7.3 Indicator-based methods & dispersion models
Indicator-based Methods
In many epidemiological studies, data on the concentration of a toxicant within the environment are unavailable. However, the locations of probable pollution sources are often known and toxicant concentrations are assumed to be higher near these sources. Such techniques, which use indirect measures of exposure instead of field measurements, are known as indicator-based methods. Indicators are thus used as a measure of last resort for assessing exposure, when no direct measures of toxicant concentrations are available.
One of the most widely used indicator-based methods is to use distance from source as a proxy measure for exposure. In many UK towns, for example, no data are available for atmospheric levels of sulphur dioxide emitted from traffic, which may be a contributory factor in respiratory disease. Consequently, buffer zones drawn around major roads may be used as a surrogate measure of inhaled sulphur dioxide. Sometimes traffic density or traffic volumes may be used as indicators either instead of or in addition to proximity to roads. In much the same way, installing monitoring equipment to record lead levels in drinking water supplies is too expensive for most water boards. A simpler, indicator-based measure of exposure to lead in drinking water would be to identify the age of housing stock. Legislation was introduced that banned lead water pipes in new housing and so lead ingested in water should be greater for residents of older housing blocks.
Think of 3 problems that there might be in using such indicators to assess exposure. When you have thought about this, click on the ‘answer’ link below.
Answer 1
There are many obvious problems with indicator-based methods. Such methods assume that pollutant levels are correlated with distance from source. In reality, however, groundwater flow or strong winds may produce elevated levels of toxicants far from sources of pollution. Source activity often varies over time and periods of high emission may coincide with unusual meteorological conditions. It can be difficult to convert an indicator into a specific dose of a contaminant. For example, it may be unclear what effect living within 500 metres of a road has on the total quantity of sulphur dioxide inhaled by residents over a year.
Hide
Dispersion Models
In some instances, not only is the location of a pollution source known, but levels of emissions over time are also available for analysis. Such data may be combined with meteorological or hydrological data within a dispersion model to produce a potentially more accurate depiction of ambient pollutant levels. A dispersion model is a mathematical prediction of how pollutants from an emission source will be distributed into the surrounding environment under given meteorological or hydrological conditions. For example, a slagheap of waste material from a mine may be a source of hazardous heavy metals. With an appropriate knowledge of groundwater flow, it may be possible to model the movement of these heavy metals in water and assess the risks to surrounding aquifers. In much the same way, the pathways followed by heavy metals from the slagheap in surface waters could be followed via a run-off model. More complex dispersion models are not implemented in most standard GIS packages, and so separate modelling software is often necessary to undertake dispersion modelling.
Think of 2 difficulties that might exist in using dispersion models to assess ambient pollutant levels. When you have thought about this, click on the ‘answer’ link below.
Answer 2
In practice, dispersion models are a simplified representation of the passage of pollutants through the atmosphere or water. In atmospheric models, turbulence is difficult to predict, depending on localised variations in temperature that are difficult to capture in a GIS. Processes other than transport between locations may affect the atmospheric concentration of a pollutant. For example, chemical reactions may affect the concentrations of specific pollutants and air-borne pollutants may be removed from the atmosphere either by rainfall (wet deposition) or through turbulent airflow (dry deposition). Similarly, in hydrological models, hydraulic conductivity (the rate at which water can move through a permeable body of soil) can vary markedly over very small scales, making groundwater flow unpredictable. In short, 2 problems are firstly that data for dispersion models are often difficult to obtain and secondly, the models necessarily simplify real-world dispersion and their output may therefore be unrealistic.
Hide
All of these problems mean that a dispersion model will not necessarily produce a better representation of environmental risk than indicator-based methods. Much depends on the model used and the specific environmental conditions. In practical terms, implementing a dispersion model is a complex task that often involves detailed understanding of meteorological processes as well as data transfer between GIS and modelling software.
Activity
Download the zip file and complete the practical exercise if you are using ArcGIS Desktop or alternatively use this zip file for ArcGIS Pro instructions. The exercise involves mapping the population at risk of exposure to air-borne dioxins from a waste incinerator using indicator-based methods.
References (Essential reading for this learning object indicated by *)
The article below is an example of an indicator-based approach to mapping air quality:
Mavroulidou, M., Hughes, S. J., and Hellawell, E. E. (2004) A qualitative tool combining an interaction matrix and a GIS to map vulnerability to traffic induced air pollution. Journal of Environmental Management 70 (4), 283-289.
The article below provides an example of dispersion modelling being used to evaluate air quality:
Scoggins, A., Kjellstrom, T., Fisher, G., Connor, J., and Gimson, N. (2004) Spatial analysis of annual air pollution exposure and mortality. Science of the Total Environment 321, 71-85.
The original study on which the practical exercise is based is described in this article:
Diggle, P. J., and Rowlingson, B. S. (1994) A Conditional Approach to Point Process Modeling of Elevated Risk. Journal of the Royal Statistical Society Series A-Statistics in Society 157, 433-440