GIS and GPS as useful tools to determine transportation noise levels D. Fabjan 1, D. Paliska 1 & S. Drobne 2 1 Faculty of Maritime Studies and Transportation, University of Ljubljana, Slovenia 2 Faculty of Civil and Geodetic Engineering, University of Ljubljana, Slovenia Abstract Road traffic is one of the biggest sources of environmental problems. Traffic needs and the motorization rate are increasing due to economic growth, which in turn tends to further generate rapid increase in traffic flow on the Slovenian road network. Thus an attempt to establish the influence of traffic flow on environmental degradation has been made. For this purpose road vehicle noise was analysed. After evaluating the mean speed of vehicles and the gradient of each road section, a model for computation of noise levels was applied. The Global Positioning System (GPS) device was used to determine the actual speed on chosen road sections which enabled the calculation of the influence of the slope and the number of crossroads on the average speed. The ArcView program was used for the Geographical Information System (GIS) approach to forecast noise levels, as well as to present the results of the analysis. The results confirmed a significant influence by traffic flow on noise. Keywords: road traffic, traffic noise, GPS, mean vehicle speed, road gradient, GIS, noise levels forecast. 1 Introduction Road traffic is one of the most widespread sources of noise. It affects practically all people, thus its negative effects are of great concern and a lot of attention is paid to its study. Every road vehicle emits noise some more, others less
34 Management Information Systems therefore a road with more intensive and more constant traffic flow is more aggravating to people and the enviroment than a road with less traffic. When assessing the intensity of traffic effects on noise, noise levels caused by vehicles circulating on roads have to be calculated considering different circumstances in which the noise is generated. The process of noise level determination is a very complex one as it takes into account as much conditions and assumptions as possible. The level of traffic noise is caused by factors such as number of vehicles per hour, mean vehicle speed, percentage of heavy vehicles, road surface characteristics, road gradient, distance of measuring point from the road edge, height of the measuring point, ground cover characteristics, view angle, reflection from the objects in the vicinity, etc. Most of the time it is not possible to consider every factor that influences the noise generation due to lack of information. In such cases the condition has to be simplified or even omitted. 1.1 Existing methods for estimation of noise levels The best way to determine noise levels is of course to go to the site and measure them or to install a permanent measuring device that measures the noise at all times. Both of these options are still quite expensive, and so most researchers base their research on measurements of representative samples, which then enable them to create models for approximate determination of noise levels considering the affecting factors as well. When deciding on the method to be used for measuring the noise levels caused by traffic, two strategies should be considered. The first one enables the determination of noise levels at points situated within a relatively regular grid of measuring points in the area studied. The second strategy offers the possibility of measuring noise levels by taking into account the previously classified studied area use, demographic density or the importance of the chosen streets and roads (Calixto et al. [1]). 1.2 Models FHWA and CORTN There are several models developed for calculation and evaluation of noise levels caused by road traffic. The most frequently used are the FHWA, developed in the United States of America and the CORTN, which was developed in Great Britain. Both models are quite precise compared those which consist of most factors that affect the intensity of noise, and at the same time they are quite simply designed. (Pamanikabud and Tansatcha [2]). The mathematical simulation model FHWA was developed for estimation, analysis and forecasting of noise levels generated on roads by a continuous traffic flow. The model divides the vehicles into three groups. For each group a separate calculation is required for basic noise levels, traffic flow correction, and shielding and barrier correction. Other adjustments such as distance correction and finite road adjustment are the same for all vehicle groups. The noise level in this model is expressed as L eq in decibels (db(a)) [2].
