PDF version

COMPARISON OF ELEMENTS IN ALL SAMPLE MEDIA, GENERAL COMMENTS AND CONCLUSIONS

by W. De Vos¹, A. Demetriades², K. Marsina³, R.T. Ottesen⁴, S. Reeder⁵, S. Pirc⁶, R. Salminen⁷ and T. Tarvainen⁷

¹Geological Survey of Belgium, Brussels, Belgium;
²Institute of Geology and Mineral Exploration, Athens, Greece;
³Geological Survey of Slovak Republic, Bratislava, Slovak Republic;
⁴Geological Survey of Norway, Trondheim, Norway;
⁵British Geological Survey, Keyworth, Nottingham, United Kingdom;
⁶ Geology Department, University of Ljubljana, Ljubljana, Slovenia;
⁷Geological Survey of Finland, Espoo, Finland.

A comparison between the geochemical distribution patterns of the different sample media was discussed at a special meeting at the Geological Survey of Finland in December 2004. This meeting also led to the final approval of most of the maps, and the rejection of some of them on the basis of inconsistencies, country boundary effects or other doubtful results, due to possible sampling or analytical errors.

For several elements, especially those having an environmental importance, conclusions have been formulated regarding possible anthropogenic patterns detected on the basis of this comparison of geochemical distribution maps. Indeed, it can generally be admitted that subsoil reflects a distribution pattern that is closer to a natural situation, whereas topsoil would be more influenced by human activities and natural exogenic effects (oceanic, eolic, meteorological inputs), and this is even more so in stream sediment and floodplain sediment. Differences in distribution patterns between these sample media should always raise suspicion of human influence. However, the sampling design was intended to maximise the natural signal, especially in soil, by selecting forest soil whenever possible, and always far removed from human influence.

Stream water for most elements shows a dispersion pattern generally different from the solid sample media, and seldom directly related to the underlying geology. The geochemistry of waters is decisively controlled by acidity, oxidation potential and organic matter concentrations and these in turn of climate, vegetation, topography and other exogenic factors. The humus patterns are difficult to interpret at the European scale, as the nature of the humus sample depends strongly on climate, topography and vegetation, as well as on the amount of mineral material that is often mixed with the humus sample when the humus layer is thin.

Similarities between the four solid sample media (subsoil, topsoil, stream sediment, floodplain sediment) are further investigated by factor analysis, which is presented in Annex 5 of this volume (Batista et al. 2006).

Differences between geochemical distribution patterns of the same element or determinand in various sample materials – subsoil, topsoil, stream water, stream sediment and floodplain sediment, are regarded as representing different and complementary aspects of the elemental dispersion in the near-surface environment. They can be assessed by statistical parameters describing the median and variation – is the element in one medium “higher”, or “lower” than in the other medium, or perhaps “equal”, and by studying the elemental maps, where the spatial patterns of “highs” and “lows” are visualised. Important conclusions on the relevant part of the geochemical cycle and processes of elemental migration can be deduced from these differences. It is important to understand the reasons behind the differences between sample media. All information on the elemental distributions is obtained from the samples that are collected, analysed and described statistically and graphically on geochemical maps, and interpreted.

The discussion on the causes of natural variability could start with the mineralogical composition and genesis of soil, sediment and water. Differences are indeed considerable. While soil is a product of pedogenic processes where the rock is transformed in situ by weathering into soil, in addition other processes take place before the formation of sediment, i.e., erosion (destruction of the weathered rock, soil and even previously deposited sediment), denudation (removal of this material from its place of origin), transport (carrying it away, mostly by running stream or river water), and deposition of the eroded and transported material on the stream bed or on floodplains. The “subsoil” – soil from deeper horizons, and “topsoil”, from the upper soil horizon, differ in composition owing to soil formation processes, and also due to input of some substances from above, often from the atmosphere, either naturally or by human activity. The material of the “sediment” differs markedly from that of soil, because of differential erosion, denudation, transport and deposition according to the nature and chemistry of the weathered material. Topography definitely influences the properties of stream sediment. A major difference between stream and floodplain sediment is that stream sediment is deposited as relatively coarse-grained material on the active stream bed, whereas floodplain sediment is deposited as very fine grained material from exceptionally high water that floods the plain beyond the river bed. The difference in grain size between them results in a geochemical difference. Both stream and floodplain sediment, represent the geochemistry of the upstream drainage basin from the point of sampling, including both natural and human produced materials.

