CNOSSOS-EU Sensitivity to Meteorological and to Some Road Initial Value Changes

Transcription

1 CNOSSOS-EU Sensitivity to Meteorological and to Some Road Initial Value Changes Panu Maijala 1, Jarno Kokkonen 2, Olli Kontkanen 3 1 VTT Technical Research Centre of Finland, P.O. Box 1000, FI VTT, Finland. 2 3 Sito Oy, Department of Environmental Studies and Engineering, Espoo, Finland. ABSTRACT During the implementation process of the CNOSSOS-EU, many questions have been arisen. There are inconsistencies in the guidance, but the most of the questions are related to the acquisition of the initial values and their validity. We evaluated the sensitivities of the noise propagation part and the road noise sound power levels to the changes of some initial values within their validated ranges. The evaluated parameters were the probability of occurrence of downward-refraction conditions and the coefficients for the road surface and propulsion noise. Measured weather values were calculated using three approaches and compared to each other and to the default values. The results shows that the use of the default weather values increase the immission levels and thus the number of exposed people. The default rolling and propulsion noise factors of the CNOSSOS-EU method are not usable in northern conditions and the national factors should be exploited. Keywords: CNOSSOS-EU, environmental noise, mapping, rolling and propulsion noise coefficients, weather effects to noise, favourable conditions, initial values 1. INTRODUCTION In 2015, an update to the Environmental Noise Directive (END) [1] Annex II was published. According to the new Annex II [2], all the EU Member States (MS) are required to use Common NOise assessment methods in the EU (CNOSSOS-EU) from 31 December 2018 onwards. Some of the MS have already started the implementation of the CNOSSOS-EU framework and found that there are still many issues to be resolved in order to complete the process successfully. These issues include derivation of some initial data such as the probability of occurrence of favourable conditions from meteorological data and determination of the alpha and beta coefficients for road pavements. The European Commission Directorate-General for Environment is currently working on a corrigendum to address the most obvious inconsistencies, such as the conflicting frequency bands of interest, found in the Annex II [2]. In Finland, a project was started to prepare the implementation of the CNOSSOS-EU. The members of the project are authorities from the Ministry of the Environment, the Finnish transport agency, the Centres for Economic Development, Transport and the Environment (ELY Centres, responsible for the regional implementation and development tasks of the central government), a consultant company, and a research centre. The objectives of the Finnish project included preparation of national guidelines for the CNOSSOS-EU and evaluation of the method whether it is mature enough for adoption as a national method. Since the mid-1990s, Finland was part of the development team of the Nordic method, Nord2000 [3], representing the best knowledge of its time. Based on the work for the Nord2000, extensive initial value databases exist for modelling environmental noise. Because CNOSSOS-EU treats the initial values otherwise, new procedures to handle the national data had to be developed. A combination of a great number of initial values and a complex calculation method gives also reason to think about how much the whole process allows MS to use their own procedures to handle initial data without any significant change in the final result. 1 Panu.Maijala@vtt.fi 2 Jarno.Kokkonen@sito.fi 3 Olli.Kontkanen@sito.fi 1367

