1. Introduction
The adverse health effects of environmental noise have been extensively studied during the last two decades and evidence of these health effects has accumulated. Sufficient epidemiological studies have found relationships of environmental noise to cardiovascular diseases, annoyance, sleep disturbance, cognitive impairment, and tinnitus [
1]. As a result, the importance of environmental noise pollution in shaping urban, environmental, and public health policies throughout the EU and internationally is increasing, albeit at a relatively slow pace [
2].
In the EU, the Environmental Noise Directive (END) 2002/49/EC [
3] has been established for management of the environmental noise problem, and noise maps have been mandatory for all large agglomerations, major airports, roads, and railways. Noise mapping is simply a means of presenting calculated and/or measured noise levels in a representative manner over a particular geographic area [
4]. Since the establishment of the END in 2002, there has been a significant improvement in awareness among the general public and policymakers about the relationship between human exposure to environmental noise and related public health concerns [
2].
The END allowed each Member State to use its national method of noise assessment during the first phase of implementation, making it difficult to compare the results from the Member States [
4]. For improving the effectiveness of implementing the END, a common approach for assessing noise levels in Europe, namely CNOSSOS-EU, was developed by the European Commission in 2012 [
5]. This common assessment method was attached as an annex to the Commission Directive (EU) 2015/996 mandating all Member States to use it from 31 December 2018 onwards [
6].
In parallel to the European Commission efforts in environmental noise management, there is an increasing trend in the EU pushing the policy to move beyond the traditional noise control methods towards a more integrated approach of urban sound planning, and some large-scale projects are currently in place [
7].
In countries outside the EU, noise mapping has not yet been mandatory. However, some progress has been found in the use of noise maps for environmental noise management. For instance, production of noise maps of several cities all over the world have been mentioned in the literature, such as in the United States [
8,
9,
10], China [
11], Hong Kong [
12], Korea [
13], Brazil [
14,
15], Argentina [
16], and many others.
In KSA, environmental noise, particularly traffic noise, has been found a persistent environmental problem [
17,
18,
19]. This is mainly because KSA has experienced fast development with rapid growth of cities and urban extensions as well as high population growth, which accelerated the demands for increased transportation facilities [
18]. For instance, traffic volumes above 300,000 vehicles/day are common on the main roads of Jeddah City [
19].
Starting from 2012, the Presidency of Meteorology and Environment (PME) established new Saudi standards of environmental noise, of which community noise standard includes daytime, evening, and nighttime noise levels for three types of area according to area sensitivity [
20]. The standard does not mandate city noise mapping, but requires noise modeling and mapping for environmental impact assessment studies. Furthermore, the standard allows using noise prediction methods where direct measurement is not practicable. Nevertheless, the standard is intended to be revised every five years. With increased interest in city (or strategic) noise mapping in KSA, it is likely to be included in a revised version of the standards.
Currently, some research institutions in KSA have shown interest in environmental noise mapping. For instance, King Abdulaziz City for Science and Technology (which is responsible for establishing KSA national research priorities) has recently considered noise mapping of different cities in the Kingdom and noise monitoring and mitigation as part of its research-supporting priority [
21]. The growing interest in environmental noise mapping in KSA is shown in some published studies in Al-Dammam City [
17] and Riyadh City [
21].
In the Al-Dammam noise study [
17], direct noise measurement was performed in 122 locations adjacent to main roads and, subsequently, imported into noise mapping software to calculate noise contours. No information was mentioned about other types of data used in the software. On the other hand, the noise map produced for Riyadh City [
21] was generated by integrating the data of real equivalent noise (
LeqA) measurements for a period of 40 min (divided between daytime and nighttime) along with GPS data in 5284 locations. These data were imported to ArchGIS9 software (ESRI, Redland, CA, USA) to produce the noise map of the measured acoustic parameter.
It seems that the aforementioned trials of noise mapping in Saudi cities depended mainly on extensive real measurement of noise levels. Due to the complexity, and mainly laborious and expensive task of measuring the acoustic situation of a place, simulations may be used through prediction software informing noise levels in a faster way and less measurement data may be used for calibration of the prediction model [
14]. Additionally, noise maps produced from measurement are static, whereas those generated by prediction models are live, allow easy updates, and can separate contributions from all other component noise sources [
22].
