Monitoring soil parametries in industry areas frequently involves mapping techniques that assist the researcher and/or the decision-maker to describe and quantify the soil pollution at locations where no measurements are available. The preparation of these soil pollution maps is a complex task, which is only feasible if a spatial correlation of the variable of interest is identified. Furthermore, the spatial correlation may not only change in time and space, but also according to the soil pollution levels. To illustrate this point, this paper investigates the fractal dimension of the spatial correlation of different levels of soil pollutions concentrations [As, Pb, Cu. Zn] in the industry area of Thai Nguyen in Viet Nam. As a result, maps defining areas with high and low probability of exceeding a specific concentration threshold will have an uncertainty that is related to the different soil pollutions levels.