2 Meteorological data in the CNOSSOS-EU is taken into account by carrying out the calculation of sound levels in favourable conditions and in homogenous conditions. Long-term level L LT is calculated [4, p. 18] as a sum of levels in favourable conditions L F and homogenous conditions L H weighted by the probability of occurrence of favourable conditions (1): ( ) L LT = 10 lg p f 10 L F /10 + (1 p f )10 L H/10, (1) where p f is the probability of occurrence of downward-refraction conditions in the long term. The p f values are expressed in percentages for day (7:00-19:00), evening (19:00-22:00), and night (22:00-7:00) periods. Typically, the p f values are determined to 18 source-receiver directions (20 sectors). Currently, no guidance is given by the European Commission for determination of the p f values. At present, the only known pre-defined values of the probability of occurence exist for 41 locations across Metropolitan France [4]. In the CNOSSOS-EU requirements it is mentioned that variation in the input parameters of the emission part should have less than 2 db effect on the calculation results [5]. Using wrong sound power levels as initial data may lead to significant systematic errors in the calculation results and to avoid this, the essential values of the CNOSSOS-EU should be mitigated to the national conditions and values. When considering rolling and propulsion noise, a national surface correction may not be enough, rather than the correction for the A and B factors should be utilised also. In the Nordic countries the current road noise calculation method is Road Traffic Noise Nordic Prediction Method (RTN96) [6]. The RTN96 values are over 20 years old and too simple for the CNOSSOS-EU road model. Much more detailed and fresh measurements are made for the Nord2000 model [7] and those values can easily be transferred to CNOSSOS. In the CNOSSOS-EU framework, the vehicles are grouped into five separate categories: Category 1: Light motor vehicles (passenger cars, delivery vans 3.5 tons). Category 2: Medium heavy vehicles (medium heavy vehicles, delivery vans > 3.5 tons, buses, etc. with two axles and twin tyre mounting on rear axle). Category 3: Heavy vehicles (heavy duty vehicles, touring cars, buses, with three or more axles). Category 4: Powered two-wheelers (mopeds, motorcycles). Category 5: Open category (future needs). For total sound power level categories 1 3 are most important and same categories are also used in the Nord2000 model. In the Nordic prediction method (RTN96) there are only two categories: light and heavy. For the light, medium, and heavy motor vehicles (categories 1, 2, and 3), the total sound power corresponds to the energetic sum of the rolling and the propulsion noise. Total sound power level of the source lines m =1, 2, or 3 is defined in (2) [2, Eq ]: L W,i,m (v m ) = 10 lg(10 L W R,i,m(v m )/ L W P,i,m(v m )/10 ) (2) The rolling noise sound power level for each octave frequency band i for a vehicle of class m = 1, 2, or 3 is defined in (3) and (4) [2, Eq ]: L W R,i,m = A R,i,m + B R,i,m lg v m v ref + L W R,i,m (v m ) (3) L W R,i,m = L W R,road,i,m + L studdedtyres,i,m + L W R,acc,i,m + L W,temp (4) The Coefficients A R,i,m and B R,i,m are given in octave bands ( Hz) for each vehicle category and for a reference speed v ref = 70 km/h L W R,road,i,m is correction for the effect on rolling noise of a road surface with acoustic properties different from the virtual reference surface. This road surface correction factor is given by [2, Eq ]: L W R,road,i,m = α i,m + β m lg v m v ref (5) 1368

3 The propulsion noise emission includes engine, exhaust, gears, air intake, etc. The propulsion noise sound power level in the each octave frequency band i for a vehicle of class m is defined as [2, Eq ]: L W P,i,m = A P,i,m + B P,i,m (v m v ref ) v ref + L W P,i,m (6) The coefficients A P,i,m and B P,i,m are given in octave bands for each vehicle category and for a reference speed v ref = 70 km/h. The assessment of people exposed to noise is based on receivers defined in front of the building façades. Inhabitants of a building are linked to noise level of façade calculation points and the number of inhabitants is weighted by the length of the represented façade. The most exposed façade noise level is used only in buildings with a single dwelling per floor level. [2] [5]. In this paper, the sensitivity of the CNOSSOS-EU to some meteorological initial value changes was evaluated in a case study from Helsinki region. The default weather values of a commercial CNOSSOS-EU software implementation were compared to the measured values in three cases. The hypothesis was that using the default weather values the immission levels increase, and thus the number of exposed people. The CNOSSOS-EU road noise sound power levels were compared to the current Nordic model (RTN96) [6] and the latest Nord2000 model levels [7]. Correction suggestions for the CNOSSOS-EU road noise sound power coefficient are presented. During spring 2016 a new road noise emission report form Technical Research Institute of Sweden (SP) will be published. If the sound power level changes are notable, then the new values will be exploited. 2. METHODS 2.1 Evaluation of the meteorological data The p f values were determined with two modeling approaches from three different types of data sets. Statistical weights were calculated for favourable conditions by using local meteorological data. The statistical weights of favourable conditions describe the frequency of occurrence of downward-refraction conditions. The meteorological data was provided by the Finnish Meteorological Institute. The three obtained data sets were: a) Helsinki Airport and Helsinki Kaisaniemi (years ), b) Espoo Kivenlahti ( ), and c) Espoo Tapiola ( ). There were weather station data including air temperature at height of 2 m, wind direction and velocity at 10, 16 and 31 metre level, and cloudiness in both a) and c) data. The c) data included also computational meteorological data: Klug-Manier stability classes and both the temperature and wind speed gradient data. Because the wind velocity data from the 10 metre level was missing from Espoo Kivenlahti data and the profile-fitting trials didn t give any plausible results, we decided to discard that data. Wind roses were calculated to describe the meteorological conditions in the Helsinki region. Wind velocity gains the maximum values in the south-west direction (225, see Figures 1 and 2). This phenomenon is reflected also to the p f values, having higher values between the south and west ( ). (a) Height 16 m. (b) Height 10 m, z 0 =0.5. (c) Height 10 m, z 0 =0.1. Figure 1. Wind roses for the data from the meteorological tower in Helsinki Airport. Calm winds 1.45%. All the meteorological data included also date, time at one hour interval, and geographical position of the weather station. All the data was validated, and only the valid data was used for deriving the occurrence 1369