Stream water constitutes a sample material on its own, with a composition reflecting the soluble, leached out part of the rocks in the upstream drainage basin where water percolates. It also reflects any human-produced inputs in solid, liquid and gaseous form. Water may in addition be subjected to drastic dilution by rainfall. Its composition is controlled by hydrochemical processes that either dissolve or cause precipitation of elements. This makes water a much more diverse, and more delicate sample medium, compared to solid media. In spite of the high variability of its composition, stream water presents continent-wide geochemical patterns that confirm its importance in an effort, such as the one carried out for the present FOREGS Geochemical Atlas of Europe. Stream water does not reflect the geochemical composition of the bedrock to the same degree as the solid sample media do, since it is susceptible to a whole range of exogenic influences – weather and climate, especially rainfall or precipitation generally, topography, distance from the ocean, vegetation, and also human-induced inputs.

The subsoil and topsoil samples represent material weathered in situ from underlying rocks, supplemented with matter introduced from above, by deposition from the atmosphere and also hydrosphere, or by human activities. The geological history includes various gradations in weathering, owing to past climates, earlier glaciations, etc.

The diversity of each sample medium is closely related to certain inherent characteristics. A “soil” sample is essentially a point sample, representative of a very restricted area that can be measured in square metres. A “stream sediment” or “floodplain sediment” sample, however, is a surface composite sample, because it contains material derived by denudation from the entire catchment basin above the sampling site. It is materially representative of the catchment basin, indicative of its average geochemical composition. The area of influence can be measured in tens of square kilometres for stream sediment, and hundreds of square kilometres for floodplain sediment. The “stream water” sample is in fact a volume sample, taking into account the groundwater input into a stream. However, most usually it is regarded as a surface sample, considering the areal character of a drainage basin. “Soil” samples are, consequently, expected to express the narrow local variability, i.e., the composition of neighbouring samples may differ considerably one from the other, including sometimes high extreme (anomalous) values; most of this could be considered noise in regional geochemical studies. “Stream sediment” and “stream water” samples express the wider regional scale variability, but the composition of neighbouring samples is not expected to differ very much. High extreme values are lower and less frequent – owing to the averaging effect of mixing material from the entire catchment basin. “Floodplain sediment” samples express the widest regional “continental” variability between the very large drainage areas. The averaging effect of material within these large catchment basins results in a limited number of extreme values, either high or low, compared to stream sediment or soil, collected from a smaller drainage basin within the same larger catchment basin.

Besides the natural reasons for differences between the geochemical maps of various sample media, several other factors associated with the research methodology play a part. Any geochemical mapping project is of probabilistic character, and its precision depends on the sample density. The FOREGS geochemical mapping of Europe is of very wide spacing, since its main objective is to disclose the continental scale geochemical variability. Hence, its precision, with respect to local variability, is low. Furthermore, owing to the very wide sample spacing, it can miss geological bodies of many tens of kilometres across, and pollution of moderate areal extent caused by human activities.

In the FOREGS geochemical mapping project the same sample materials were collected and prepared according to the same schemes (Salminen, Tarvainen et al. 1998). Then for each sample material, the same suite of determinands were analysed with the same analytical methods and in the same laboratories. If this procedure had not been followed, large discrepancies in results would have arisen owing to differences in sampling, sample preparation and the performance of the laboratories with different analytical methods, instruments, etc. Sample preparation and analysis were strictly controlled, and whole datasets for certain elements were eliminated due to lack of reliability whenever necessary. But, laboratory analysis still remains an important source of error –for differences between the sample media. Another source in laboratory performance is different physical properties of various media – grain size, distribution of fine fractions, organic matter, etc. that necessarily differ between soil and the two varieties of sediments, stream and floodplain.

Differences in results can occur also due to unequal numbers of samples in different sample materials owing to omissions in collecting, analysing or statistical considering certain samples. The reasons might be accidents during sampling, handling and analysis of samples, determinations below the detection limit and removal of samples with extreme values (outliers) of certain determinands.

An important source of differences between the geochemical distribution maps of the various sample media are (a) the ranges for presentation of concentration patterns with colours, and (b) the variable size circles again representing a range of values. In both cases the concentration scales used are relative and not absolute. For every distribution the lower three blue shades represent the lowest 25% of data, the shades from pale blue, green to orange display the intermediate 50% of data, and the three red shades the highest 25% of data. The deep shades represent the two extreme 5% of data, the deep blue the lowest 5%, and the deepest red the upper 5% of data. The purpose of the comparison is to establish whether the element or determinand in the compared other sample medium map is “lower”, “equal”, or “higher”. This can be estimated by comparing the areas of the same colour, but this is correct only if the element or determinand concentrations are “equal” on the two maps. For an objective comparison, the surfaces delimited by absolute boundaries, expressed in concentration units, must be compared, and not the relative boundaries, as they are used on the present maps. Therefore, the comparison made is not absolute, but relative.

Acknowledgements

Ms. Eeva Teräsvuori from Geological Survey of Finland prepared the box-plot figures.

PDF version