Improving geostatistical predictions of two environmental variables using Bayesian maximum entropy in the Sungun mining site

摘要

In this paper, the spatial distributions and temporal changes of electrical conductivity (EC) and pH in the Sungun mining area (in the East Azarbayjan province, Iran) were assessed. These variables were measured in 2005 in three parts of the mine considered for: the mining pit, waste dump, and tailings dam. A follow-up study was devised in 2016 with a new sampling round, at almost the same locations to examine the environmental status of the study area and its changes during this time interval. First, the general statistical evaluations were conducted. After distribution assessments and spatial variability modeling, the EC and pH were predicted at unsampled locations using three geostatistical methods of kriging, Sequential Gaussian Simulation, and Bayesian Maximum Entropy (BME). BME can also efficiently take the soft information into account. Moreover, the predicted variables and their estimation variances were mapped using these methods. The hazardable zones on these maps were also noted.