4 (a) Height 31 m. (b) Height 10 m, z 0 =0.5. (c) Height 10 m, z 0 =0.1. Figure 2. Wind roses for the data from the meteorological tower in Helsinki Kaisaniemi. Calm winds 0.32%. values. Weather stations have a very good coverage through Finland; hence the weather station data should be preferred in deriving the local p f values for the use of national noise calculations required by END. Modeling approach 1: Stability classes The first method to calculate p f values uses weather station data. The required values are: temperature, wind direction and velocity, cloudiness, and declination of the sun (derived from time and date). Method is described in [8][9][10]. The meteorological conditions are divided into stability classes and furthermore in 25 classes with different sound speed profiles. The 25 classes was reduced to favourable conditions and to homogenous conditions. This simplification method was derived from the method described by Plovsing in [11, p. 12], the number of meteo-classes was reduced to four classes: unfavourable, neutral, favourable, and very favourable. A Matlab program developed by Delta was modified and used to calculate the p f values. Modeling approach 2: Temperature gradient and wind velocity gradient The second method is based on the temperature gradient and the wind velocity gradient. Both the wind and temperature data from two heights is needed, preferably from 2 metres and 10 metres levels. The gradients were analysed and p f values were derived with method described in the reference literature [4, p. 18]. A Matlab program was used to calculate p f values. Noise calculations and the number of people exposed to noise The effect of p f values were evaluated along with the number of people exposed to noise and also by looking after differences in noise levels in grid calculations. The noise calculations were done in the test area in Helsinki region on an urban area of 34 km 2, where lives residents, from which 50% live in small residential buildings or row houses, 49% in high-rise buildings, and 1% in non-residential buildings. Road traffic noise was calculated using a preliminary implementation of the CNOSSOS-EU model in commercial noise calculation software. L de and L n, the day and evening noise level and the night noise level indicators, were used with a calculation height of 2 metres above the ground. Two different approaches were used for the assessment of people exposed to noise: 1) all inhabitants of a building are linked to noise level of the most exposed façade 2) residents are distributed to all calculation points on the façades and are weighted by the length of the represented façade. 2.2 Evaluation of the road data The CNOSSOS-EU sound power levels were calculated along the guidance given in the new Annex II and using the basic values (e.g. without acceleration, without studded tyres, etc.) [2]. The basic values of the current Nord2000 model [7] are given in third octave bands and for the sound power level calculation the values were converted to octave bands. The calculated A-factors in octave bands are the energy sum of corresponding third octave bands and the calculated B-factors are the average from corresponding third octave 1370

5 bands. The RTN96 light and heavy vehicles sound power levels were calculated from the sound exposure levels L AE,10 m using (7) where r = 10 m and v m is expressed in m/s. L W A = L AE + 10 lg(2 r v m ), (7) The correction for the reference surface condition was calculated according the Nord2000 and the Harmonoise surface correction [7, Eq. 2.3] [12, Table 6.9]: L Road = RS (CS 11) db, (8) where RS= +0.3 db for the SMA surface and CS is the maximum chip size in mm (8 16 mm). 3. RESULTS 3.1 Case studies weather in Helsinki region The calculated p f values for the meteorological data sets described in Section 2.1 are presented in the Tables 1, 2, and 3. The derived p f values differ significantly from the default values: day 50%, evening 75%, and night 100%. Table 1. Values of p f, Helsinki Kaisaniemi weather station, method Direction, p f,day, % p f,evening, % p f,night, % Direction, p f,day, % p f,evening, % p f,night, % Table 2. Values of p f, Helsinki Airport weather station, method Direction, p f,day, % p f,evening, % p f,night, % Direction, p f,day, % p f,evening, % p f,night, % Table 3. Values of p f, Espoo weather model, NMPB2008 method. Direction, p f,day, % p f,evening, % p f,night, % Direction, p f,day, % p f,evening, % p f,night, % At largest, the differences in the calculated p f values are 27, 48, and 67 percentages smaller than the default values at day, evening, and night periods, respectively. When the Espoo weather model based, more 1371

