The Precision of Methods for Soil Analysis
Soil analysis and its influences on risk assessment of contaminated land
Over the past few years there has been a movement to increase the precision of methods of analysis of soils for the assessment of contaminated land, not least with the introduction of MCERTS.
I question whether this is the area where most fundamental errors are found within the contaminated land assessment process and therefore whether we are mistaken in spending all our time and effort going down this route. I go on to discuss some of the problems that can occur specifically with the data produced from the analysis of soils and its effect on the subsequent risk assessments. I conclude by propounding the need for communication between professionals in the assessment process.
In 2003 the Environment Agency established the Monitoring Certification Scheme (MCERTS) to attempt to ensure the delivery of high quality environmental measurements. The scheme ranges from the product certification of instruments and the competency certification of personnel to the accreditation of laboratories conducting chemical testing on soils.
Why did they introduce it? For, amongst others, one important reason. Chemical test data from soils is used by the Environment Agency to support its regulatory activities under a number of regimes, including:
- Part IIA of EPA 1990, Pollution, Prevention and Control (England and Wales Regulations) 2000
- Waste Management Licensing Regulations 1994
In addition, the Environment Agency state that ‘MCERTS “Performance Standard for Laboratories Undertaking Chemical Testing of Soil” requires that, where results are to be submitted to the Agency for regulatory purposes, laboratories must both fulfil the general requirements of the European and international standard BS EN ISO/ IEC 17025:2000 (accredited by United Kingdom Accreditation Service (UKAS)), and meet specific method validation and performance requirements laid down in the Performance Standard.’
The provision of quality, reliable data is therefore extremely important in circumstances where the costs of falling foul of the legislation and of the regulators are potentially great. Demonstrating the validity of analytical methods is an essential requirement of the MCERTS scheme. Every laboratory must have utilised the essential statistical techniques needed for the evaluation of the validation data for its analytical method, produced to meet the requirements of the MCERTS performance standard for soil analysis.
The MCERTS performance standard for laboratories undertaking the chemical testing of soil provides:
The selection and validation of methods
Sample pre-treatment and preparation
The estimation of measurement uncertainty
Participation in proficiency testing schemes
The reporting of results and information
The application of MCERTS is dependent upon the proper use of suitable methods, standards, services and equipment, trained and qualified personnel, effective planning, quality assurance and quality control. MCERTS does not specify the techniques used for obtaining measurements of a determinand but provides performance requirements and testing procedures and therefore, in essence, it is only concerned with how well a determinand is measured. In brief the MCERTS fundamentally aims to improve the quality of the analysis of soils and not necessarily the applicability of the data.
Precision and accuracy
How does this move to more precision in the analysis of soils then fit into the risk assessment of land affected by contamination? How does the precision and accuracy of the result of any analysis influence the assessment process?
Before discussing this it is best to explain the difference between precision and accuracy. Precision relates to the quality of an operation by which a result is obtained, and is distinguished from accuracy, which relates to the quality of the result. Precision is an indication of the uniformity or reproducibility of a result whereas accuracy relates to the truthfulness of a result.
Which of these particularly interest us in contaminated land investigation? – well both of course but what use is extremely precise analytical data if the samples tested are not representative of the site, location or depth that they are collected from? The question is rhetorical. Why is it, therefore, that there has been a significant movement over the past few years, including the MCERTS, to more effectively control the methods by which we analyse soils rather than establish appropriate data on which to base a risk assessment.
I must state that I am not against the MCERTS at all but I am concerned that the more focus we place on what we can easily achieve and can influence the more we will tend to ignore, however unconsciously, the imponderables which may generate the more important errors when it comes to the assessment of risk associated with contaminated land investigations. It is human nature to examine those issues that are easy to control, such as the analytical method, rather than those which are difficult to control and standardise.
Errors from sampling
From my perspective, as a consultant and investigator in the contaminated land field, I wish we could introduce a larger degree of consistency in the sample collection phase of contaminated land investigations. We need to remove the often subjective approach to the collection of say 1kg samples representative of a possible 1m thick layer of made ground in a trial pit 3m long by 1.5m wide. It is this issue that provides the most significant error factors in the whole approach, a situation which is well understood by the assessment fraternity.
