Preface of special issue on probabilistic and soft computing methods for engineering geology

Probabilistic and soft computing methods are increasingly adopted in various fields of engineering geology. In the past decades, extensive original and innovative studies of various engineering geology problems have been conducted using probabilistic and soft computing methods, and substantial new theories, methods, insights, experiences, and data have been attained. This special issue is aimed at providing a timely overview on the current state-of-art of probabilistic and soft computing methods in engineering geology. Through peer-review processes, we have secured 16 papers for this special issue, which may be divided into four categories: (1) slope stability and landslide hazard assessment; (2) development of empiricalmodels for engineering geology applications; (3) site characterization considering uncertainties; and (4) calibration of geotechnical and geological models. We envision that this special issue will provide guidance on how probabilistic and soft computing methods can benefit engineering geology research and practice, point out directions for future development, and facilitate and promote the use of such methods in engineering geology. The first part of the special issue covers the topics of slope stability and landslide hazard assessment. For probabilistic slope stability analysis, one important practical challenge is the long computation time required in the search for the most critical slip surface in each random realization. Various response surface methods (RSMs) have been proposed to replace the original mechanical models in an effort to reduce the computation time. Li et al. compared the effectiveness of these RSMs in four types of slope reliability problems, from which an appropriate RSM is suggested for each type of slope reliability problem. Chen et al. suggested a multi-hazard cell-based risk assessment platform for regional rainfall-induced slope failures and debris flows, and the method is applied to analyze the risk posed to a highway close to the epicenter of the 2008 Wenchuan earthquake. Massimi et al. showed how the artificial adaptive system could be used to obtain the displacement pattern of an extremely slow moving landslide when the measurement error is substantial and when there is no agreement between displacementsmeasured fromdifferent techniques in a redundant measurement system. Arzu Erener et al. assessed the performance of the GIS-based multi-criteria decision analysis, logistic regression and association rule mining in the determination of landslide susceptibility map of Savsat, Artvin province in Turkey. They found that the association rule mining method provided a higher overall prediction performance, whereas the percent of landslides being correctly identified was higher for the logistic regression method. Park et al. proposed a GIS-based methodology to predict instability in spatially distributed slope faces based on kinematic analysis, in which the uncertainty in the orientation of a discontinuity is explicitly considered through Monte Carlo simulation.