Why use virtual outcrops?

(and some things to consider when making them)

Photogrammetry is not a new method. Calculation of the distance between objects in a photograph is a technique that has been around since the mid-19 th century - nearly as long as modern photography itself. Digital photogrammetry, or Structure-from- Motion, is the modern equivalent: a technique that can reconstruct, in detail, the relative location of millions of features, from overlapping images. 3D reconstruction of outcrops by digital photogrammetry is increasingly popular in geological research but the jury is still out as to how useful this technique really is. Many geoscientists fall into the: ‘it looks pretty, but what is the purpose?’ camp, while proponents see virtual outcrops as heralding a ‘virtual geoscience revolution’. Sceptics tend to be concerned about ‘Sofa Geologists’ and the apparent threat to traditional fieldwork by ‘moving further and further away from reality’.

Virtual outcrops are not supposed to replace field geology - simply they provide a reconstruction of geological features in digital form. Much like seismic lines, field photographs, or even geological sketches, they are a way to record and catalogue the physical world. They are about adding to the arsenal of available methods and techniques for geological analysis, to improve understanding. They will not replace fieldwork. As with other methods, they are a supplement to having your feet on the ground and your nose to the rocks. For those unable to access field sites, virtual outcrops provide the next best thing – great for teaching and diversity in education (see www.e-rock.org).

When doing outcrop-based research, fieldwork time can often be limited because of expense or logistics. Virtual outcrops allow you to examine and re-examine your outcrops in digital form when back from the field. They can provide a wealth of data, ripe for harvesting: digitised features and extracted geometric attributes can be analysed, projected, tweaked and manipulated, all in the digital environment.

The validity of these measurements and observations is clearly dependant on the accuracy and resolution of the reconstruction (Cawood et al., 2017). Virtual outcrops, no matter how good, do not reproduce the detail you can observe directly. Accuracy and precision of the reconstruction, suitable mesh resolution and orientation and scaling of the virtual outcrop are all of critical importance for robust geological interpretation. If you can acquire your own images and process models yourself, then you should. In this way, you will know if the digital object you are interpreting or measuring is an accurate representation of the real world, and you’ll have an idea of what the inaccuracies are.

Virtual outcrops have limitations and are not the whole answer, but they can definitely be used to enhance our geological experiences. They provide unique viewpoints, opportunity to revisit our favourite outcrops, and can provide a plethora of data. The future of outcrop geology is both digital and field based.

Tips for virtual outcrop creation:

  • Know what you want the virtual outcrop for – this will determine resolution, size etc.

  • You can use almost any camera, but the better your camera sensor and lens, the better chance you have of generating good, clear results.

  • Take overlapping photos orthogonal to the surface you’re interested in – anything less than 60% overlap between images will result in model holes.

  • Aim to use 100-200 images per virtual outcrop to start. Any more and you’ll likely need a powerful computer with large amounts of memory.

  • Work at a sensible distance from the outcrop or sample. Too far and you lose resolution; too close and it’s difficult to get the required overlap. Balance desired resolution with desired coverage.

  • Use ground control points (preferably with a very precise GPS) – these help for georeferencing and let you make sure that everything is where it should be.

  • Use the built-in tools within photogrammetry software for determining reconstruction accuracy – consider discarding data that fall below an accuracy threshold.

  • Ground-truth your virtual data by collecting data by hand in the field – measure orientations of features or distances between known points – then compare to your model data.

References

Cawood, A.J., Bond, C.E., Howell, J.A., Butler, R.W. and Totake, Y. (2017). LiDAR, UAV or compass-clinometer? Accuracy, coverage and the effects on structural models. Journal of Structural Geology 98, p 67-82.