It is worth mapping environmental noise in KSA metropolitan cities using environmental noise prediction models. These models can be used also in generating public noise annoyance data that are believed to be important in identifying the highly annoyed people who need priority action plans to protect their health. The use of noise mapping and annoyance models requires a large amount of data, e.g., digital maps with GIS data, terrain, buildings geometry, traffic data (road characteristics, traffic flow, speed, transportation type, etc.), population data, and others.
The objective of the current study is to test the adequacy of the available data in KSA municipalities for generating urban noise maps and using them for public annoyance analysis, and to verify the applicability of available environmental noise mapping models for KSA. This paper evaluates a model recommended by the END for traffic noise prediction using measurements and noise annoyance surveys in the Al-Fayha District.
3. Results and Discussion
3.1. Data Availability
Table 2 shows the availability of data necessary for noise mapping at the GIS Department of Jeddah Municipality. Except for population and traffic data, all other data needed for noise mapping were available in formats compatible with noise mapping software, such as GIS and CAD files. These file types can be easily imported into the noise mapping software and converted to the proper file type for further processing.
Fortunately, missing data such as number of inhabitant per dwelling are available in other governmental organizations, such as the Central Department of Statistics and Information (CDSI) and Jeddah Urban Observatory (JUO). Also, details of traffic flow in roads and streets can be obtained and regularly updated from Jeddah Traffic Department. In spite of being unavailable in GIS format, building heights (in the form of number of floors) are available from the Municipality as one of the required points of information for issuing building licenses. Furthermore, these data could be verified by using the databases of population statistics, where all demographic characteristics of buildings are collected during the statistical surveys. Through formal collaboration agreements, data exchange between GIS Department in Jeddah Municipality and other governmental organizations, such as CDSI and JUO, can be initiated for the benefit of public health.
3.2. Day, Evening, and Night Noise Levels
Generally,
Figure 5 shows that predicted noise levels were high as compared to the Saudi community noise guidelines. For instance, noise levels higher than 50, 45, and 40 dBA (the Saudi standards for LAeq,D and LAeq,E and LAeq,N, respectively in sensitive areas) were predicted near the premises of schools and mosques in the corresponding period of the day. Furthermore, noise levels in the two areas A1 and A2 mostly exceeded the standards shown in
Table 1 during the corresponding periods, except for scattered villas surrounded by wall fences and few multistory buildings that were far from main roads. In agreement with this,
Table 3 shows the average façade noise levels of the planning zones in Al-Fayha District, which were higher than the standard levels during daytime and nighttime.
Noise levels from
Figure 5 show that, despite the area A1 was a combination of sensitive and mixed area (mainly one or two-story buildings, schools,
etc.) with lower traffic flow, it was affected by noise levels almost equal to the corresponding levels in the area A2 (multistory buildings, commercial,
etc.). This might be a combined effect of the lower heights of area A1 buildings, fewer buildings, and more open areas, all resulting in limited barriers to noise propagation, contrary to the area A2 where the buildings might play a role in cutting off the road traffic noises [
13]. In addition to these, the very high traffic flow of the surrounding roads from all sides allowed for those high noise levels. The traffic data of the main roads and streets surrounding Al-Fayha District showed very high average daily flow, relative to the population density. These roads received vehicles from/to a large university campus, ring road, and the industrial zone.
Because of the hot climate of Jeddah City, the recreational, social, and leisure activities are normally practiced during the evening period. The traffic flow during this period is high and uniform throughout four hours of the evening. Although the average hourly traffic flow during the evening period is generally higher than that of the daytime period, the predicted LAeq,E values were almost equal or slightly lower than those of LAeq,D. This might be attributed to the sharp reduction of heavy truck flow during the evening period. Some types of these trucks are prohibited by traffic rules and others avoid movement during this peak period. In a recent study carried out by the author in Jeddah (data are not published), it was found that water tanker flow increases community noise levels by up to 1.6 dBA in some residential areas.
The high predicted noise levels during the three periods might be attributed to many reasons. Firstly, high small vehicle flow was observed at the main roads and side streets of Al-Fayha District during daytime, evening, and nighttime. The relatively high traffic flow at side streets resulted from a combination of factors, such as high average passenger vehicle ownership in Jeddah City (about 440 per 1000 persons, as calculated from the official statistics) and the unique pattern of vehicle use in such hot areas where using vehicles for transportation is preferred over walking, even for short distances. Secondly, it was estimated that about 13% of the population in Jeddah depended on water tankers for water supply due to inadequate supply from municipal water network. Furthermore, about 53% of the houses discharged their sewage in septic tanks rather than the municipal sewage network (which does not cover most of the districts). These two factors necessitated the use of water and sewage tanker trucks for the supply and discharge of water, loading the roads/streets excessively with these types of noisy heavy vehicles.
