about 5.5 [Table 2]. The (arbitrary) limit of one additional AWR per night referring to the DLR protection criterion of night noise, [6] is exceeded by approximately 3.5% (summer) and 2.5% (winter) of the population. 0.2% (winter) respectively 0.5% (summer) of people are even estimated to experience more than two additional awakenings per night. About 6570% of the 1,247,906 residents in the canton are estimated to awake at maximum 0.2 times per night (on average) from church bell ringing. Given the normally occurring number of spontaneous AWR in a night, which on average amounts in healthy sleepers to about 1530, the figures may appear low. Yet, a considerable number of people [ca. 44,000 in summer and 31,000 in winter, [Figure 3] have been calculated to experience more than one AWR per night on average, which can, according to previously published night noise protection recommendations, [17] be considered a potential threat to health Hence, a practical application of the night noise protection policy for church bell noise could mean measures to reduce the noise impact from churches through, e.g. suspension of ringing for certain hours, reducing the sound pressure level of bells (e.g. installing absorbers or encapsulation of the bell cage) or a combination of both. Regarding the suspending of bell ringing for certain hours, the most efficient time period would be between 03 h and 06 h cf. [Figure 2]. Simulations of suspending nightly bell ringing for a period of only 2 h from 04 to 06 showed a reduction of more than 50% [Table 3]. Suspension of four hours from 02 to 06 would reduce the number awakened more than once per night, to about a tenth of the original value. Suspension of the bell ringing for eight hours from 22 h to 06 h, which many countries define as the "legal night," would even reduce the number of people experiencing more than one additional awakening per night to virtually zero. Measures to reduce L AF, max of the church bells would also be quite effective [cf. [Table 4]], combinations of sound attenuation and suspension periods show promise as well [cf. [Table 5]]. The results of these simulations could therefore provide the foundations for the regulation of nightly church bell ringing in Switzerland, and elsewhere. Still, such as music, the sound of church bells may please or not. Therefore, protection measures could plausibly be required or equally be considered unnecessary. Bell ringing is an old tradition, which some people connect with and believe it has its rightful place, also in the night. Opposing this tradition stands the need for sleeping in quiet conditions. Remediation of the unwanted church bell noise effects is therefore rather complicated and potentially conflictual. As one important predictor for noiseinduced awakenings is the maximum sound pressure level of a ringing event, lowering the acoustic level of bells would be an efficient means to reduce awakenings while retaining the ringing tradition for the many that cling to it. Without measures taken, the number of the church bell induced awakenings will most probably increase in the future as in the wake of counteracting further urban sprawl, higher urban density is currently aimed for in spatial planning, especially in city centers, where also most churches are located. Limitations The presented extrapolation exercise was based on an exposureeffect model a distribution of sleep stages from one single observational field study. [3] This study was conducted with only 27 subject and was also rather explorative by nature. Since, variables such as noise sensitivity, age and gender are highly dependent on the individual, [20],[21],[22],[23] the predictions reported in the present paper are reliable only when the distribution of the individual sensitivities in the sample does not deviate too much from that in the population. As far as possible, we therefore, assessed the representatively of the sample of volunteers by means of a nonresponseanalysis whose results are reported in. [3] The nonresponse analysis revealed slight differences between responders and nonresponders insofar as the nonresponders showed, e.g., higher noise sensitivity, which could have resulted in an underestimation of the awakening effect in the population. There are a few other potential sources of error, which will briefly be discussed in the following: (1) Most importantly, the estimated number of awakenings per night could have been underor overestimated due to the tentativeness of the representativeness of the (very simple) sound propagation model that we used to derive the distribution of the L AF, max, 1 m [Figure 1] and for the calculation of exposure at the rp. The model is based only on the sound level produced by the nearest church. Yet, in some urban areas, several churches may be heard and not always the nearest church produces the highest sound level. (2) In the calculations of the estimated maximum soundpressure levels at the ear of a sleeper (cf. Eq. 6), an attenuation of 15 dB between outdoor and indoor soundpressure levels was assumed for an open or tilted window on the basis of figures published in. [24],[25] However, the 15 dB value might slightly underor overestimate the effective attenuation as we measured in our own field study an average attenuation of 16 dB  in contrast, e.g. [26] suggests only 12 dB. For closed windows, an attenuation of 25 dB was assumed. (3) A small number of churches with reduced nightly ringing schedules (with bell tolls less frequent than every 15 min between 21 h and 09 h) were not considered in the calculationsthis contributes to a slight underestimation of the effect. (4) The population data used in the present calculations rather led to a slight underestimation of the overall number of AWRs in absolute terms, because the population figures we used date back to the census of the year 2000, whereas the total population in the canton has since grown by about 10%. While, we gathered detailed operational information about the nightly ringing of church bells in the Canton of Zurich, it is not clear if the distribution of the ringing schedules is representative for the whole of Switzerland (or any other geographic entity). All in all, we believe thatwith regard to exploring the effects of measures to reduce awakenings caused by church bell sounds  the degree of reliability of the present extrapolation is acceptable. In absolute terms, the estimated numbers of AWRs in the study population may, for the reasons given, be subject to considerable uncertainties. Acknowledgments We are grateful to the Swiss Federal Office for the Environment (BAFU) as well as ETH Zürich for their financial support. Many thanks are due to all our volunteers whose participation in the field study set the basis for the extrapolation of awakening effects to a larger population. We would also such as to express our appreciation to Reto Pieren, Beat Schδffer, and JeanMarc Wunderli for their assistance and cooperation regarding acoustic measurements and calculations. Thanks are extended to Basner for information regarding the night noise protection concept at the Leipzig/Halle airport. Data to calculate the values in [Table 1] have gratefully been provided by the Swiss Foundation for Research in Social Sciences. Finally, we like to acknowledge the helpful comments and suggestions that we have received from one anonymous reviewer. References