Management Information Systems 35 The model CORTN is also used for forecasting the noise levels in uninterrupted traffic flow conditions. The noise levels are expressed here in decibels (db(a)) as L 10 (noise level that is excessive in 10% of the time of measurement) for the loudest hour, and as L 10 for an 18 hour time interval. The basic noise level in this model is calculated considering only passenger cars, adjusted by speed correction, percentage of heavy vehicles, gradient correction, road surface with propagation correction [2]. Most models for estimation of noise levels use the value L 10 expressed in db(a), which represents an average value of the 18 hourly values L 10 for the period from 6 a.m. to midnight on working days. The Slovenian decree on noise owing to road and railway traffic defines a methodology to estimate sound pollution, which is based on the German standard RLS 90 and enables the evaluation of road vehicle noise emission percieved at the distance of 25 meters from the road axis from an approximate hight of 2.25 meters and by the mean speed of 100 kilometers per hour. In this paper a method to calculate and forecast the noise levels that would be perceived at the distance of 10 meters from the road edge and by the mean speed of 75 kilometers per hour will be used and simplified with the assumption that the ground is flat and that there are no barriers or other objects in the road vicinity. 2 Noise levels on road sections Data analysis and graphical presentation of the study results were effective, fast and simple due to the GIS approach using the ArcView program. This approach enables road traffic noise simulation and forecasting, thus makes the understanding of traffic effect on noise levels better. Geographical and attribute data from different databases were merged to make the analysis simplier and clearer, which increased its efficiency in changing and testing the parameters and conditions. The reference database in this analysis was the Slovenian road network database with several attribute data and defined road sections that represented the basic unit on which noise levels were estimated. The major road network is shown in figure 1. The data on road sections enabled also the merging of other databases such as the attribute data on traffic count on the Slovenian road network. Based on traffic count data it was possible to calculate the hourly traffic flow and the percentage of heavy vehicles (vehicles >3.000 kilograms) for each road section. 2.1 Calculation of the road section mean speed It has been proven that vehicle speed affects the noise level and that is why all noise level forecasting models contain the speed correction factor. The question arises how to determine the speed at one point or on one road section. Most of the models use the speed limit defined by regulations according to the road category. A lot of the times roads of any category are equipped with traffic signs
36 Management Information Systems that additionally limit the speed, or road characteristics and road conditions cause traffic flow to be slower or faster than the set speed limit. Figure 1: Slovenian major road network. In our study we developed a model to evaluate the mean speed within road sections using the GPS device. The device enabled us to gather a sample of data for geographical coordinates and speed at different points during each trip on chosen road sections. Trips on roads of different categories and in different traffic conditions were made in the coastal region of Slovenia. After collecting the GPS data an attempt was made to consider and estimate the influence that crossroads have on speed. Therefore the number of crossroads on each road section was defined by the query in the Slovenian road network database. Settlements were treated as crossroads. The coefficient of crossroads influence on the mean speed of road sections was calculated with the coefficient of determination 0.68 applying linear regression analysis by taking into account the regulated speed limit. The analysis showed that the coefficient (-1.4638) is statistically significant, which was confirmed by the global F test as well. The mean speed for each road section can be then defined by the following equation: v...mean speed C...number of crossroads on road section v l...speed limit for different road categories v = -1.4638 * C + v l (1)
Management Information Systems 37 The negative value of the regression coefficient shows that crossroads play the role of a barrier to traffic flow, thus causing a decrease in speed. The linear model introduced is simple and at the same time represents a good challenge to work on its improvement. The mean speed of course depends on several further factors that at this point were not considered due to lack of data. The estimated mean speed of each road section was then used in further calculations of noise level correction. 2.2 The estimation of noise levels The model CORTN was used to estimate noise levels (Cardiff University [3]) The formerly estimated mean speed and the road gradient were considered as well as the traffic flow structure and the immision point distance. The noise level was expressed as L 10 in one hour intervals to better understand the intensity of the traffic influence on noise. 2.2.1 Basic noise levels Basic noise levels for each road section were estimated using the equations provided by the CORTN model. Matematically the basic noise level can be defined as: L 10 = 42,2 + 10 * log 10 q (2) L 10...basic noise level for one hour intervals (db(a)) q...number of passenger cars in one hour intervals (vehicle/h) 2.2.2 Mean speed and heavy vehicles percentage corrections In reality, noise levels tend to be much higher than basic noise levels, which is due to several factors that cause such an increase. Thus the correction due to heavy vehicles percentage was considered additionally to the previously calculated mean speeds of road sections. The equation follows: L v, p = 33 * log 10 (v + 40 + 500/v) + 10 * log 10 (1 + 5p/v) 68,8 (3) L v, p...mean speed and heavy vehicles percentage correction v...mean speed of vehicles (km/h) p...percentage of heavy vehicles (%) number of heavy vehicles divided by number of passenger cars 2.2.3 Road gradient correction Road gradient also affects noise levels caused by road vehicles. Vehicles use more power to move uphill, creating more noise, and brakes used to move downhill are also an additional noise generator. Thus, evaluating the road gradient for each road section was essential to show the intensity of gradient influence on noise levels.
