Spatial Estimation of Soil Hydraulic Conductivity Using Multivariate Geostatistical Model Case Study: Tabriz Subway Domain

Considering the importance of soil permeability in Civil Engineering and its impact on constructed buildings, the precise study and accurate estimation of the factor in site have a significant role in accurate geotechnical studies. However, usually due to technical and economical problems, drilling boreholes in the site is not feasible and it is necessary to produce reasonable data with current data of boreholes. Therefore, in this research, spatial distribution of soil permeability in Tabriz Subway Domain was estimated by the use of geostatistics tools. The monthly piezometric heads of 21 boreholes obtained from Water Organization of Eastern Azarbaijan were used. In this way, the data of 16 boreholes were used for estimation and the rest 5 boreholes data for the crossvalidation purposes. Generally, the geostatistical tools include two advantages, one is its unbaisness property and second it is possible to compute the estimation variances. These make it possible to verify the performance of the model at each modeling step. Directional variograms showed non-isotropic condition of both soil hydraulic conductivity and ground elevation. It should be noted that to eliminate the dimension dependency of the models in the co-kriging, the values of the data were normalized prior to impose into the model. The study indicates that spherical variogram model is suitable for the geostatistical modeling. Also, the Cokriging system which employs the ground elevation as the secondary variable leads to more accurate results than the Kriging model which is a univariate estimation system.