An introduction to a new Faber Maunsell contaminated land risk assessment tool by its developer, Jonathan Welch.
Assessment of brownfi eld sites has undergone a revolution in the UK since the release in March 2002 of the Contaminated Land Exposure Assessment (CLEA) model and associated contaminated land reports by the Department for Environment, Food and Rural Affairs (DEFRA) and the Environment Agency (EA).
CLR7-10 introduced a new risk based methodology for assessing the possible adverse affects to human health from exposure to chemicals in the ground. Under the new regime, human health risk assessments of potentially contaminated sites must consider levels of exposure to chemicals and the sensitivity of human receptors for the specific land use.
DEFRA and the EA also initiated a programme for the delivery of toxicology (TOX) and soil guideline value (SGV) reports. These provide authoritative sources for chemical properties for use in risk assessments as well as SGVs for standard land uses to simplify routine assessment of priority substances. SGVs represent the concentrations of individual substances posing either a minimal or tolerable risk for residential sites (with and without vegetable consumption), allotments and commercial or industrial sites.
Two major hurdles existed for consultants attempting risk assessment in accordance with CLR guidelines. Delivery of SGVs proved to be very slow. SGVs were available for seven substances in March 2002, but only three more were finalised by January 2007.
Also, the CLEA2002 software proved to be very restrictive in its ability to model alternative land use scenarios or add new chemicals.
CLEA2002 became outdated with the release of DEFRA/EA Briefing Notes 1-3 which modified the CLR10 methodology for dermal contact and inhalation of indoor vapour, and its successor spreadsheet CLEA-UK became eagerly awaited from March 2003.
Faber Maunsell watched development of the CLEA model with interest. None of the computer models available to produce site specifi c assessment criteria (SSAC) for contaminants were fully compliant with the CLR10 guidelines. To take account of the variability and/ or uncertainty in human behaviour and sensitivity to contaminants, selected input parameters in CLEA are approved statistical distributions rather than discreet values.
In a probabilistic model, distributions of input parameters are randomly sampled over hundreds or thousands of simulations to produce a distribution of possible outcomes.
By policy and convention the 95th worst case percentile from the output distribution is taken as the solution.
Faber Maunsell realised that its Microsoft Excel techniques for probabilistic modelling of groundwater could be used for human health risk assessment. This led to development of the Contaminated Land Risk Assessment Tool (FMCLaRAT) which is a full implementation of CLR10 and subsequent Briefing Notes 1-4.
FM-CLaRAT comes in two versions. The user version contains standard databases of soil type, land use, buildings and chemical properties.
Site specifi c data is accepted for pH, soil organic matter content, depth of contaminant and bioaccessibility for ingestion pathways.
The developer version allows complete access to underlying databases to add site specifi c conditions, risk groups or chemical properties.
FM-CLaRAT has a unique graphical representation of the probabilistic solution. Site specific assessment criteria from multiple iterations are graphed in decreasing order from left to right against percentile. The strip area above each concentration is divided by colour in proportion to the relative significance of each pathway.
Figure 1 shows the FM-CLaRAT output for arsenic for a residential land use with consumption of home grown vegetables and default CLEA conditions. The assessment criterion is dominated by soil ingestion over consumption of home grown vegetables, with an insignifi cant contribution from dermal contact and dust inhalation.
The ragged interface between soil ingestion and vegetable consumption pathways is a consequence of both pathways being probabilistic. For naphthalene, Figure 2 shows that consumption of home grown vegetables and inhalation of indoor vapour are equally important. Inhalation pathways are not probabilistic, which leads to a smooth contact between indoor inhalation and ingestion pathways.
Since the release of CLEA-UK (beta) in November 2005 it has been possible to provide independent verifi cation of Faber Maunsell's calculations. FM-CLaRAT provides more consistent results than CLEA-UK, without any significant variation between runs. It also integrates vegetable consumption and inhalation pathways correctly when exposure by these pathways is limited by maximum aqueous solubility and vapour pressure, respectively.
The software was fi rst used to calculate generic assessment criteria (GAC) for routine assessment of contaminants for which no SGVs have been published. But a big advantage is speed. A typical run of 1000 iterations takes less than 5s in FM-CLaRAT compared to several minutes for CLEA-UK.
This, combined with the graphical display, lends itself to routine use of the software to carry out risk assessment, perform sensitivity analysis and select remedial options. With FM-CLaRAT it is practical to make assessments site specifi c in keeping with the principles of the new risk assessment regime.
Jonathan Welch is principal hydrogeologist at Faber Maunsell and developer of FM-CLaRAT.