22.5.2018

Science Blog: GECCO – High-performance geoscientific computing in multi-scale mineral potential studies

Teksti:
Eevaliisa Laine, Senior Scientist

A central task in mine site evaluation and 3D mineral potential studies is to collect data on multiple spatial scales and subsequently use inverse methods to infer the location and extent of economically interesting mineral deposits. The data sets comprise, for example, airborne and ground geophysical data, drill-hole data, geological maps and cross-sections, drill-core logs and geochemical data. Directly observed geological information is often sparse (e.g. drill holes) and subsurface geology has to be inferred through the interpretation and inversion of measured geophysical data.

The project GECCO combines expertise in high-performance computing and geomodelling with the aim of developing tools for faster geological modelling in a powerful computing environment. The common earth model (CEM) is a new concept in the multidisciplinary integration of data and work processes. Modern 3D earth modelling software enables the sharing of common digital 3D representations of the subsurface and rapidly incorporating new information into existing models. 3D models and all geoscientific data can be included in the same CEM model. The lithological, geochemical and geological properties in the CEM fulfil geological and geophysical constraints. Geological constraints include geological 3D models, geological maps, geochemical data and geological rules inferred from structural geological observations.

The increasing amounts of data have led to the need to use high-performance computing techniques in geosciences, with parallel computing and the use of modern accelerator technologies such as graphical processing units (GPUs) to speed up calculations. The first computational results indicate that GPUs can speed up the direct calculation of Newtonian gravitation by a factor of 200. For instance, a 500 x 500 x 40 source grid and an 84 x 84 receiver grid can be calculated in 5.2 secs. In addition, preliminary results also indicate that the use of the exact formula for the gravitational field of a cube will result in an insignificant deviation from the use of a spherical source, i.e. less than one percent, for practical distances.

Figure 1. GECCO workflow: Stochastic modelling and imaging of subsurface geology and the use of GECCO tools for gravimetric and magnetic forward modelling. Geometrical changes in the geological model (position/boundaries/properties) are immediately visible in the calculated geophysical responses.
Eevaliisa Laine

Teksti: Eevaliisa Laine

Eevaliisa Laine, D.Sc. (Tech.), works at the Bedrock Construction and Site Assessment unit of the Geological Survey of Finland. She is specialised in 3D modelling and responsible for the GECCO project at GTK financed by the Academy of Finland during 2015–2019. The whole project is led by professor Jan Westerholm from Åbo Akademi University and is being carried out in collaboration with the University of Western Australia, the Geological Survey of Norway and Pyhäsalmi mine. GTK researchers Soile Aatos, Gardar Gislasson, Suvi Heinonen, Heidi Laxtröm, Ilkka Suppala and Marit Wennerström are contributing to the project this year.