Data analytics in geotechnical and geological engineering

Data analytics is indispensable in geotechnical and geological engineering, particularly the analysis of geo-data, which is often spatially varying or even temporally varying. With the fast development of sensing and digitalisation technologies, recent years have seen an unprecedented growth of data, also known as big data, with salient characteristics of volume, velocity, variety, and veracity (e.g. uncertainty and incompleteness). The surge of big data accompanies a rapid development of data analytic methods that target to deal, in a computationally efficient manner, with a large volume of uncertain and incomplete (or even sparse) data in a variety of forms for extracting knowledge and value from the data and decision-making. There is an on-going paradigm shift from traditional physics-based models to data-driven models in digital intelligence and digital economy. Rather than relying on physics, or insights into the mechanisms concerned, to develop models and hoping that the data in hand fit the models, a data-driven model is featured by its adaptability to data, and the model automatically adapts itself to fit the data in hand. Data-driven models are probably most beneficial when the physical insights, or concerning mechanisms, are unclear or too complicated to model quantitively. Some long-lasting challenges in geotechnical or geological engineering fall within this category, such as developing high-resolution three-dimensional (3D) subsurface geological models from sparse site investigation data.

There have been exciting developments in data analytics in geotechnical and geological engineering recently. This special issue, entitled “Data Analytics in Geotechnical and Geological Engineering”, aims to report the state-of-the-art developments in data analytic methods, or data-driven methods, in geotechnical and geological engineering. The special issue appears as Volume 16, Issue 1 of Georisk, and it contains ten papers from the mainland China, Croatia, Germany, Hong Kong, Japan, Norway, Singapore, Taiwan, the U.K., and the U.S.A. Various topics are covered in this special issue, including Bayesian learning of unconfined compressive strength of rock, machine learning of geological details from borehole logs for the development of the high-resolution subsurface geological profile, determination of optimal sampling locations using Gaussian Process Regression (GPR), performance-based assessment of soil slope displacement under seismic ground motions, Bayesian learning of landslide runout distance from sparse data in a regional scale, multivariate model for sparse and incomplete 3D soil data, generalised additive model for predicting geological formation interfaces, non-parametric representation and simulation of spatiotemporally varying geo-data, physics-informed neural networks for unsaturated groundwater flow problems, and Gaussian mixture model for estimating debris-flow exceedance probability.

Last but not least, we would like to thank all the authors for contributing the papers, all the reviewers for reviewing the manuscripts, and the Georisk editor-in-chief (Professor Limin Zhang) and editorial office staff for their generous support. Without their contributions and efforts, this special issue would not have been possible. All papers in this special issue will be presented in a special session of the 4th International Conference on Information Technology in Geo-Engineering (4ICITG) which will be held in Singapore in July 2022. Our gratitude also goes to Professor Jian Chu, the 4ICITG conference chair, for this arrangement.