Descriptive statistics and stationarity in environmental geochemical variables

ABSTRACT The aim of this work was to analyze geochemical data in order to check their stationarity and to correlate the statistical normality using the ordinary kriging technique. The ordinary kriging technique was chosen as the geostatistical method applied to work because such technique is advised for studies in areas where there are data with variables that might present spatial dependence, like the geochemical variables, and also because it is indicated for data presenting stationarity. The methodology used for this research involved, besides literature review, data collection of trace metals (Cu, Zn, Mn, Fe, Cr and Pb) that were partially extracted from surface samples (0 to 10 cm) of soils and sediments collected in the field. We also determined the values of pH, salinity, total nitrogen, phosphorus, organic matter and particle size. Statistical analyzes, semivariogram development, ordinary kriging use and, lastly, cross validation were performed to measure the uncertainty of the previous measurement of data. It was found, in this work, by means of the variograms that although data were ordinary, they showed stationarity. In addition, the parameter of descriptive statistics that mostly correlates directly with the ordinary kriging is variance.