The distinction of anthropogenic (man-made) and geogenic (natural) compartments in complex pedogeochemical anomalies

by Andreas Kluge, Bernd Voland, and Uwe Schlenker

Institute of Mineralogy, Freiberg University of Mining and Technology

Brennhausgasse 14, D-09596 Freiberg/Saxony, Germany

submitted to Mineralium Deposita - March 1991

abstract in german language

Introduction

Problems of environmental influences especially the contamination of agricultural land by heavy metal, have increasingly become a subject of investigation. The main points of the discussion are the determination of heavy metal contents in soils, ecotoxical limits, the behavior of heavy metals in foodchains and their biological availability. A factual discussion of these problems and the choice of suitable ways to control the heavy metal metabolism in plants is based on both the knowledge of the dynamics of trace element metabolism and the anthropogenic and geogenic sources of these elements. The trace element metabolism in soils is controlled by numerous parameters. The main sources ("from below") are natural processes forming lithogenetic- and mineralization-based anomalies geochemical provinces. Environmental factors (e.g. changes of pH-value) are superimposed upon the element metabolism in soils, thereby forming another type of pedogeochemical anomalies. Geochemical changes in soils of a industrialized landscape are mostly controlled "from above", where the input of fertilizers, the application of sewage or manure, the influence of long standing waste disposals and the deposition of dust from the atmosphere control the formation of anthropogenic anomalies. Also for the problems of environmental technologies and the geological exploration it is of interest to investigate the causes of such anomalies. An example for the solution of these problems can be given by an investigation of an area in the district of Freiberg/Saxony (fig. 1 and fig. 2). A heavy metal ore deposit is situated in the central part of this area. Furthermore, there are some anomalies of different types (lithogenetic, anthropogenic and environment-controlled). The location and quantitative differentiation of these anomalies with respect to their formation has a practical significance for lowering or controlling the trace element influences in soils. On the other hand a successful method of differentiation of complex anomalies is of great importance for fundamental research on environmental protection or geochemistry. Some possible ways of solving this task are:
  1. Analysis of element speciation (form of chemical bindings): The determination of element speciation is a promising method, because the anthropo- genically emitted elements may have special chemical binding forms. It needs of course special analytical methods and equipments which require a large amount of research and development work.
  2. The investigation of selected soil profiles, which are characteristic and representative for a large area: The aim of this method is the detection of transport currents and the dynamics of trace elements in soils. This relatively expensive method does not allow widespread application.
  3. Univariate statistical comparison of element contents in top and sub-soils: This method is a simplification of method No.2, but it fails in areas with strong interactions of complex parameters. A further method - used the first time in such a case study -is the mathematical differentiation by application of multivariate statistics.

Theoretical possibilities for a mathematical differentiation of anthropogenic and geogenic influences in geochemical anomalies

Data sets produced by geochemists, which contain information about possible geochemical anomalies, are the starting point for mathematical differentiation. How successful the differentiation can be depends on the actual local conditions and the obtained analytical results. In a first step the anomaly must be mathematically delimited from the geochemical background. If anthropogenic and geogenic anomalies exist simultaneously, different cases with increasing levels of difficulty occur and the following is assumed:
  1. Geogenic and anthropogenic anomalies are different in terms of their localities; they are distinguished by their spectrum of anomalous element contents. The statistical differentiation succeeds in producing a two-dimensional presentation of univariate data (e.g. z-scores), mostly by using cluster-points or factor-score maps. The interpretation of the resulting associations can be done on the basis of typical anomalous element spectrums.
  2. There is a local differentiation, but the anomalous element spectrum is similar so that statistical differentiation can be accomplished by the same method. For the interpretation of associations a-priori-knowledge is necessary (ore veins, man-made disposals etc.). Nevertheless the associations are not obvious in all cases.
  3. There are complex, locally superimposed anomalies of different genesis. Such anomalies are characterized by a uniform factor and consequently by the same class of cluster analysis. The explanation of anomaly reasons is possible only when they are characterized by different element associations (RENTZSCH et al. [1985]).
  4. There is a superimposition of complex anomalies with similar geogenic and anthropogenic element associations. This case has happened in the investigated area with simultaneous non-ferrous metals mining and metallurgy, but it occurs also in highly urbanized areas (WOPENKA [1981]).
Possible solutions were found for the cases 1 to 3 Case 4 should be investigated in greater detail. At the same time this case is the common one, whereas the others only represent special cases. A common model is described by the method of element equilibrias: The element content of a sample is considered as a linear combination of the single concentration patterns of single sources with different influences. 

