There are an increasing number of portable instruments (and test kits) that environmental geochemists can take into the field to get rapid measurements of contaminant concentration. They can measure a range of different inorganic and organic contaminants in many different media, such as soils, wastes, waters and gases – and give results in a few minutes. This seems a very attractive way to enable rapid decisions on whether contaminants might pose a threat to human health, or just whether and where more measurements need to be taken. It seems to be a much better option than waiting weeks to get samples analysed in a remote lab. The key issue is how reliable are these on-site measurements? Can sound decisions be made using them, or should we stick to measurements made in the remote lab?
Recent research at University of Sussex has been investigating this issue, by focusing particularly
on the uncertainty in both the on-site and lab-based measurements. The key concept here is that all measurements are wrong to some extent, but the value of the uncertainty tells us the range of concentration within which the true value lies. We applied a range of different on-site techniques to a several different contaminated sites, and compared the results with those from remote labs. There are two main ways of applying on-site analytical techniques in the field. The ‘in situ’ approach leaves the sample material in its original position; for example a portable x-ray fluorescence spectrometer (PXRF – Fig.1) can be placed directly onto an area of bare soil, or a pH electrode placed in a river. The second more common ‘on-site’ approach, is to remove a sample from its position (e.g. in the top soil), prepare it in some way (e.g. mix a soil as far as possible within a polythene bag), and analyse this prepared sample near the original location, or in a local field-lab.
The results for the in situ approach, in an example for As determined by PXRF, showed that the areas of land identified as being contaminated by As above a threshold concentration value, were very similar to those found using the lab measurement. Somewhat unexpectedly, the uncertainty of both the in situ and lab measurements were also similar. The was because the main source of uncertainty was shown to be the sampling, rather than the analytical process, and therefore caused by the small-scale heterogeneity of the As distribution in the soil. The optimal level of uncertainty was calculated, at a level that minimizes the overall potential financial loss. This cost arises from both the taking of the measurements, and from decision errors in the land classification caused by the uncertainty. Neither the lab or the in situ measurements had this optimal level of uncertainty, which is required to achieve full fitness-for-purpose. Both type of measurement would therefore benefit from the taking of composite samples or measurements. This is achieved by taking several sub-samples across each small area, rather than just one sample, to reduce the effect of the small-scale heterogeneity to an acceptable level. The bias the in situ measurements was estimated as -43%, by comparison against the lab measurements across all locations. This was partially due to the moisture and pore spaces included in the soils measured in situ. Once quantified, this bias can be corrected for, to improve the agreement. Using this approach, it was argued that in situ measurements can be not only sufficiently reliable, but they can also be more fit than lab measurements for some purposes, such as hot spot delineation.
For another case study of on-site measurements, with local removal of the samples from two sites, the sampling process again dominated the uncertainty values. For the organic contaminant total petroleum hydrocarbons (TPH), the uncertainty of the lab analysis (U = 54%) was too high to make an effective comparison of the on-site results from the test kit (U = 30%) against them. This study also indentified that the different definitions of terms, like TPH including aromatic but not aliphatic compounds, is another factor limiting the interpretation using such comparisons.
Overall the conclusion was that on-site (including in situ) measurements can have distinct advantages over lab-based measurements, especially in terms of their rapidity and lower overall cost. The consideration of the uncertainty of both types of measurements revealed that both can be very high (e.g. ~50 - 100%), but the reliable interpretation of both types of measurements requires the value of the uncertainty to be estimated for that application. Knowing this information, on-site measurements can be as useful to environmental geochemists as those made in the lab.
Professor Mike Ramsey, University of Sussex.
 Ramsey M.H. and Boon K.A. (2012) Can in situ geochemical measurements be more fit-for-purpose than those made ex situ? Applied Geochemistry 27, 969-976 . http://dx.doi.org/10.1016/j.apgeochem.2011.05.022