3.3. Public Annoyance Analysis
The high noise levels depicted from the produced noise maps necessitated the study of public annoyance due to traffic noise.
Table 3 shows that the average conflict levels (difference between average façade noise levels and the corresponding noise standards) were positive, being in the range of 4.6–8.7 dBA during daytime and 7.1–11.2 dBA during nighttime.
Figure 6 shows the details of average residential façade noise conflict levels during the three periods of the day for all buildings of the District. It is obvious that the conflict levels changed, generally, from day to evening then nighttime in ascending order. Furthermore, conflict levels as high as >15 dBA were found on many buildings, particularly those adjacent to main roads, indicating that people were exposed to severe noise levels. The high conflict levels and percentage of affected population during nighttime relative to other periods necessitates giving priority to the reduction of nighttime traffic noise [
13].
Since the planning zones of the area A1 were more sensitive than those in the area A2, their façade noise conflict levels were generally higher. For instance,
Table 3 shows that the average façade noise levels during day and night times at the pure residential zone 1 were 57.1 and 52.9 dBA, respectively, and the corresponding daytime and nighttime values at the general residential zone 3 were 61.7 and 57.3 dBA, respectively. However, the average façade conflict values during the same two periods, respectively, were 2.1 and 7.9 dBA at zone 1, and 1.7 and 7.3 dBA at zone 3. This could be explained by two factors. Firstly, the strict standard of the pure residential and educational planning zones included in A1 (zones 3, 5 & 6 in
Figure 2) compared to that of the general residential zone of A2 (zone 1 in
Figure 2) increased the conflict values at A1, knowing that both areas were adjacent to roads with high traffic density. Secondly, lower floors were exposed to noise levels higher than those at the upper floors. Therefore, the lower building heights at A1 (villas) compared to A2 (multistory) was another reason for higher average façade noise conflict levels. On the other hand, the average conflict levels at the mixed zones 2 (in A2) and 4 (in A1) were close to each other with a difference of 0.2 dBA.
Considering the distribution of population through the District zones,
Table 3 shows that 64.5% and 94.9% of the population were affected by noise levels higher than the standard levels of day and night periods, respectively. For instance, 46.7% of the population were exposed to average façade conflict level of 5.2 dBA during daytime whereas 69.4% were exposed to average façade conflict level of 8.9 dBA during nighttime. These numbers reflected the existence of a considerable percentage of highly annoyed persons among the whole populations of the district of interest.
Table 3 shows also that the total Noise Index was 145,663 and 326,497 (decibel.person) during the day and night periods, respectively. This index is important in valuing the cost of environmental or community noise in terms of its effect on the annoyed population as well as the cost of mitigation plans. Valuing noise impacts and control measures and balancing their costs is a key challenge in managing the problem from an economic perspective [
35].
The LKZ method provides a simple and explainable approach [
27], and its dependency on a given standard noise level makes it useful in long-term planning for community noise management. However, it fails to predict the highly annoyed people who should be given the priority in noise action plans. Instead, the %HA and %HSD metrics indicate the probabilities that certain percentages of the population, exposed to certain levels of road traffic noise, would be highly annoyed or have high levels of sleep disturbance, respectively, at a given spot [
30,
31,
36]. Therefore, they were used to predict the burden of disease from environmental noise [
1]. In light of this, the two metrics were calculated using the predicted average day and night façade noise levels at the five zones, as shown in
Table 3. About one-fifth of Al-Fayha district were estimated to be highly annoyed during daytime and 10.5% were estimated to have high level of sleep disturbance during nighttime. Zones 2 and 4, which are mixed residential/commercial and are adjacent to roads with high traffic flow, had the highest % of annoyance. However, zone 1 had the highest level of annoyance in terms of the number of affected people. For instance, the data of
Table 3 can be used to estimate that 52.2% and 54.8% of the total highly annoyed and highly sleep disturbed people of the district, respectively, belonged to zone 1.
As mentioned earlier, a traffic noise annoyance survey was conducted to test the validation of the predicted annoyance levels.
