The 32nd International Congress and Exposition on Noise Control Engineering Jeju International Convention Center, Seogwipo, Korea, August 25-28, 2003 [N492] Using GIS in Noise exposure analysis Author: Hardy Stapelfeldt Stapelfeldt Ingenieurgesellschaft mbh, Wilhelm-Band-Str. 7 D 44141 Dortmund, Germany Email address: info@stapelfeldt.de Second Author: Andrew Jellyman Birmingham City Council Birmingham,UK ABSTRACT In 1998 Birmingham City Council (BCC) in the United Kingdom commissioned a pilot study that lead to a city wide Noise Map taking into respects the road and rail net as well as a terrain model and the buildings. Most of the model data was provided in a GIS system based on Ordnance Survey data. Grid results of noise levels were imported back into GIS and in a second phase the GIS system was used to assess the noise impact on the population. The unique alphanumeric Ordnance Survey Addresspoint Reference or OSAPR was chosen as the common geo-reference for dwellings. Façade point calculations of each dwelling helped to provide a database of buildings and related noise levels. This was used to define average and extreme noise exposures per dwelling. Two main data sources of population data per dwelling were linked with these results in a relational database. GIS has since been used to query these results for various aspects.
The views and opinions expressed in this paper are those of the authors and not necessarily those of Birmingham City Council. BACKGROUND Back in 1997 debakom GmbH, Germany, produced Noise Maps of Birmingham for road, rail, aircraft and some industrial sources. The model data was set up in the LIMA program system for environmental acoustics, based on map data provided by the Ordnance Survey. This model and the results were imported into the GIS system already in use at Birmingham City Council. In a second stage of this study the council has used this data to evaluate the noise exposure of Birmingham residents from road traffic noise. In a simple procedure this could have been a task for a GIS system by just combining the grid based noise levels of the hole mapping area with average number of inhabitants per area for regions, such as suburbs, planning zones or building blocks. Due to a lack of more detailed data such investigation has recently been performed for the 18000 km2 area of Thüringen. Fig. 1 Birming Noise Map Fig. 2 Noise expose base on inhabitants per planning zone A MORE DETAILED APPROACH FOR EXPOSURE ANALYSIS The EU directive requires a more detailed analysis of the noise exposure of inhabitants that will look at the situation at each dwelling. As normal GIS data does not provide information as detailed as this some aspects of such analysis will be discussed based on inhabitants per building.
Based on a theoretical example, this paper will describe the influence of some of the parameters involved and then describe the procedure followed in Birmingham. RELATIONSHIP BETWEEN NOISE EXPOSURE AND POPULATION DENSITY To define noise exposure, levels at building facades and free field levels can both be taken into account. In our example we will only look at the façade noise levels. Noise exposure can be documented in an non-rated manner by showing the distribution of inhabitants depending on the average and the maximum façade levels of their building (Fig.3). Fig. 3 Table showing distribution of inhabitants Fig. 4 Noise-Annoyance relation A number of noise exposure definitions have been suggested and implemented in the software products. For the example case we use a study by German Umweltbundesamt that gives a relation between the likelihood of someone being severely disturbed by a noise and a certain noise level. The original study additionally distinguishes between different types of noise sources, but the Landesumweltamt Nordrhein- Westfalen suggested an integrated formula, based on the functions in Fig. Xx. Using these formulae, the number of severely disturbed people may be estimated in a manner which, unlike some other methods, is easy to interpret.