Ci  = concentration of element i in the analyzed sample
mk  = influence of source k
n   = number of samples
xik = concentration of element i  formed by source k
The equation for the special case of a superimposition of anomalies formed by veins and metallurgical emissions could be:

This model drastically simplifies the anthropogenic and geogenic relations. Also, the geogenic and the anthropogenic components could be subdivided into more detailed factors with different influences of the source and different element relations. It cannot be assumed that the element ratio of one source is equal over the whole area and time-independent. Related results were obtained by KLUGE [1985] from the investigation of aerosols. The following reasons can be considered for anthropogenic sources:

Thus, means the element concentration patterns of particular sources become functions of the immission conditions. An investigation of the influence of sources just by solving equation (1) was rejected by WOPENKA [1981], because of the fact that a definite solution is impossible for similar concentration relations of two or more sources. In fact, the same elements are also characteristic for two or more sources, and the use of label elements is not available to assess of the influence of the source. By using the factor analysis it has been attempted to solve this problem in another way. The starting point of this analysis is the information about the causing sources which is obtained from the variance of element concentrations determined from a large number of an amazon spectrum (WOPENKA [1981]). The estimation of the factor analysis (3) leads to an equation similar to the explained equation (1). The single sample is described as a linear combination of factor loading patterns of different source influences (factor value). 

             i=1,2,...,m
             j=1,2,...,m
zij : standardized measuring value for feature i on sample j
ail : factor loading of feature i on factor l
plj : estimated value of factor l on sample j
r   : number of mathematically determined factors (from ÜBERLA [1971])
The standardized values (z-scores) used as target value of (3) represent quantitative values reproducible for the starting data. In spite of the similarity of this estimation with the method of element balances, there is a fundamental difference. Instead of element concentration relations (xik in (1)) the factor loadings ail represent only correlation coefficients. The factor values plj differ from source influences mk first because of their vectorial behavior and secondly because of their relationship to concentration relations. This has the following consequences:
  1. Quantitative conclusions about the element spectrum cannot be drawn by factor analysis.
  2. The qualitative conclusions of the factor analysis concerning the characterization of trace element sources with similar element spectrum are reduced by the required single structure. Only elements with high factor loadings on one factor are stressed, which is characteristic for one, or at most two factors.
  3. If the same elements are characteristic for two factors then both sources show a uniform variance and cannot be distinguished.
  4. Should there be a relation of one (or more) features to two factors as result of factor analysis, the data set has to contain at least two features which correlate weakly with each other but have a high correlation to the common features.
Taking into account the problems of differentiation of geogenic and anthropogenic influences the following conclusion can be drawn: features which contain distinct information about differentiation have to be under additional consideration. In certain circumstances the number of features has to be reduced in order to increase the influence of these features. Possible features are elements and phases with higher relationships to one or the other source, or features containing information about source-related element speciations (e.g. content of humic substances, organic compounds, mineral phases). A further source of information is the different collection of samples from the top- and subsoil. It can be concluded that in spite of strong interactions of the trace element metabolism between the horizons the influence of anthropogenic immissions ("from above") in the topsoil (A-horizon) respectively the influence of geogenic sources ("from below") in the sub-soil (B-horizon) can be described as stronger.

