Statistical and geoestatistical characterization of heavy metal concentrations in a contaminated area taking into account soil map units

Soil map units were taken into account in geostatistical prediction of heavy metals concentrations in a contaminated site of Fuyang Valley, Zhejiang Province, China. To compare the spatial variabilities of Cu, Zn, Pb and Cd concentrations in different soil map unit or combination of soil map units, a total of 94 samples distributed in three map units (Yellowish-red soil, Gravelly-yellowish-red soil and Gritty-yellowish-red soil) were collected in this area. The values of Cu, Zn, Pb and Cd concentrations were natural logarithm transformed to fit normal distribution. The variations of four heavy metal concentrations in Gravelly-yellowish-red soil was similar to that in Gritty-yellowish-red soil, but they were different to that in Yellowish-red soil. Then, the 94 soil samples were grouped into two classes (samples of Yellowish-red soil, and samples of Gravelly-yellowish-red soil or Gritty-yellowish-red soil). Semivariogram analysis revealed that all the four heavy metal concentrations in the study area showed moderate to strong spatial dependency, and local spatial variability (i.e. the variance between soil map units) played an important role in their spatial prediction. The spatial distribution maps of the four heavy metals were drawn by using only the measured data and by using the measured data plus taking into account soil map units, respectively. The results showed that spatial prediction by taking into account soil map units could reveal the huge spatial variability of the heavy metals, much better than the prediction without using soil map information in this contaminated site.