EXAMPLE The geometry of the example situation is presented in Fig. 5. and grid results in Fig. 6. It shows a symmetrical layout of buildings with heights of 6 or 15 m and different density with some parts near an elevated road and others at a distance. All together 8 zones can be distinguished. Five thousand inhabitants have been apportioned among the buildings in proportion to floor space. Exposure has been calculated using three different ways to define the receptor positions in front of the buildings: I) Receptor at a height of 4 m, only placed at the façade that faces the road II) Receptor at a height of 4 m on all facades III) Receptor at a height of 2.8, 5.6, {8.4, 11.2, 14} (m) on all facades Fig. 5 Model situation Fig 6 Grid results As result we receive the number of severely disturbed people for the different zones: Zone Method I Method II Method III dense high near no. no. no. A: Y Y Y 532 401 491 B: N Y Y 226 195 217 C: Y N Y 256 256 259 D: N N Y 60 60 60 E: Y Y N 443 205 352 F: N Y N 205 152 163 G: Y N N 210 206 211
H: N N N 52 50 51 Conclusions from the example could be: For low building heights, choosing an average calculation height of 4 m produces results that do not vary significantly for different densities and do not depend much on the selected relevant façade. For higher buildings the results of method I or II significantly deviate from a comprehensive analysis of all facades (method III). THE BIRMINGHAM APPLICATION In Birmingham, many of the citys residential areas s consist of 2 story residences. Thus, working on an average receptor height of 4 m is easily justified. It can also be shown that interpolating the façade noise levels from neighbouring grid results is an acceptable method as long as the grid results identify receptor positions inside buildings. When comparing the results of the calculation of façade levels (method II) and the interpolation from grid calculations for a test area, the results for all façades showed an average interpolation error of 0.34 dba with a standard deviation of 0.78 dba. Of the 2457 façade positions in our test case, only 19 (just over 0.7%) had an error of between 3 and 4 dba and there were no errors above 4 dba. To enable each individual residential building to be identified, Addresspoint, a database marketed by the United Kingdoms mapping agency, the Ordnance Survey, was used. This product assigns a unique 20 character alphanumeric code to every property in the UK that receives a postal delivery. Other fields in the database include the full address and, importantly for this project the X and Y co-ordinates of a point approximately in the centre of the property. A subset of this data was output as a text file listing the addresspoint for each residential property along with Tile Addresspoint N day night NW day night W day night their respective X and Y ASP1385.BNA APW5878G5J6498G01Q 36 26.7 42.6 34 0 0 ASP1385.BNA AP72878G5J7498GG1Q 36 26.7 42.6 34 0 0 ASP1385.BNA APPD6L855J4499HGCQ 40.5 31.7 0 0 42.2 33.7 coordinates. This data was ASP1385.BNA APFG6L855J4499H0CQ 40.5 31.7 0 0 42.2 33.7 ASP1385.BNA APKE0C855J349AJGVQ 39.8 31.1 0 0 41.4 32.9 then used to add an additional ASP1385.BNA APP987805J349AG03Q 39.8 31.1 0 0 41.4 32.9 ASP1385.BNA AP3B6B8C5JU498JGUG 0 0 41.9 33.3 0 0 ASP1385.BNA APWEA3825JT498J0X0 0 0 41.9 33.3 0 0 attribute called addresspoint to the building database used by Fig 7 Addresspoints and façade levels related to direction Lima. When the façade level interpolation is carried out, the results of each calculation are linked to their respective property using the addresspoint (Fig. 7) As can be seen, to associate an individual façade with its noise level, the values are linked to
one of eight compass points e.g. N, NW, W etc. Having obtained façade noise levels, information on population density is required before an estimate of noise exposure can be made. This information has been obtained from two sources, the register of electors and the cities Education Department database of children of school age. In the UK, with a few minor exceptions, everyone who is eligible to vote in an election is registered with their local authority from the age of 17 (the register of electors). This database has been stripped of all personal information to leave us with the number of adults at each residential address in the city (Fig. 8). The Education Departments database lists all of the children over three and less than 16 years old. This has again been stripped of any personal information to leave only the number of children at each address. Both of these databases also use Addresspoint as their reference and this has been used to facilitate merging them into one large database giving the number of people living at each address in the city. It has to be said that it has not been easy to get all of this data together. Different agencies and even other Fig 8 Population at addresspoint departments within an organisation use many different systems of data collection and storage. These systems have evolved over a considerable period of time and their origins pre-date modern database practice. This has resulted in a great deal of work in re formatting data. Inevitably, some people have been omitted due to limitations in the data sources. At present, we are unable to include children less than three years of age. One other source of error is students living in university halls of residence. The address point reference for these halls is often the university itself, which may be some distance away from where the students actually live. Other potential sources of errors are residential homes for the elderly and children s homes. Despite these problems, checking the number of people we can account for against the population figures estimated from census data shows that less than 2% of the population of Birmingham are unaccounted for. Finally, on the subject of data collection, it is vitally important not to underestimate the time needed to gather data together and to get the data into a format that is usable by you.
Using Addresspoint as the common geo reference throughout has enabled all of the data sources, both noise and population density to be combined readily in a relational database from which a variety of queries can be run (Fig. 10). Since originally writing this paper, the project has moved on. Using the Circulating Points facility in Fig. 9 GIS View Lima, we have recalculated noise levels at the facades of all buildings in Birmingham and have imported this data into a geographical information system. A further development has been the introduction of a new Ordnance Survey product called Master Map. Unlike its predecessor Landline, this product contains all buildings, including terraces and semi-detached properties, as individual polygons. This enables a greater degree of accuracy when assigning population data and noise exposure data to each individual residential unit. This approach enables spatial queries to be run. Such queries could include by road type or administrative area thus assisting with efficient and cost effective targeting of resources. (Fig. 10). This same GIS will also contain data on Fig. 10 CONCLUSION Estimating noise exposure by merging façade noise levels with numbers of inhabitants could develop into a powerful political and planning tool. Judging noise exposure on the basis of a uniform height can be acceptable for achieving global targets. However for more specific uses, for example local planning, the evaluation should take into account noise levels at different heights. Façade noise levels derived by interpolation from (existing) grid data can deliver acceptable results and, therefore, offers a fast alternative to a dedicated façade calculation.