DescriptionThe aim of this project is to gain understanding of the role of statistical analysis by developing an important application in the study of contaminated land. Developers are increasingly using contaminated or brown field sites due to rising environmental awareness and pressure from UK governmental policies (e.g. the government's commitment to construct the majority of new housing on brown field sites). Due to the hazardous nature of the redevelopment of contaminated land, it is essential that the risk posed is fully assessed, based on a comprehensive site investigation. Risk assessment identifies factors such as hazards to end users, site workers and the local environment and gives an indication to the size and cost of the remediation programme (if required). As more information regarding the site becomes available at different stages of the investigation process, the risks posed by the site can be assessed in more detail. However, the amount of testing that can be carried out as part of a site investigation is always limited by time and budgetary constraints. There are inevitably considerable uncertainties both in terms of the levels of contamination present and also the soil properties that affect the degree of hazard posed by the contaminated materials. Therefore, it is vital that any modelling recognises these uncertainties. Statistical techniques can be used to evaluate the scale of uncertainties, and sensitivity analysis used within risk assessment can identify the potential significance of inherent uncertainties. The process of site inspection, especially where it involves sampling and analysis of different substances in different media, requires careful design. The site investigation process could be better planned if the major areas of uncertainty are highlighted. This could be used to provide a sampling design to focus the limited amounts of testing to best effect. This project will consider current practice in contaminated site inspection, based on comparing measured results against deterministic soil guideline values determined by the Contaminated Land Exposure Assessment model. We will look at alternative approaches, based on spatial statistical models which may be used to quantify the uncertainties in the spatial distribution of contaminants. Such models may be used: (i) to make inferences about contamination levels for any given collection of inspection outcomes; (ii) to direct efficient land inspection schemes; (iii) to develop appropriate Bayesian decision models for treatment of contaminated land, balancing the uncertainties for the various outcomes with the utility of their respective outcomes. This analysis may be applied both to simulation models and to field studies.Prerequisites3H Statistical Methods, and some experience with R. 3H Decision Theory would be useful, but is not essential.ResourcesCL:AIRE Technical Bulletin 7, Improving the Reliability of Contaminated Land Assessment using Statistical Methods: Part 1, Basic Principles and Concepts March 2004 (available from the CL:AIRE website http://www.claire.co.uk/bulletin.php) [Discusses uncertainty in the context of a contaminated site with a few useful contaminated land references] Nathanail C.P, The use and misuse of CLR 7 acceptance tests for assessment of risks to human health from contaminated land, Quarterly Journal of Engineering Geology and Hydrogeology 2004; 37: 361-367 [An overview of the Environment Agency's approach to technical guidance relating to statistical tests and its shortcomings.] Freeze R.A, et al. Hydrogeological Decision Analysis: 1. A Framework, Ground Water 1990; 28: 738-76 [Detailed description of an approach to decision making for engineering projects where the hydrogeological environment plays a role. A risk-based approach is broken down into a decision model, a simulation model and an uncertainty model.] Diggle P.J, Statistical Analysis of Spatial Point Patterns, Arnold 2003 Ripley B.D, Spatial Statistics, Wiley 1981 [Good general books on spatial statistics.] |
email: Michael Goldstein