Using GIS and statistics to study influences of geology on probability features of surface soil geochemistry in Northern Ireland

The probability features of non-normality and non-lognormality are widely observed in geochemistry due to the influences of multiple factors that are difficult to quantify and model. In Northern Ireland, the pseudo-total concentrations of 14 elements (Ca, Cd, Co, Cr, Cu, Fe, K, Mg, Mn, Na, Ni, P, Pb and Zn) from 6138 topsoils were measured, and GIS mapping showed that the spatial distribution of these data were in line with the spatial distribution of geology in the area. Investigations into the influences of geology on the concentration data and their probability features were carried out using GIS and statistics in this study. The whole raw data sets for each element were positively skewed and none of them followed either normal or lognormal distributions. Logarithmic transformation was found to have “over-transformed” most of the data sets, changing their skewness from positive to negative values. When soil samples were classified by rock type using a GIS overlay function, obvious differences were observed in the chemical concentrations of soils derived from different rock types. Soils in basalt areas displayed the highest concentrations for most elements under study (Ca, Co, Cr, Cu, Fe, Mg, Mn, Na, Ni, P and Zn) but the lowest concentrations for K, while the highest levels for Cd and Pb occurred in the shale areas. Classifying soils by rock type produced more normally distributed data sets, especially for the igneous rock areas. To restrain the influence of soil type and land cover, samples from both gleys and pastures were extracted via a GIS and it was found the data sets then showed generally greater tendencies towards normality. However, many of the data sets would still not pass a test for normality unless the sample size was small (e.g. of the order of a couple of hundreds). Geology, soil type, land cover and sample size all played important roles in determining soil chemical concentrations and their probability features. However, the influences from other factors were still evident. Attempts made in this study show that it remains a challenging task in geochemistry to separate all the factors and to model their influence at the regional scale.