Table 4 presents a summary of the sampled population characteristics as well as the results of public annoyance. Only results from zones 1, 2, and 3 are shown as the responses from other zones were too limited to be included. Obviously, all traffic noise annoyance metrics were highest in zone 2, followed by zone 1, and finally zone 3. This is in agreement with the order of average façade levels of the same zones during day and night (
Table 3).
The predicted %HA and %HSD were compared to the survey %HA (as %highly annoyed day-night in
Table 4) and survey %HSD (as %highly sleep disturbed in
Table 4). The comparison is presented in
Table 5 which shows that there was a good agreement between the predicted and survey high annoyance metrics. The differences were statistically insignificant at the 0.05 level as shown by the Z-test for two independent proportions. The survey results of zone 2 were relatively higher than the predicted ones, which might be attributed to a relatively high percentage of those living in the most crowded eastern section of zone 2 in the sample. Their higher perception of traffic noise lifted up the %HA and %HSD. On the other hand, the lower perception of noise annoyance by zone 3 inhabitants as compared to the prediction could be due to better acoustical insulation of their houses. For instance, most zone 3 houses were provided with additional screens above the guard walls of their houses for more privacy, which resulted in noise reduction as well.
3.4. Comparison of Saudi and END Grid Calculation
The new Saudi noise standards recommend measuring community noise at a height between 1.2 and 1.5 m above ground [
20], whereas the recommended calculation height for noise mapping in END is 4 m above ground [
25]. In this study, the calculation of traffic noise was made at the two heights 1.5 and 4 m. The results from the two calculation heights were compared to each other in terms of affected inhabitants at various façade noise levels during daytime and nighttime.
Table 6 shows that the % inhabitants affected by façade noise levels higher than the standards were higher when using the END recommended height of 4 m, which means that the calculation height recommended by the Saudi standard was relatively less conservative. However, the differences in % inhabitants affected by exceeding levels between the two heights were slight with maximum values of 1.1% and 0.9% for daytime and nighttime periods, respectively. In spite of being slight, these differences could be reflected in the form of savings in the reduction of noise levels.
Annoyance analysis was made by interpolating grid calculation results in this study. It was found in another study that grid interpolation resulted in higher percentage of annoyed population as compared to direct façade calculation [
37]. Therefore, analysis of annoyance using grid calculations based on the height above ground recommended by the Saudi standard might give more accurate façade noise levels than that of the height recommended by the END.
3.5. Comparison of the Predicted and Measured Noise Levels
Table 7 shows the measured noise levels (locations 1–10) and façade conflict levels (locations 11–14) compared to the corresponding levels predicted by the model. The ranges of the difference between the measured and the predicted noise levels at the selected locations (1.5 m free-field levels) were from −3.3 to 0.2, from −2.7 to 0.5, and from −3.1 to 0.7 dB for day, evening, and nighttime, respectively. On the other hand, the difference ranges of façade conflict levels at the last four locations were −2.9 to 0.1, −1.9 to 2.1, and −4.5 to 2.6 dBA for day, evening, and nighttime, respectively. These deviations could be acceptable in comparison to deviations found in other noise mapping applications [
16].
Figure 7 shows the strong agreement between the predicted and the measured noise levels in this study. This good agreement was obtained despite the dynamic error was set to be 1.0 dBA in the prediction model and some settings were modified to accelerate the calculation on the debt of accuracy. It is, therefore, expected that if these settings were modified for maximum accuracy, the average error (variation) of the predicted noise levels from the measured values would, most likely, be lower. However, the calculation time would be much longer.
It should be kept in mind that there were many sources of uncertainty in noise mapping, such as model inputs, uncertainty propagation, model structure, and evaluation data [
25,
38]. Similar to the study of Ausejo
et al. [
16], the higher uncertainty source was due to road traffic flow (uncertainty group B: 0.5–1.0 dBA) and speed (uncertainty group C: 1–3 dBA) for both light and heavy vehicles (heavy vehicle flow <30% of the overall flow). This resulted in overall uncertainty of about 3.0 dBA in noise predictions of day, evening, and night noise levels. Traffic flow data in this study were collected in a way classified as moderate in terms of cost as proposed by WG-AEN [
25], resulting in uncertainty of about 3.0 dBA.
