Digital geoscience applications
Digital innovations have the potential to revolutionise the applied geosciences. A recent thematic collection entitled Digitization and digitalization in engineering geology and hydrogeology touches on various elements of this transformation, including the potential for ‘big data’ and the use of tools such as artificial intelligence and machine learning.
This open collection currently contains four papers. A study by Alexander Hall and colleagues at the British Geological Survey, Edinburgh, presents a ‘proof of concept’ method for automating sample comparison based on thin section analysis and comparison with existing sample images in an image library. They train a neural network to analyse the pore-space geometry in sandstones and assist in the evaluation of grain-scale similarity, with a potential application being to source a suitable match for stone required for repairs to heritage buildings. While the trained model identified the correct source for samples with an accuracy of only 48%, this prototype method forms a benchmark for further development and refinement of such models.
A second paper highlights the barriers to digitisation in rock engineering due in part to a lack of standardisation in rock classification systems. Beverly Yang at the University of British Columbia, Canada, and co-authors review the definitions of and limitations to these systems and discuss how we can pave a way towards greater use of applications such as machine learning and digital acquisition systems in rock engineering problems in the future, including the challenge of capturing failure mechanisms in automated forms of rock mass characterisation.
In a similar vein, Charlotte Gilder at the University of Bristol, UK, and colleagues highlight the increasing importance of data sharing and use of digital tools in the effective delivery and management of large civil engineering projects. They use the results from a survey and various interviews to better understand current opinions, working practises and potential barriers to data sharing for ground investigations and geotechnical engineering in the UK. Currently, ground investigation data are only shared for large infrastructure projects and the researchers suggest that current data sharing practises are directly linked to drivers of risk relating to geotechnical aspects of a project.
The final paper, which comes from Matthew Crawford at the Kentucky Geological Survey, USA, and co-authors, uses high-resolution LiDAR (Light Detection and Ranging)-derived datasets from a digital elevation model and detailed landslide records for Magoffin County, Kentucky to investigate connections between landslide occurrence and slope morphology. They present a combined machine-learning and statistical method that can identify important geomorphic variables (further evaluated using a logistic regression to determine landslide probability occurrence) and thereby provide a useful approach to geomorphic-based landslide susceptibility mapping.
Together these four papers convey the message that digital innovation in the fields of engineering geology and hydrogeology is rapidly developing. The novel and interesting applications discussed will stimulate further research and be key drivers for development and innovation in this exciting field.
DETAILS: J. Eng. Geol. Hydrogeol. thematic collection (open)
55, qjegh2021-039; https://doi.org/10.1144/qjegh2021-039
55, qjegh2020-183; https://doi.org/10.1144/qjegh2020-183
54, qjegh2020-177; https://doi.org/10.1144/qjegh2020-177
54, qjegh2019-138; https://doi.org/10.1144/qjegh2019-138