CLEA: light at the end of the tunnel?
Jo Strange, associate director of Card Geotechnics Ltd, explains the government’s newly published Contaminated Land Exposure Assessment, CLEA.
After many tears of waiting, the government has published its Contaminated Land Exposure Assessment (CLEA) model in the form of R&D Publications Contaminated Land Reports (CLR) 7-10. These publications represent the next step forward for contaminated land assessment in the UK and the Environment Agency has adopted the guidance within these documents as the preferred methodology to be implemented in the UK.
In order to promote a unified approach to the implementation of CLEA, Card Geotechnics have devised a protocol for applying it to practical projects. In applying CLRs 8-10, it is worth noting some of the assumptions made in the guidance:
- The guidelines were developed to address issues arising from Part 2a (EPA1990) and the Town and Country Planning Act, based on assessment of industrial land- not green field or agricultural land.
- The guidance applies the ‘suitable for use’ principle and not multi-functionality and uses a probabilistic model based on a ‘Monte Carlo’ simulation.
- It is not statutory…but beware of not using it without good justification.
- Soil Guidance Values (SGVs) are intended to replace the Inter-departmental Committee of the Redevelopment of Contaminated Land (ICRCL) Action levels. Some apparently increased ‘limits’ only reflect this approach.
- SGVs are based on scientific information on health effects including toxicological data and ‘background exposure’, but the model only addresses human health issues in terms of chronic long term affects. The model is NOT appropriate for other receptors, eg controlled water.
- The quoted SGVs have only been determined for a few substances and limited S-P-R conceptual models. Outside this range, specific SGVs should be developed using the CLEA or another appropriate model.
- When used as a clean-up criterion, the SGVs can be applied for substances which have threshold effects. For non-threshold substances, site specific criteria are based on the ‘as low as reasonably practicable’ (ALARP) principle.
- Conservative ‘highest measured values’ are no longer used, instead ‘averaged values’ are assessed taking account of statistical accuracy.
- The model recognises the impact of capping on the SGVs but with no guidance on modelling. Instead, it relies on professional judgement and indirectly advocates capping as a remediation methodology.
Effect of uncertainties
The last on the above list may yet prove to be the quantum change in approach introduced by CLR 7. It is recognised that there are unavoidable uncertainties associated with sampling and analysing soils for contamination, with variations occurring throughout the process from sampling to reporting of results. True comparison of results requires the variations to be minimised. This is being addressed by the issue of BS 10175:2001, investigation of potentially contaminated sites, and the introduction by the Environment Agency of MCERTS as a quality assurance accreditation of laboratories.
The Monitoring Certification (MCert) scheme is a topic for discussion in its own right, but in summary, it should result in greater consistency between laboratories and record those areas where uncertainties may have occurred. The main impact on those specifying laboratory testing is ensure that they use an MCert laboratory, or run the risk of results being rejected by the Environment Agency. Close liaison will be needed with laboratories to minimise reporting of sampling uncertainties.
CLR7 works not on the principle of assessing ‘worst case’ results against a quoted standard but requires the assessment of mean and maximum results across specific ‘averaging areas’ by the application of statistical tests. This assumes that the mean of the test results may not be a true average for the material and that “elevated results” may represent the upper range of a population and not necessarily an ‘outlier’ or ‘hotspot’. CLR 7 provides a statistical test method based on an assumed normal distribution.
There is likely to be controversy over the definition of averaging areas and the use of statistical testing as they assume that sampling and analysis is carried out to a set random pattern. In reality, in accordance with BS10175, all investigations will be preceded by preliminary data assessment and will be ‘targeted’ and affected by obstructions such as existing buildings and services. Therefore, the assumed background condition for the statistical test is already invalid.
Defining realistic averaging areas for assessing investigation data will rely on an understanding of possible ‘hotspots’ and future land use layouts – often impossible at the time investigations are carried out. However, for this to be achieved, the approach taken must be both pragmatic and practical in order to keep the number of necessary test results within reasonable limits, i.e. one housing unit is not a suitable averaging area.
Where the statistical tests will be extremely useful is for remediation validation against a stated SGV. Previously one measurement, slightly too high, may have resulted in considerable additional work.
It is also worth noting the uncertainty inherent in developing SGVs for additional determinants. DEFRA has not been able to publish additional SGVs, presumably due to the lack of definitive toxicity data. It is therefore unlikely that specific limits derived by others will be any more certain. Calculation of site specific SGVs also requires assumptions on the soil types and pH, usually in relation to intrinsically non-homogeneous made ground. There is no guidance on how these parameters should be ‘averaged’ and adherence to the model implies the use of sample specific SGVs!
One area of debate has always been the blanket testing for the ‘ICRCL suite’. This is driven by ignorance, by the way laboratories are set up and partially due to clients’ comfort zones. CLR8 provides a new list of potential contaminants, which should be reviewed as a basis to test scheduling. Not all of these contaminants will be applicable, but now the guidance requires application of thought and judgement to avoid very expensive ‘blanket suite’ testing.
As with the introduction of Part 2a and S-P-R risk assessment techniques, a protocol involving additional professional assessment is being imposed on the industry through publication of guidance and its adoption by the regulators. Whilst nobody questions the need for a uniform approach, there is a cost implication in terms of assessing sites, excluding time lost in gaining a working knowledge of the model.
In reality, these costs must be eventually passed on to clients. Absorbing extra cost in already ‘squeezed’ consultancy fees will result in sites being assessed on the basis of too few samples. This, in turn, will lead to recommendations for remediation, when an increased or staged budget would have provided additional data possibly leading to a different conclusion.
In terms of the costs of sampling and testing, laboratory testing costs must go up to cover MCert accreditation and its additional QC requirements. A possible scenario is that samples will be scheduled on a ‘minimum necessary’ approach, but additional or larger samples will be taken to allow for subsequent re-testing without requiring further intrusive site works. Laboratories will then be left with large numbers of ‘unwanted’ samples in storage ‘just in case’. The testing of these at a later date will also create issues of storage deterioration and eventually add to overall analysis costs.
CLRs 7-10 can be seen to be a shadow of the tool it had been expected to be, being both limited in scope and substance. In the fullness of time, it has great potential to standardise the assessment of human health risk but, even with a full set of TOX reports and SGVs, it will still only be one tool in the overall risk assessment process. Whatever methods are adopted will be constrained by the lack of relevant toxicity data.
Practical experience has also demonstrated that remediation is more often driven by groundwater risks rather than human health. This issue is not addressed by CLEA which raises the question of why the Environment Agency, with its current remit, to protect controlled waters has become so closely involved with the model. Despite this, CLEA should be assumed as being here to stay, and consultants/clients should be prepared to provide human health assessments in that format.
Until the regulatory bodies have become familiar with the process there is certain to be ‘discussions’ and associated delays whilst a common understanding is reached. Some clients may then opt for remediation rather than delay, but perhaps more likely is that they will look to abandon seemingly uneconomic developments.
However, contaminated land need not be a barrier to successful development. Experience to date, using the CLEA model and appropriate quantitative risk assessment, has indicated that, while consultancy fees have increased to cover the more detailed assessment, in general it has been possible to reduce actual remediation charges thereby cutting the overall project cost.