6 accurate p f values, are compared to p f values based on weather station data (see Tables 4 and 5) it seems that Helsinki Airport p f values are closer to the accurate data than Helsinki Kaisaniemi p f values. Kaisaniemi p f values were also calculated with the wind speed that was converted to the height of 10 metres (original data was from 31 metres). Conversion is based on a constant value of the roughness length z 0. The conversion was made with three z 0 values (0.025, 0.1, and 0.5). [10, Eq. 11] The differences in p f values are about 0%, when z 0 = 0.025, between 3% and +1% when z 0 = 0.1, and between 14% and +6% when z 0 = 0.5. There is a comment in the Delta s Matlab script, which says that the calculation method is valid for a roughness length z 0 = m. Table 4. Difference between the p f values in Kaisaniemi and Espoo Direction, p f,day, % p f,evening, % p f,night, % Direction, p f,day, % p f,evening, % p f,night, % Table 5. Difference between the p f values in Airport and Espoo Direction, p f,day, % p f,evening, % p f,night, % Direction, p f,day, % p f,evening, % p f,night, % To evaluate the sensitivity of the CNOSSOS-EU sound propagation model to the changes of meteorological conditions, noise calculations were performed in the Helsinki region on an urban area. The evaluated parameters were the values of probability of occurrence of downward-refraction conditions. Default p f values, day 50%, evening 75%, and night 100%, were used towards each direction, and compared to the values derived from the measured data in three cases. The effect of p f values was evaluated by the number of people exposed to noise. The values of the number of people exposed to noise are shown in the Tables 6, 7, 8, and 9. The results show, that using the default weather values for the day evening time, the number of exposed people becomes from 10% to 20% higher than compared to the to the p f values derived from the measured data. At the night time the number of exposed people becomes 60% 90% higher than compared to p f values derived from the measured data. Thus, the use of default values increase the immission levels, and, the number of exposed people. There were no significant differences between number of people exposed when the measurementsbased p f values were used. In Finland, the exposure assessment method has been used where inhabitants are linked to the most exposed façade. The new CNOSSOS-EU guideline says that the inhabitants have to be allocated to all façade calculation points [2, Sec. 2.8.]. The definition of the exposed people has a significant effect on the total number of the people exposed to noise. The effect of different exposure assessment methods results in difference of 70% in the number of people exposed to noise. In a recent master s thesis similar results were obtained with the percentages between 50% and 70% in the number of people exposed to noise [13]. In the Figures 3 and 4 the differences between the results of the grid calculations with two different p f values (Airport minus Default) is visualised. The results of the grid calculations show, that at a close distance (between metres), there is only a db difference between the default and measurements-based weather values. The difference becomes db at longer (over 100 metres) distances at the day time (see Figure 3). At the night time, the difference between the calculated and the default p f values varies between 4 and 2 db at distances over 100 metres (see Figure 4). In this Helsinki region case the p f values 1372

7 have effect only on the main roads, where over 55 db noise levels are reached up to distances of metres (see Figure 5). At low traffic count streets, with narrow noise zones, there is no effect in the number of people exposed to noise. Table 6. The number of people exposed to noise L Aeq,day-evening, method: the most exposed façade Weather data, Calculation method db db db 65 db sum of over 55 db Kaisaniemi weather station, Airport weather station, Espoo weather model, NMPB2008 Default values, (day 50%, evening 75%) Table 7. The number of people exposed to noise L Aeq,night, method: the most exposed façade Weather data, Calculation method db db db 65 db sum of over 50 db Kaisaniemi weather station, Airport weather station, Espoo weather model, NMPB2008 Default values, (night 100%) Table 8. The number of people exposed to noise L Aeq,day-evening, method: all façade calculation points Weather data, Calculation method db db db 65 db sum of over 55 db Kaisaniemi weather station, Airport weather station, Espoo weather model, NMPB2008 Default values, (night 100%) Table 9. The number of people exposed to noise L Aeq,night, method: all façade calculation points Weather data, Calculation method db db db 65 db sum of over 50 db Kaisaniemi weather station, Airport weather station, Espoo weather model, NMPB2008 Default values, (night 100%)

