Spatial Analysis of odour nuisance

 A GIS based-approach to investigate odour complaints in a water company operational area.

M. Fornés.

University of Leeds, September 2011.
  
Wastewater treatment works and their impact on environment, particularly on air quality, have been reported since air pollution started to become a serious threat in populated areas. Odour nuisance from wastewater treatment installations can also be linked to concerns on air quality especially since studies on odour impacts on human health started to show their effects in local communities.

The study intends to understand spatial patterns of odour complaints as well as to recognise socioeconomic characteristics of the company’s operational area and potential odour sources from the wastewater treatment facilities.

Aims:

The aims of the project are to understand the spatial and social distribution of odour complaints in the company’s operational area. Consequently, the analysis was divided into three main objectives:

· Firstly, to show where the concentration is, especially if any specific social group complains more.

· Secondly, to explain the concentration of complaints in selected areas in terms of socioeconomic factors linked with concepts such as environmental justice.

· Thirdly, to recognise the odour complaint patterns observed.

In order to achieve this, the researcher approached the problem in two ways. Initially by developing first a spatial analysis throughout GIS and then, once the whole operational area was analysed, the study tried to respond by characterising the areas on socioeconomic terms as well as the spatial behaviour of odour complaints. 

Georeferenced data (point shapefiles) on odour complaints provided by the company and collected in the last ten years in the company’s operational area were used to carry out a spatial distribution analysis. In addition, the water company provided also another point shapefile with the location of wastewater treatment works that permitted to carry out analyses by type of treatment and legal status (who operates the installation?...public? , private?). 

Finally, English wards were the spatial units chosen to perform the study which is optimal for representing socioeconomic differences and very widely used in studies of deprivation.

Objectives:

The first objective was the study of the spatial distribution of odour complaints throughout different techniques on spatial patterning of odour nuisance in the company’s operational area.

The second objective was the study of socioeconomic characteristics of the areas in which odour complaints are located:

·         Can odour complaints be explained by socioeconomic factors and their spatial distribution (e.g. Townsend index vs. Complaints/facilities)?

The third objective was focused on the investigation of factors linked with odour nuisance:

·         What explains the level of complaints?

·         Are smell complaints more affected by proximity to WWTW? or by other factors?.

All these questions were formulated as the researcher was working on the project proposal as well as after studying the type of data provided by the Water Company and relevant literature was consulted.

Methodology:

Methodology was divided in three parts. These parts were divided by objectives in order to meet the aims of the project.

Three types of analysis were performed to achieve the first objective:  

        Average neighbour distance analysis and spatial correlation analysis

·         Hot spot analysis (Getis–OrdGi) to investigate the clusters of high and low values (complaints rate)

·         Kernel density estimation.

Then, the study was moved to the investigation of socioeconomic characteristics of population in the company’s operational area. The objective was to know who is complaining by establishing a correlation between complaints and level of deprivation. The second objective was consequently divided as follows:

·         Study of the spatial distribution of deprivation related to complaints in the company’s operational area

·         Analysis of Townsend index and the distribution of complaints by deciles

Finally, the investigation finishes with the study of factors that affect the level of odour emissions. In order to investigate what explains odour nuisance the study focused at this level on legal status of the WWTW, their location and size, as well as the correlation between complaints and the index of deprivation.

Kernel censity estimation

The spatial distribution of odour complaints.

In this first approach the objective was to investigate the spatial patterning of odour complaints. In order to analyse the spatial pattern of odour complaints, four different methods of spatial analysis were performed. Firstly, the average neighbour distance analysis was conducted; then, the spatial autocorrelation tool was carried out to establish the relationship between the location and the rate of complaints in the company’s operational area. 

Following this, the hot and cold spot analysis (Getis – Ord GI*) was carried out and, finally, a simple Kernel density estimation was performed to complete the analysis of the spatial distribution of odour complaints.

The average nearest neighbour method from GIS was conducted in order to analyse if the complaints are clustered or dispersed in space (Lloyd, 2010). This method was useful to depict the spatial pattern of odour complaints which reflects that the complaints take place more in some places in the company’s operational area than in others.

Then, the spatial autocorrelation (Morans I) method was carried out to investigate if the points are clustered or dispersed depending not just from their geographic location but from other variable simultaneously (Rogerson 2001). In order to perform the analysis a complaints rate was derived. Firstly, number of households per ward was downloaded from Casweb, then complaints were summarised in every ward and then divided by the number of households in each ward. 

Among many other potential applications, the spatial autocorrelation tool permits the identification of an appropriate neighbourhood distance helping to find out the distance where spatial autocorrelation is strongest (Resources ESRI 2011). The spatial autocorrelation tool was carried out giving a positive Moran’s index that described clustering given the position of data and the odour complaints rate.

Hot and cold spot analysis method was applied in order to evaluate and detect the location of the clusters of high and low values (Longley 2005).

Socioeconomic aspects and distribution of complaints.

The second part of the analysis focused on socioeconomic aspects of the analysed area. In order to conduct the analysis on poverty, the Townsend index was downloaded from casweb. The index was chosen as the most suitable method to investigate socioeconomic differences in the company’s operational area. Dr. Norman (2007) performed a Townsend index at English wards level which is the most appropriate level of analysis to investigate the socioeconomic factors of the analysed area.

Initially, the study was conducted by mapping and extracting data from both the Townsend index and the location of odour complaints. Townsend deprivation index was mapped dividing the index into five quintiles. Then, GIS intersect tool was used in order to obtain the number of complaints in each quintile. Quintiles from 1 to 5 were selected and then intersect tool was carried out to get the number of complaints in each quintile. Quintiles 1 and 2 represent the less deprived areas where the total number of complaints reached to 1662 complaints.

On the other hand the most deprived areas represented by quintiles 5 and 4 have 1600 complaints. Then the analyses changes from quintiles to deciles.

Although quintile analysis is usually used in studies of deprivation, decile-ones were considered in this case for being more suitable. 

Dividing data into deciles permitted to identify variances in number of complaints depending on level of poverty. Therefore the investigation focused in differentiating the less and most deprived areas and the number of complaints in each other. The graph below left allows checking differences in the number of complaints between decile 1 which represents less deprived areas in the company’s operational area and decile 10 which corresponds to the most deprived areas.

Furthermore, the analysis in deciles permitted grouping less and most deprived deciles, allowing testing differences in the number of complaints between the less and the most disadvantaged areas.

Odour complaints and legal status.

This part of the analysis focused on the legal status of the wastewater treatment works in order to know the impact of water company’s facilities by differentiating from public and private installations where the company may have little or no direct responsibility.

The analysis intended to obtain the number of complaints in different buffer ring distances.

The chosen distances (<300m, 300-600m, 600-1000m) were based in relevant literature on odour impact assessments from WWTW (Micone, 2009). Thus, the multiple ring buffer tool from GIS was performed to get the number of complaints from the company’s facilities, public and private with the company maintenance and the rest of the installations. Consequently, public – private facilities maintained by the company and independent public – private facilities were extracted either from the original file in order to conduct the analysis.

This approach permitted to differentiate complaints which might be related to the company’s facilities or those which are maintained by the water company from public and private installations.

The investigation revealed a higher number of complaints related to public – private facilities than those to the water company. This information might be relevant in order to delimit the the company’s legal responsibility on odour nuisance which might have importance and positive legal implications.

Data extracted from the intersect analysis of buffer distances and date of complaints permitted to analyse complaints not just by legal status but by distance and dates.