38 Management Information Systems A digital model of relief (DMR) had to be made to determine the road gradient of each road section. A DMR added a third dimension to a two dimensional map of the Slovenian road network. Based on the data given a DMR could be made applying the IDW (Inverse Distance Weighing) interpolation. The added third dimension enabled a calculation of average road gradients for all road sections. Further calculations were made under the assumption that the traffic moves uphill on each road section. This approach does not show the real condition of road slopes, but it is a good approximation to estimate the correction for the gradient factor on each road section. It was estimated that each additional percent of the gradient increases the correction of noise level by 0.3 db(a) [2]: L g...road gradient correction G...gradient (%) L g = 0.3 * G (4) 2.2.4 Immision point distance correction The lack of information on ground characteristics, road surface material, and other factors made it impossible to evaluate their corrections. However, the immision point distance correction was considered for the case of the immision point 2 meters high and 4 meters away from the road. The following equation derives from the CORTN model: L d = -10 * log 10 (d/13.5) (5) L d...immision point distance correction d...the shortest distance between the emission and immision points The shortest distance between the emission and immision point was calculated simply by the Pitagora rule: r...horizontal distance of immision point h...height of immision point d = ((r + 3.5) 2 + (h 0.5) 2 ) 1/2 (6) Using the above formula for immision point distance correction that rests on the assumption that the surrounding ground is hard, flat and without barriers, which is quite rare in reality, it is possible to come to 7.65 meters of total shortest distance between the emission and immision points. The immision point distance correction is then 2.47 db(a). This value was included into a final estimation of noise levels and was constant for all road sections.
Management Information Systems 39 2.3 Total noise levels on road sections The final estimation of noise levels for all road sections is actually a sum of all values explained previously. First the basic noise level is calculated to which we then add the speed correction, the heavy vehicle percentage correction, the road gradient correction, and the correction of the immission point distance for the hard and flat ground without barriers. There is a sum assigned to each road section. This sum represents the noise level (L 10 ) perceived by the 2 meters high and 4 meters distant immission point and exceeded by 10% of one hour traffic flow. These noise levels are the best indicators for traffic flow influence intensity on generating noise. Figure 2: Noise levels on road sections of Slovenian road network with a detailed map of a chosen geographical area. Figure 2 presents noise intensity on road sections of the Slovenian road network. Figure includes a detailed map showing the noise bands for a chosen geographical area in the north-east of Slovenia. Each road section is represented according to the noise intensity wide band using the average noise level L 10 value for a better understanding. 3 Conclusion The band width is subject to noise intensity generated on road section by vehicles. The pictures confirm the assumption that noise levels are higher the traffic flow is more intensive. Thus the more vehicles using the road section,
40 Management Information Systems the higher the noise levels. It is possible to notice though that some road sections with a relatively lower annual traffic flow are represented with a much wider band, which in turn means that the noise levels are higher compared to other road sections with similar annual traffic flow. This leads to the conclusion that the correction factors in such cases are of greater importance. Traffic flow speed plays an important role on some road sections. Compared to the road sections with a slower traffic flow, noise levels can be higher on road sections vehicles can reach higher speeds. Noise is also very dependent on the traffic flow structure. Following the rule that heavy vehicles generate more noise, noise levels on road sections with a high percentage of heavy vehicles tends to be higher. The same situation is found when considering the road gradient factor. In fact, road sections with relatively more slope add more to a road gradient correction factor than those with less or no slope at all. Providing that data on location of the surrounding barriers such as trees, buildings and other objects, and data on condition of the road surface were available, adjustments could be calculated and later on added to total noise levels, which would certainly increase even more. It is important to consider the barriers especially in the case of estimating noise levels in an urban area the sound pollution is the most damaging for people (Ramiset al. [4]). Noise generation depends also on the quality of road surface. Relevant data on road condition is quite difficult to be managed with geographical coordinates and kept up-to-date due to unequal use and material characteristics. Therefore in this case the approach described in this paper would not make sense. However, the introduction of the simple yet challenging linear model for the computation of mean speed of road sections is a step towards the implementation of valuable tools like GIS and GPS technology, and a motive for improvement. The possibility of simulation and more and more precise estimation of sound pollution caused on our roads by vehicles represents also a good tool to perceive the problem accurately, to make better decisions and to act against the negative effects of the intensity of traffic flow influence on our health and lives in general. References [1] Calixto, A., Diniz, F.b., Zannin, P.H.T., The statistical modelling of road traffic noise in an urban setting. Cities, Vol. 20, No. 1, pp. 23-29, 2003. [2] Pamanikabud, P., Tansatcha, M., Geographical information system for traffic noise analysis and forecasting with the appearance of barriers. Environmental Modelling & Software, Vol. 18, Issue 10, pp. 959-973, 2003. [3] Cardiff University. Square one research PTY Ltd. Web Site, www.squ1.com/flash-menu.html. [4] Ramis, J., Alba, J., Garcia, D., Hernandez, F., Noise effects of reducing traffic flow through a Spanish city. Applied acoustics, No. 64, pp. 343-364, 2003.