With regard to this viewpoint the following statement is taken directly from BSI 10175 ‘Investigation of potentially contaminated sites, Code of practice’ – “It is important that any sample submitted to the laboratory is representative of the location and depth from which the sample was taken… The errors associated with sampling in site investigations are generally greater than those associated with the analysis. It can therefore be more informative to analyse a greater number of samples using methodology fit for the purpose, than to analyse a smaller number of samples using a more accurate method.”
The British Standard, the bible for contaminated land investigation, goes on to give advice on sampling by providing information on what methods are available to reduce the sampling errors such as the use of nine point samples. This consists of taking nine increments of the same volume and combining to form a single representative sample. The points of the increments can be localised in order that the sample is representative of that sample location. However, from experience, it is very rare that this is carried out by any investigation consultancy.
It may appear easy to standardise a method of analysis, with our long history of development of test methods in this country, although
having served as chairman of the BSI Soils Quality committee for the past few years, I can tell you it actually is not. However, it is easier to standardise test methods than to attempt to standardise the means of sampling soils or made ground from a trial pit.
There is an old acronym expression used in the computer industry ‘GIGO’, which means garbage in garbage out, the sentiment of which neatly explains my viewpoint when it comes to the whole concept of contaminated land investigations. If the samples on which analyses are conducted are not representative of the ground then all the data produced from the analysis, however precise and accurate, is completely meaningless.
There are other problems, for example the question of how much sample do you collect for the soil to be representative. The laboratory that will analyse the soils is well aware of the fact that in general the larger the sample size, the lower the potential for error. So why is that they discourage the provision of samples consisting of bulk bags with a large sample volume and weight?
A number of reasons. If a large amount of soil is sent to a laboratory, they have to analyse it all – which results in the use of a large amount of space taken up with air-drying of the sample and the time needed to physically remove those materials with a greater particle size than 2mm. Further, there will be a need to either grind the whole sample or else reduce the sample volume to one that is easier to handle – with the probability of error factors.
If a large volume of soil is sent for analysis – say a 10 kg bag – only a small portion of this sample will be destructively tested which would result in a significant weight of material left to be ultimately disposed of. This will create storage space problems at the laboratory and, after the requisite amount of time for storage to allow the client to instigate additional testing, a significant cost of disposal of the material. Because of the changes in Waste Management legislation over the past year the cost of disposal has increased markedly.
Analysis of soils
Soil has a wide range of variability not least with respect to its matrix – furthermore, the type of soil the sample is made from can have a significant impact. The pH of a soil may influence the end result, as would the presence of any material that could interfere with the determination – this is certainly a problem when it comes to the colorimetric testing of samples. All these can have the influence of reducing the accuracy of the data produced by analysis. There are other problems primarily with consistency of data.
I have been assessing analytical data from the analysis of soils collected from contaminated land sites for the last twenty-five years and with time as a consultant you develop a feel for a set of results. There have been numerous occasions when, after I have queried data with the laboratory, they have amended their results.
You may consider from my previous statements that I believe laboratories to be, in some instances, ineffectual but while I will question their approach occasionally I do not consider them incompetent. However, I do believe there are often inherent problems in their method of approach. I will give an example. On one of the sites we have looked at recently we analysed for a range of determinands, one of which was nickel. There had been a previous investigation by another consultancy, which had identified that nickel was not regarded as a problem. They had identified total nickel concentrations on average between 30 and 40 mg/kg on dry weight. Our investigation gave a new set of data with nickel concentrations between 40 and 70 mg/kg.
This is not a particularly large discrepancy but, since the Soil Guideline Value under CLEA (Contaminated Land Exposure Assessment) for nickel is 50 mg/kg, the site became overnight contaminated with nickel. The results were obtained both within made ground and within samples collected of the natural strata. The laboratory confirmed the results as being correct since their QC standards had fallen within the accepted range.
Our next set of data confirmed the approximate range of results, but the next data, including more samples from each of the earlier investigated areas of the site gave results in the range 20 to 40 mg/kg. Unfortunately the results had already been incorporated in our interpretative report which had been issued to the local authority. Where did this leave us? With a significant amount of explaining to do to the client and persuasion of the Environmental Health Department of the scale or otherwise of the problem. Since some of the contaminated samples had been of natural ground with the samples from above being essentially free from contamination an argument was propounded that indicated the results were either suspect or it was natural contamination. Further testing proved the results were suspect.
