Quality control of geochemical data
There are numerous sources of error that may negatively affect data results. If your aim is to present reliable data, awareness of possible sources of error must be taken into account at the start.
During field work, it is important to be aware of how potential contaminates (pollutants) can effect a sample collected in the field. At NGU we avoid using shovels and other tools that might contain trace samples of the material we are mapping and similarly, we carefully examine sample packaging prior to go out in the field.
There have been cases where traces of a gold wedding ring were found in the sample, which meant that the entire set of results had to be discarded. Hence, all rings must be removed during field work and/or gloves should be worn. Smoking can also contaminate samples, and therefore prohibited. In addition, you have to make sure that the material collected from a previous site does not contaminate an subsequent sample, as could traces of soil residue on a shovel. In the field, we regularly collect duplicate samples. A duplicate sample is also called 'double check' and is always taken in the same locality, and about 2-10 metres away from original sample. In our survey we make sure to collect a minimum of 20 duplicate samples when doing field work. By comparing the results from the original samples with the duplicate sample can assess variation in the geochemical landscape we are accessing, which can influence of the accuracy of the results.
When the samples come in from the fields, they are in usually dried and screened. At NGU, we use the nylon screen to avoid transferring metals to the sample. In addition there is a risk of mixing samples. If a sample repackaged, it is again important to be aware of how substances the new packaging can influence new sample (risk of contamination). During the sample preparation, we collect replicate the sampling (field duplicate).
Before the samples are sent to the lab, they must put those in a randomized order. In this way, inevitable bias will be distributed randomly throughout the test series, so that the errors are also spread out randomly on the map, thereby avoiding false anomalies. Standardsample (sample one knows the contents of) extends beyond the randomiseringen and be set on a regular basis in the entire test series. These samples will provide answers on how a lot of the instrument varies across in the analysis, and can uncover unwanted breach or time trends from the analysis (see figAnalyseforløp). Our experience is that this need not be certified material, but must be homogeneous and well documented (analyzed many times).
- Andersson et. al., 2012. PCB contamination from sampling and packaging. Applied Geochemistry 27, pp. 146-150. doi:10.1016/j.apgeochem.2011.09.026
- Reimann et. al., 2009. The EuroGeoSurveys geochemical mapping of agricultural and grazing land soils project (GEMAS) - Evaluation of quality control results of aqua regia extraction analysis. NGU Report 2009.049 http://www.ngu.no/upload/Publikasjoner/Rapporter/2009/2009_049.pdf
- Reimann et. al., 2011. The EuroGeoSurveys geochemical mapping of agricultural and grazing land soils project (GEMAS) – Evaluation of quality control results of total C and S, total organic carbon (TOC), cation exchange capacity (CEC), XRF, pH, and particle size distribution (PSD) analysis. NGU Report 2011.043. http://www.ngu.no/upload/Publikasjoner/Rapporter/2011/2011_043.pdf
- Reimann et. al., 2012. The Eurogeosurveys geochemical mapping of agricultural and grazing land soils project (GEMAS) - Evaluation of quality control results of particle size estimation by MIR prediction, Pb-isotope and MMI-extraction analysis. NGU Report 2012.051. http://www.ngu.no/upload/Publikasjoner/Rapporter/2012/2012_051.pdf