Multivariate geostatistical modelling of geotechnical characteristics of the alluvial deposits in West Thessaly, Greece.

Land subsidence due to ground-water overexploitation in the Western Thessaly basin resulted in the manifestation of ground deformations. In order to investigate the mechanism of this phenomenon, borehole data regarding geomechanical properties of soil were examined. Based on these data, the spatial variability of soil compression index (Cc) was evaluated in relation to other geotechnical parameters, by means of multivariate geostatistics. Multivariate statistics and geostatistics were combined and compared to evaluate the estimation methods of Cc. Principal component analysis (PCA) was applied in order to highlight the relationships between the geotechnical parameters. Through cross-validation, multiple linear regression, kriging and cokriging were tested as estimators of the Cc. The results show that spatial correlation of the target variable is less important than its correlation to the auxiliary variables. Cross-validation demonstrates that kriging with two PCA factors and application of inverse transform gives the best results for Cc estimates. [ABSTRACT FROM AUTHOR]