The same uncertainty sources apply also for façade noise level prediction. However, since the calculations of façade noise levels are based on 3D models, building geometry could be an important source of uncertainty. In façade noise level calculations, the majority of buildings were modeled as cubes by eliminating actual façade details for simplification of data processing and reducing calculation time. The effect of this can be illustrated using the data of façade conflict levels in
Table 7, which shows that the mean difference between the measured and predicted façade conflict levels at location 11 was high and negative in all cases (from −1.9 to −4.5 dBA) compared to locations 12, 13, and 13. The buildings in the later three locations were regular multistory houses (consisting of almost similar apartments), as shown in
Figure 8a. On the other hand, location 11 represented zone 3 which consisted, mostly, of detached houses (villas) having irregular structures as shown in
Figure 8b. Façade level real measurements showed that, besides the above-mentioned uncertainty sources, simplification of the complex geometry of zone 3 houses further increased the predicted levels above the real values.
Another potential source of uncertainty associated with the model simplification was the canyon effect resulting from simplification of a parcel of adjacent similar buildings into one building, as shown in
Figure 3. This affect the predicted traffic noise propagation by increasing the surface area of parallel building facades which cause multiple sound reflections as well as sound scattering within the urban canyon [
39,
40]. The result of this was, probably, increased noise levels.
In addition to the uncertainty associated with the calculation model, there were many uncertainty sources during measurement such as errors from measurement chain (e.g., sound level meter), source, meteorological conditions, measurement location, and residual sound [
32,
41,
42]. The combined uncertainties of measurement of the 14 locations were calculated and converted to the expanded uncertainty using a coverage factor of 2 to obtain a 95% confidence level [
16,
32,
41]. The expanded uncertainty of the 10 free-field 1.5 m noise level measurements ranged between 1.6 and 2.0 dBA for the three periods, as shown in
Table 7. On the other hand, the measurement expanded uncertainty of the four façade noise levels were higher, being in the range 1.8–2.5 dBA. This was a result of the added uncertainty of the microphone position in façade noise measurements.
4. Conclusions
In this study, traffic noise levels were predicted and public annoyance analysis was performed in Al-Fayha District, Jeddah City, Saudi Arabia using noise mapping approach. The predicted noise levels during the three time periods, i.e., daytime, evening and nighttime, were found higher than the maximum recommended levels established in the Saudi environmental noise standards. Annoyance analysis revealed that high percentages of the District inhabitants were highly annoyed, depending on the type of planning zone and period of interest. These results should be an alarm that the environmental noise, in particular traffic noise, might constitute a serious public health problem in the large cities of Saudi Arabia. For controlling this problem, environmental noise reduction plans should be in place in large cities. The first step in these plans is the assessment of the magnitude of the problem through noise mapping and public noise annoyance analysis. In this study, prediction of traffic noise using noise mapping in a Saudi environment was proven to be in agreement with the real field measurements. Furthermore, the results from annoyance analysis using the percentage of highly annoyed people approach, in particular, were very close to the results obtained from a field survey of traffic noise annoyance in the same District. This should encourage the application of noise prediction and mapping models in all large cities as a main element in environmental noise management plans to overcome the high cost of field measurement, which is the most commonly used approach in Saudi Arabia so far. However, noise mapping requires the availability of various types of datasets and supporting tools. The current study revealed that the infrastructure required for the application of noise mapping in large Saudi cities is almost provided in city municipalities, such as updated GIS data and maps, and proper noise map publishing platforms that are accessible to the public. For instance, Jeddah Municipality established an attractive and highly accessed (by the public) web application called Jeddah Geographical Explorer, which presents a lot of information related to the City in the form of layers on a map of Jeddah. Noise maps and annoyance data of Jeddah districts can be presented to the public through this application for raising their awareness of the problem. On the other hand, some other data, such as traffic and population data are available at governmental authorities other than the municipalities, and they can be easily accessible to the municipalities through collaboration within the context of a national environmental noise reduction plan.
The current study aimed at studying the potential for noise mapping and noise annoyance analysis in Saudi Arabia through application in one district in Jeddah City. However, further research is necessary to cover larger urban areas and the whole City as well. This will allow consideration of more types of input data, such as complex topography forms. In addition, further studies are recommended on the various options available to the decision makers of large cities in Saudi Arabia for management of the environmental noise problem. It is also recommended that public awareness and perception of environmental noise annoyance be studied in more details to include larger samples among Saudi residents.