Data statistic principles and methods for regional ecogeochemical assessments

How to establish a relation equation of various chemically partitioned component concentrations for elements in soils and their total concentrations or other soil parameters according to the statistical results of certain numbers of representative samples is one of the problems that need urgent solution in regional ecogeochemical assessments. This paper suggests that we should adhere to the dominant principle, concise principle, direct principle, best fitting, recognizable principle and principle of theory uniformity in the establishment of such regressions, i.e. selecting dominate factors, using concise regression models, taking factors that is directly linked to the results as independent variables, best fitting and using the uniform theory. The validity of the equations must also be tested by selecting the validation dataset and determining the prediction intervals. Examples are given in this paper by performing a statistical anal- ysis of toxic and nutrient elements in soils in the Chengdu economic region according to the aforesaid principles, and the regression relations of the concentrations of various species of elements, especially those of the available species, and the total concentration of elements or other important soil properties. These equations provide a basis for regional ecogeochemical assessments of areal survey data.