8 INTER-NOISE 2016 Figure 3: Difference of noise levels between two different calculations and pf values: Airport default. Day time. Figure 4: Difference of noise levels between two different calculations and pf values: Airport default. Night time. 1374

9 Figure 5: Buildings that are exposed to over 55 db noise when default p f values are used. Color coded difference of noise level at most exposed facade with calculated and default p f values. Red buildings: difference below 1 db. Blue buildings: difference over 1 db. Figure 6. CNOSSOS, Nord2000, and RTN96 light vehicle L WA. Table 10. Rolling noise, Nord2000 light vehicle (cat1) CNOSSOS-EU light vehicle (cat1) Speed, km/h Frequency, Hz L W A

10 3.2 The Nordic road parameters One of the main questions was, if it is enough to use a national surface correction L W R,road,i,m (5), or should there also be national coefficients for the A and B factors, (3) and (6), for the rolling and propulsion noise parts? Nordic road surfaces are much rougher compared to the middle Europe, because of the winter conditions and the use of studded tires. In the Nordic models, the rolling noise increase in a steeper angle as a function of speed, and the overall level is much higher than in the CNOSSOS-EU (see the Table 10 and the Figure 6). In the Table 10 it is shown that the rolling noise speed coefficient β (5) differences depend on the frequency band. The propulsion noise of light vehicles is practically meaningless at speeds over 50 km/h. At 30 km/h it is at about the same level with the rolling noise. So, for the light vehicles, it would be accurate enough to add a national surface correction, that also could adjust the A R,i,m and B R,i,m factors. The surface correction speed-dependent β factor is the same for all frequency bands, so there will be some error if A R,i,m and B R,i,m factors are corrected with the surface correction. This error can be minimized, if the β factor is averaged with the most important frequency range (500 Hz 2 khz) and the surface correction β factor is defined. The calculated surface correction, correcting also the A R,i,m and B R,i,m factors, is presented in the Table 11. Below the correction, a minor error is shown. Table 11. Surface correction for Nordic reference surface and error after surface correction α m α m α m α m α m α m α m α m β 63 Hz 125 Hz 250 Hz 500 Hz 1 khz 2 khz 4 khz 8 khz SMA 0/ Speed, km/h Frequency, Hz Figure 7. CNOSSOS, Nord2000, and RTN96 heavy vehicles L W A. The propulsion noise is important for the whole speed range with the vehicle categories 2 and 3. So, if the national values vary remarkable, then it is practically mandatory to use the national A P,i,m and B P,i,m coefficient for the propulsion noise. The sound power levels of the Nordic and the CNOSSOS-EU heavy vehicles are shown in the Figure 7. The category 2 in the Figure 7 and the Table 12 show that at Nordic countries the heavy vehicles have much higher propulsion sound power levels. Also, the higher rolling noise emission has a significant effect on the total sound power level (2) at higher speed levels. The reference road surface is not very common in the Nordic countries, so that s why 1.55 db higher levels are presented with the SMA 0/16 surface (8). 1376

11 Table 12. Propulsion noise medium heavy vehicle (cat2) Nord CNOSSOS Speed, km/h Frequency, Hz L W A SUMMARY The implementation process of the CNOSSOS-EU framework requires a lot of work at the national level. The guidance given in the new Annex II of the Environmental Noise Directive contains a number of deficiencies, and it is not fully consistent. Some of the Member States, including Finland, have already started the implementation process. In Finland and the other Nordic countries the some existing national data differs significantly from the CNOSSOS- EU default values and one of the first major tasks will be acquisition of new, reorganising some old, and developing the routines to convert the data to meet the needs of CNOSSOS-EU initial data. These routines include the meteorological and road data, for which CNOSSOS-EU sensitivity was evaluated. It was shown in this paper, that giving the Member States a freedom to develop their own methods to acquire and derive the initial values, a significant risk of variation to the final results will emerge. 4.1 Conclusions The main findings of this study are the following: The use of the default weather values increase the immission levels and thus the number of exposed people. The exposure assessment method has a significant effect on the total number of the people exposed to noise. The p f values derived from a simple weather station data represent well the more accurate p f values based on weather model data. It is possible to achieve adequate occurrence values with simple meteorological data. Only minor deviations to p f values were found, when considering the roughness length. Roughness length can be utilised in generation of the p f values. Local A and B coefficients has to be used for the propulsion and rolling noise in the Nordic countries. Existing Nord2000 data is usable for this purpose. In this article, we presented the effects of two methods on exposure assessment: 1) inhabitants linked to the most exposed façade method, and 2) inhabitants allocated to all façade calculation points method. The effect of different exposure assessment methods results in difference of 70% in the number of people exposed to noise. Meteorological data from local weather stations should be used to derive the local p f values for the use of national noise calculations. It is not recommended to use default weather values in Finland. As a limitation, p f values are valid only for certain environments that meet the assumptions required by the micrometeorological models used to establish these values. For example, these weather values cannot be used for the street canyons in an urban environment. For such an urban environment it is recommended to use the maximum of p f values for each period and direction. In the Nordic countries, the national road noise emission data significantly differs from the CNOSSOS-EU model default values, and the national factors for the rolling and the propulsion noise sound power coefficients 1377

