Prediction with Geostatistics (in free software) of Aluminum Soil Concentrations in a Forest in Southeast of Ecuador

Geostatistics is a discipline that deals with the statistical analysis of regionalized variables. In this case study, geostatistics is used to determine high values of Aluminum, from a soil sampling done in a forest in the southeast of Ecuador. Free statistical software R and mainly gstat and RGeostats libraries were used. RGeostats is the Geostatistical Package (under R platform) developed by the Geostatistical Team of the Geosciences Research Center of MINES ParisTech. First, exploratory data analysis (EDA) was performed. The second step was the review of a probable presence of drift or trend in the data through the analysis of directional clouds: East and North. Geostatistical tools such as: variogram cloud and variogram surface were used to detect outliers and anisotropy, respectively. An automatic model fitting by Iterative Least Squares was employed to fit the variogram. Finally, Kriging with moving neighborhood was used to get a detailed map of Aluminum soils concentration. The standardized error was reached in the interval [-2.5; +2.5]. The highest values obtained in estimation map belong to the central area. For additional conclusions in relation with toxicity of Aluminum is required to develop a multivariate analysis that includes variables such as pH and average plant growth.