I have discussed metal contamination but what do the results obtained from the analysis mean in reality? Most of the readers of this article will consider that when the analysis from the laboratory is reported as ‘total metal’ concentration that it means total metal. It will surprise some of you that this is not the case and that the analysis normally involves the digestion of the sample with strong acid, which may only extract 80- 90% of the metal. One might wish therefore to define the result as “strong acid extractable metal”… perhaps we should stick to ‘total’ and educate people more. It is the best approach we currently have so we had better live with it.
Does this actually matter in practice? The answer is no – so long as the data which has been used to determine soil guideline values under CLEA has been generated using the same approach. But has it? In any eventuality, if the SGV is based on toxicology of the individual metal species – the toxicology will change dependent on what compound the metal is present as… nickel oxide will have a different toxicology to nickel sulphate for example. Since the analysis is of acid extractable metal how can the results be compared? I think there is a fundamental difficulty here. This is a strong argument against the need to spend more time and effort on greater precision.
If we look at the analysis of soils for Total Petroleum Hydrocarbon (TPH) then the situation becomes even more problematic. Each laboratory has its own favoured method or methods of analysis. Quite often they measure different things, because the analysis is general in nature and is of a range of contaminants, and the direct comparison between what is reported by one laboratory as TPH and another is almost impossible. You almost have to be a qualified Organic Chemist, and then with a specialism in hydrocarbon assessment, to interpret the data. This is normally beyond the typical consultancy but not normally beyond the laboratory.
Communication between professionals
As a consultant I am in a better situation than the laboratory when it comes to the collection phase of the investigation of soils from land affected by contamination. I know where the sample came from, what the historical industrial development of the site was, what the matrix of the soils was, why the sample had been taken and where and what the final interpretation of the data needs to achieve i.e. is it the redevelopment of a site for residential reuse or is it merely an estimation of liability for due diligence purposes. However, although I am an experienced individual when it comes to the assessment of contaminated land, I am not a qualified analytical chemist. My knowledge of the specific test methods to determine hydrocarbon parameters cannot be as great as the specialists in the laboratory.
Why is it, therefore, that the average consultant does not communicate with the laboratory to obtain the laboratory’s view of what the correct analytical testing should be to achieve an end result? It is almost as if there is an “us and them” mentality where communication between professions is to be avoided at all costs. Is this logical? Decidedly not.
The analytical chemist is an ‘expert’ in the analysis of samples. The consultant may be an expert in contaminated land assessment but how often has he ever conducted any analysis? I would postulate very rarely if ever. In order to achieve the most out of the chemical analysis of soils we have to introduce the concept of integrated thinking between the individuals involved.
There has to be communication at all levels between:
- The individual who collects the samples from a site
- With the person who has conducted an assessment of the history of the site
- With the person who will decide what suite of testing is needed
- To the analytical chemist who can identify the most effective test suite
to supply the final goal of the assessment. That goal is to produce data which is as meaningful as possible and which allows decisions on the risks present on a piece of land to be made as value judgements rather than, at best, informed guesses
So to conclude, having stated the above why are we bothering trying to determine the degree of precision of the results? I sometimes wonder whether we approach the analysis of soils in the right manner. I am aware that there is a need to provide consistent results which are as accurate as possible, because when we come to assess results for different sites we shall all be singing from the same hymn sheet, but we must be aware of the inadequacies of the data. Benjamin Disraeli was spot on when he said “to be conscious that you are ignorant is the first step to knowledge”. I wholly agree with his sentiments.
Clive Griffi ths is Environment Director at STATS Limited, where he provides advice to property developers and investors on a wide range of environmental and pollution control issues. Clive’s distinguished career has been unusual in that he has held positions of responsibility within organisations from each of the main environment sectors, namely in environmental regulation, waste management and remediation, analytical laboratories and consultancy. He was one of the fi rst practitioners in the UK to become a registered Principal Environmental Auditor with the EARA scheme. For the last twenty years Clive has been at the forefront of developments in the investigation, assessment and remediation of contaminated land and he is currently Chair of the British Standards Committee on Soil Quality.
STATS Limited Porterswood House, Porters Wood, St Albans, Hertfordshire AL3 6PQ Tel: +44 (1727) 833261 www.stats.co.uk [email protected]
Published: 01st Jun 2006 in AWE International