12 A and B have to be used. The Nordic A and B coefficients for the rolling and propulsion noise are based on the measurements in Sweden and Finland, and are already implemented in the Nord2000. Also, the Nord2000 road surface correction can be converted to CNOSSSOS-EU. If the commercial software implementation of the CNOSSOS-EU doesn t allow to adding the coefficients for the basic sound power levels, then one option is to add the rolling noise coefficients to the national surface correction. The drawback is that the error of the sound power levels of heavy vehicles won t be fixed properly. REFERENCES [1] Directive, EN. Directive 2002/49/EC of the European Parliament and of the Council of 25 June 2002 Relating to the Assessment and Management of Environmental Noise, June [2] Directive, EN. Commission Directive (EU) 2015/996 of 19 May 2015 Establishing Common Noise Assessment Methods According to Directive 2002/49/EC of the European Parliament and of the Council, May [3] Jørgen Kragh, Birger Plovsing, S. Storeheier, and Hans Jonasson. Nordic Environmental Noise Prediction Methods, Nord2000. Technical Report AV1719/01, Delta Acoustics & Electronics, March [4] François Abbaléa, Savine Andry, Marine Baulac, Michel C. Bérengier, Bernard Bonhomme, Jérôme Defrance, Jean Pierre Deparis, Guillaume Dutilleux, David Ecotière, Benoît Gauvreau, Vincent Guizard, Fabrice Junker, Hubert Lefèvre, Vincent Steimer, Dirk van Maercke, and Vadim Zouboff. Road Noise Prediction, 2 Noise Propagation Computation Method Including Meteorological Effects (NMPB 2008). Technical Report EQ-SETRA 09-ED32 FR+ENG, Sétra report, June [5] Stylianos Kephalopoulos, Marco Paviotti, and Fabienne Anfosso-Lédée. Common Noise Assessment Methods in Europe (CNOSSOS-EU). Technical Report EUR EN, Institute for Health and Consumer Protection, aug [6] Jørgen Kragh, Hans G. Jonasson, Ulf Sandberg, Svein Storheier, and Juhani Parmanen. Road Traffic Noise: Nordic Prediction Method. Number 525. TemaNord, April [7] Hans G. Jonasson. Acoustic Source Modelling of Nordic Road Vehicles. Technical Report 12, SP Swedish National Testing and Research Institute, [8] Renez Nota, Robert Barelds, and Dirk van Maercke. Engineering Method for Road Traffic and Railway Noise after Validation and Fine-tuning. Technical Report HAR32TR DGMR20, January [9] Raimo Eurasto. NORD2000 for Road Traffic Noise Prediction, Weather Classes and Statistics. Technical Report VTT-R , VTT Technical Research Centre of Finland, April [10] Raimo Eurasto. Sääolot ympäristömelun laskentamalleissa [ Weather Conditions in Environmental Noise Prediction Models) ]. Technical Report 655, Ympäristöministeriö, November [11] Birger Plovsing. Noise Mapping by Use of Nord2000, Reduction of Number of Meteo-classes from Nine to Four. Technical Report 18, DELTA Danish Electronics Light & Acoustics, [12] Hans G. Jonasson, Ulf Sandberg, Gijsjan van Blokland, Jurek Ejsmont, Grek Watts, and Marcello Luminari. Source Modelling of Road Vehicles. Technical Report HAR11TR SP10, December [13] Olli Kontkanen. Ympäristömelulle altistuvien ihmisten määrän arviointitarkkuuden parantaminen [ Improving the accuracy of environmental noise exposure assessment methods ]. Master s thesis, Aalto University School of Electrical Engineering, Department of Signal processing and Acoustics,