Differentiation of a complex anomaly by factor analysis of selected features

The considered area of investigation around Freiberg/Saxony includes about 580 km². Figure 1 gives topography of this area. About 65 % of this area are used for agriculture, 25 % as forest and 10 % in form of settlements, roads, water streams or industrial area. In the center of this area a hydrothermal Pb-Zn-Ag-ore deposit is situated. This deposit was the object of intensive mining activities for some hundred years. The mining was followed the by corresponding types of metallurgy. Now the towns of Freiberg, Muldenhütten and Halsbrücke are centers of processing of waste materials, local and imported ores (tin ore concentrate, zinc sulphide etc.). These geogenic (ore deposit dispersion zone) and anthropogenic (smelters and other industrial activities) influences are significant reasons of anomalies. Furthermore, other types of anomalies (anthropogenic, lithogenetic and environmen-based) can be discussed. The main attention is directed towards the superimposition in soils of the dispersion zone of the ore deposit and the influence of emissions from smelters arriving on the soil as dust and aerosols. The special problem is the identity of the element spectra of both sources in a qualitative sense. During the environmental research work at the Mining Academy of Freiberg been done an area-covering collection of soil samples from the A- and B-horizon was done. The sample collection followed a rectangular network with 1 km distance between the sampling points. The following total element contents were analyzed: Ag, B, Ba, Be, Co, Cr, Cu, Ga, Mn, Ni, Pb, Sn, Ti, V, Zr and the HNO3-extractioncontent of Pb, Cu, Zn, Mn, Cr. By a general factor analysis of all element contents from the A-horizon the conclusion was drawn that the elements Pb, Ag, Cu, Sn and Zn characterize the complex anthropogenic- geogenic factor of smeltery emissions and mineralization. If the variance of these features also contains information about the differentiation of geogenic and anthropogenic influences, this part will disappear because of its relatively small values in the unspecified single variance. One possibility to use this information could be the extraction of other, more variance-weak factors. Because of the assumption that the necessary information is only contained in the investigated features, a factor analysis has been done with these (logarithmic) features (see table 1). At that, two factors with an eigenvalue > 1 could be extracted. These factors were named in relation to factor I of the factor analysis of the A-horizon as factor AI1 and factor AI2. Factor AI1 has an eigenvalue of 4.51, this is related to a fraction of variance of 56.4 %; factor AI2 has a eigenvalue of 1.31, that means a fraction of variance of 16.5 %. (There are three other factors with an eigenvalue less 1 carrying a fraction of variance of total 20.1 %.) Both variance-strong factors carry the following factor loadings: Factor AI1 is obviously determined by Sn, factor AI2 by Zn. Here occur the most important differences of loadings. The pH-value in factor AI1 has a small negative loading. The interpretation of the areal distribution of factor values (factor maps) indicates a clear association of factors to geogenic or anthropogenic influences respectively. Factor AI1 reflects the influence of smelteries. The highest factor value appears in the site Muldenhütten (lead smelter), the location of the Freiberg smelter (zinc and tin smelter) and inside the Tharandt forest. Factor AI2 is associated with the remaining part of the total anomaly, which is mostly geogenically influenced. All anomalies are related to known ore veins. The anomalous areas are situated side by side and do not cover each other. However, there are overlappings in their peripheral fields. By superimposing the maps of both anomalies on each other, the total area of the complex heavy metal anomaly can be reconstructed (fig. 5). Analyzing the element contents of the anthropogenically or geogenically indicated anomalies, it will be evident that the differences of the relations are the reason for the extraction of these two factors. In table 2 the arithmetic and geometric means of both partial anomalies are represented. The obvious differences in the element content ratios reflect the elements Zn and Sn. While Zn is enriched in the geogenically interpreted partial anomaly, Sn enrichment could represent the anthropogenic impact (of smelters). The mean ratio Sn/Zn in anthropogenically influenced samples is 1.94 : 1 (with a minimum of 0.118 : 1) and 0.043 : 1 in geogenically determined samples (with a maximum of 0.1267 : 1). A distribution-free rank test after Wilcoxen (STORM [1986]) is used for both elements. The hypothesis that both element contents (Zn and Sn) come from the same basic data set has been proved. Concerning the Sn values, the hypothesis failed with a error probability =1 % (proof value: zSn=2.74; test value z =0.01=2.326), in case of Zn even with a error probability of =0.1 %(proof value zZn=-3.48; test value z =0.001=3.090). Consequently the hypothesis is statistically proven that the Sn or Zn content in samples of both anomaly types comes from different basic data sets and therefore from different sources. From this differentiation the following conclusions can be drawn in this step:
  1. The anomaly caused by the smelters is the most powerful anomaly with a fraction of variance of 56.4%.
  2. The geogenically determined anomaly has a larger area than the anthropogenic area (at least 16 geogenically determined anomalous samples as composed to 12 anthropogenic anomalous samples).
  3. The anthropogenically caused anomalies reach up to the area around the Freiberg, Muldenhütten and Halsbrücke smelters and also to the forest area in a distance of 10 km from the smelters (see fig. 3).
The more anthropogenically influenced anomalies have distinctly higher contents of Sn and also some higher contents of Ag and Pb. Thus the conclusion can be made that the elements of these anomalies mostly follow the chain:

tin ore concentrate/lead and noble metal scrap -> smelting -> aerosol -> deposition in the surrounding of the smelters or recombination from aerosols in the forests -> fixation in the soil.

