GRSG Conference 2022: Orbit to Outcrop
Title: GoldenEye: Earth observation, drone and proximal data integration in 2D and 3D using machine learning for cost-efficient exploration targeting – Case study about stratiform tin mineralisation in Bockau, Erzgebirge/ Germany
Author: Andreas Knobloch
Abstract:
The Erzgebirge is a Variscan orogenic belt is located at the northern margin of the Bohemian Massif, bordering Germany and the Czech Republic. The Erzgebirge mining district has been mined for 800years and is most famous for its mineralisation of Sn, W, U, Ag and Li. In the western part of the Erzgebirge, the Bockau deposit is located. There, potentially economic ore bodies in a type of stratiform tin mineralization have been discovered. The mineralization consists of disseminated and often finegrained cassiterite in phyllites and quartzite schists.
The presented use case in Bockau focuses on the exploration of these potentially economic ore bodies, which have so far received little study. Within an up to 200 m thick package extending for about 10 km laterally and to depths of several hundred meters enhanced tin contents are known to exist. Within this package, several meters thick layers with potentially economic tin grades were mined in the 18th century. The characterization of the lateral extent and depth range of such layers and potential ore concentrations within them is the goal of the current exploration phase.
Firstly, in a general study for the whole Erzgebirge district, data from different public and commercial satellites, including TerraSAR, Sentinel-1 and Sentinel-2 was used for mapping soil moisture for identification of structures and layers. Secondly, detailed studies focused on the Bockau deposit itself.
A smaller area was covered by drone-base electro-magnetic (EM) survey. In addition, a drone-based LiDAR survey was conducted to map mine workings and tectonic structures that can be identified on the surface through elevation mapping. For proximal (spectral) field measurement, a new active hyperspectral sensor (AHS) and a new portable RAMAN device were used at hand specimen, surface outcrop walls and old underground mine workings. Finally, machine learning was applied using Artificial Neural Networks (ANN) and Random Forests for mineral prospectivity mapping in 2D.
Currently, 3D modelling and predictive mapping is being finalised until end of 2022. The final results of the data fusion will allow for reduction of drilling costs through improved ore body mapping in 3D showing structures and metallic bodies in the underground on a local scale, as well as increasing the potential on discovery of new mineral deposits on a regional scale.
The paper has been compiled in the frame of “Earth observation and Earth GNSS data acquisition and processing platform for safe, sustainable and cost-efficient mining operations (GoldenEye)” project. This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 869398.