Stationarity for the non-normal distribution of regionalized variables

Based on Johnson distribution curves and the theory of product process quality accuracy control, a method to transform non-normal distribution data was proposed from the view of regionalized variable distribution, and the transformation flow was also designed. With SPSS and Surpac mining software as tools, the applied efficiency of traditional data and Johnson transform methods was analyzed through several cases. It is proved that by using the transformed data of the proposed method, the statistical analysis, fitting and validation of the variation function model, and Kriging estimation could meet the need of stationary, the estimation error can be minimized, and the precision of prediction can be improved.