IAMG 2001--Cancún Technical Program--Session C

Home Page Technical Program Session C Intro Session C Papers

Statistical Treatment of Soil Geochemical Data From a Classical Karstic Polje--An Example from Sinj Polje, Croatia by Esad Prohic*, University of Zagreb, Faculty of Natural Sciences and Mathematics, Croatia, Slobodan Miko, Institute of Geology, Croatia, and Zoran Peh, Institute of Geology, Croatia Estimation of the anthropogenic over natural contribution in the chemical content of soils and sediments is very important but not an easy task. The normalization based on either aluminum or iron, as referent conservative elements having supposed natural distributions, is an useful tool to solve this problem. However, since several independent factors can influence the normalization procedure, which if not taken into account could lead to completely wrong conclusions about the origin of particular elevated concentrations of certain elements, a great deal of caution should be exercised during application of the normalization procedure. Analytical data from 95 soil samples from Sinjsko polje enabled us to critically apply this procedure. Linear regression analysis, evaluation of enrichment factors and trend surface analysis with construction of appropriate maps were used to test the procedure. Presuming an existing linear dependence between the conservative element and the heavy metal it is possible with the use of linear regression to simultaneously define the heavy metal geochemical background (baseline) and to isolate natural and/or anthropogenic outliers (anomalies). Coupled with geographic plotting facilities this type of outlier-screening is used to locate areas of man made pollution. In addition, the factor analysis was performed on 96 samples collected in a regular grid 1 x 1 km which covered the entire area of the Sinjsko Polje valley. Another factor analysis was performed in a regular grid 5 x 5 km that comprised 41 sample in the karstic surroundings. In this manner two factor model were created and closely inspected for possible differences due to the peculiar properties arising from two geochemically different environments. From within the polje we were able to evaluate three factors while four factors are judged to be significant for the explanation of the geochemical structure of data assessed in the regular 5 x 5 km grid (Tab. 2). This grid encompasses the area of Sinjsko Polje and its surroundings together Statistical treatment of selected elements also included the calculation of means, standard deviations from the mean, and other estimates of basic statistic parameters. Evaluation of enrichment factors reveals dependence of this parameter upon four main factors: the choice of referent element; the choice of referent material; the homogeneity of referent element distribution in the study area; and correlation of particular element with referent element. Examination of trends was used together with other statistical procedures as a common procedure in the analysis of geochemical maps. Its purpose is to segregate the underlying trend that possibly pervade the study area, from the local variations, that is, the "noise" or "background". Ba/Sr index reveals striking differences in drainage patterns between polje itself and its surrounding. Statistical analysis, as suggested by the results in this work, would be more properly used over the wider areas in the karstic terrain, with different geological, geomorphological and hydrogeological characteristics, rather than being focused on the relatively small phenomena such as poljes, with many local geologic, hydrogeological and geochemical constraints.