engleski

Application of multivariate geostatistics to describe soil pollution by heavy metals in Northwestern Croatia

Environment protection depends on how well land use and remediation practices are fitted to variable soil conditions. The assessment of soil contamination has then become a crucial issue. Multivariate geostatistics uses the information coming from relationships among variables in order to improve estimation precision and to separate the different causes of variation. Scale-dependent correlation structure of some variables was investigated by means of Factorial Kriging Analyses which consists in modelling the coregionalization of a set of variables, cokriging specific factors and mapping them. The present study was conducted in an area of about 2000 km2, located in the Zagreb region in Northwestern Croatia. The land is widely heterogeneous with respect to lithology, geomorphology, pedological features and land use. The topsoil was sampled according to a square grid with a variable mesh of 2 or 1 km in the areas of higher urbanization. Basic soil chemical properties were determined and metal contents were extracted by aqua regia digestion. The Linear Model of Coregionalisation was fitted using three nested spatial structures including: nugget effect and two spherical models with ranges of 10 km and 35 km. The Factor Kriging Analysis allowed to define two regionalized factors at different spatial scales. On the first factor at shorter range Zn, Pb, Cd, Cu and Ni loaded mainly and positively, indicating different sources of anthropogenic contaminants. On the contrary, most variation of the first longer range factor was determined by lithology and parent material composition. The methodology used has showed to be a useful tool to separate geological and anthropic inputs and identify common pollution sources

soil pollution; heavy metals; spatial variability; geostatistics; factor kriging.

nije evidentirano