The stronger binding of Sn to the anthropogenic anomalies can also explained in such a way that Sn is a minor element of the ore deposit (only in ore veins containing stannite in the Zn- Sn-Cu succession of the kb-formation and in form of cassiterite in the Sn-W-succession (BAUMANN [1958]). Also, the migration capability of Sn is reduced because of the fixation as cassiterite (SnO2). From the tin smelter Sn is emitted as SnO2. In the dust particles coming from the lead refining plants of Halsbrücke and Freiberg, SnO2 could be extracted as well (VOLAND et al.[1989]). SnO2 settles in the anomalous range and is enriched because of its low mobility in the top-soil. 4. Zinc is generally enriched in the atmosphere and with an atmospheric interference factor of 23 it is determined as obviously anthropogenic (VOLAND [1987]). Additionally the zinc and sulfuric acid production emits ZnO (VOLAND et al.[1987]). In comparison to Pb and Ag the zinc smelter has existed in Freiberg for a short period (direct smelting in Freiberg since 1886, VOLAND [1984]). For the soils above the ore deposit the anthropogenic zinc impact plays a minor role. The higher mobility of zinc from this deposit has a higher level of influence. A high zinc content shows most of all mineralization caused anomalies. In table 3 it is shown how an overweight or a superimposition of anthropogenic and geogenic influences bear upon the concrete measuring values of selected samples. It is evident that connections such as the distance to the emittents or the location of ore veins are reflected in factor values.

Summary

The separation of spheres with natural (geogenic) and artificial (anthropogenic) environmental influence is a fundamental problem of the environmental assessment. A separation of complex anomalies into interpretable partial anomalies is possible using factor analysis of geochemical data. In our application example we were able to distinguish two areas within the Freiberg heavy metal anomaly which either can be traced back to the dispersion halo of the ore deposit or are determined by the sedimentation of smelter emission dusts. Using the results of the factor analysis a qualitative characterization of both sources of trace elements could be given, moreover both influences could be plotted in map form. By this way, it will be possible to estimate or to emulate the scope of efficiency of environmental protection measures in the smelting plants. The results, however, simultaneously show areas in which changes in the emission of the smelting plants will cause no or only slight alterations of the trace element status of the soil.

Literature:

BAUMANN, L.: Tektonik und Genesis der Erzlagerstätte von Freiberg (Zentralteil). - Freib. Forsch.h. R. C 46, Berlin (1958)

KLUGE, A.: Entwicklung von Rechnerprogrammen zur mathematischen Behandlung geochemischer Daten.- Freiberg: Bergakademie, Skt. Geow., Studienarbeit, 1988.

RENTZSCH, J.; SCHULZE, W.; BIRKE, M.; MÜLLER, H.; EICHBERG, M.; PILLATZKE, L.: Geochemische Prospektion in anthropogen beeinflußten Gebieten. - In: Z. angew. Geol., Berlin: 31 (1985) 12.- S. 290-295.

STORM, R.: Wahrscheinlichkeitsrechnung, mathematische Statistik und statistische Qualitätskontrolle.- 8.verb. Aufl. - Leipzig: VEB Fachbuchverlag 1986.

ÜBERLA, K.: Faktorenanalyse: Eine systematische Einführung für Psychologen, Mediziner, Wirtschafts- und Sozialwissenschaftler. -2. Auflage - Berlin - Heidelberg - New York: Springer Verlag 1971.

VOLAND, B.: Schwermetallemissionen der Hüttenindustrie.- In: Spurenelemente in der Umwelt/ Fiedler H. J. und Rösler H.J. - Jena: VEB Gustav Fischer Verlag 1987, S.130-135.

VOLAND, B.: Charakter und Genese anthropogener Veränderungen der Geochemie der Landschaft: Ein Beitrag zur Umweltgeochemie.- Freiberg: Bergakademie, Diss. B 1984.

VOLAND, B.; SCHLENKER, U.; KLUGE, A.: Multivariate Methoden in der Umweltgeochemie.- Freiberg, Vortrag auf dem XL. Berg- u. Hüttenmänn. Tag Freiberg 1989.

WAGENBRETH, O.; WÄCHTLER, E.: Der Freiberger Bergbau - Technische Denkmale und Geschichte.- VEB Deutscher Verlag für Grundstoffindustrie.- 2. Auflage, Leipzig 1988

WOPENKA, B.: Mathematische Methoden zur Erfassung atmosphärischer Aerosolquellen.- Wien: TU, IAC 1981.

kluge@mineral.tu-freiberg.de