Predictive gold potential in entire southern West Greenland assessed by geological experience, artificial neural network and self-organizing maps analysis

Bo Møller Stensgaard (Redaktør)

Publikation: Bog/rapportRapport (offentligt tilgængelig)


This report presents the results of three different approaches to gold mineral potential mapping in the entire southern West Greenland. In a rather inaccessible and large area like Greenland, efficient use of data and the precise targeting of a geological phenomenon (in this case potential gold mineralised environments) are of great importance. Prior to this study, application of multivariate statistical modelling for mineral potential mapping in southern West Greenland was restricted to the Nuuk region as reported in Stensgaard (2008) and references therein. They used both a qualitative empirical approach and a model-driven approach for prediction of areas with gold mineralisations. The current study extends the analysis of the gold potential to the entire southern West Greenland (~61° to 67°N).

The first approach to the mapping of geological environments and thereby mineral potential mapping is a qualitative knowledge-driven approach, in which geochemical distribution functions from regional stream sediment samples are assessed and interpreted based on the geologists’ experience and knowledge. Certain signatures in regional aeromagnetic data and the locations of known mineralisations are used to support the interpretation of the geochemical data. Single or combined parameters in the data sets used to recognise specific geological domains and the potential for gold mineralisations are presented

The second approach is a quantitative unsupervised data-driven, artificial neural network analysis carried out on a variety of data sets. The selection of data sets are partly guided by an earlier multivariate statistical study in the Nuuk region, in which data signatures of gold occurrences were identified (Stensgaard 2008 and references therein). Forty-five training points covering the central Nuuk region, each representing a 200×200 m area in which one or several rock samples have yielded ≥1 g/t Au (our heuristic definition of a ‘gold mineralised environment’), were used to train the neural network to recognise the characteristic data signatures of the training points in a variety of re-analysed data sets. The ‘experience’ gained by the artificial neural network in this way was then used to find similar signatures in data sets that covered the entire southern West Greenland. Several areas with favourability for gold are predicted by this approach. Follow-up in one of the areas together with investi-gations of the possible causes of the data signature of the specific geological environments predicted to be favourable for gold were conducted during fieldwork in 2009 and reported in Schlatter & Stensgaard 2012.

The third approach is a Self-Organising Map analysis (SOM) which identifies data distribu-tions and correlations between different data types in a multidimensional data space. The selection of the geochemical elements included in this approach were guided by those tra-ditionally seen as pathfinder elements for gold, and by elements that had been identified as characteristic for gold in previous studies of southern West Greenland.

The three different approaches are complementary to each other. The purpose of this re-port is therefore not to rank the different approaches, neither is it the intention to make a conclusion on the exact extent and position of favourable tracts (metallogenetic provinc-es/areas) for gold. The purpose is primarily to discuss and illustrate the different approaches of accessing the mineral potential for gold in southern West Greenland and provide exploration with input and inspiration for their own decisions and evaluations on where to go and do further investigations.
Antal sider89
StatusUdgivet - 8 mar. 2013


NavnDanmarks og Grønlands Geologiske Undersøgelse Rapport


  • Greenland


  • Programområde 4: Mineralske råstoffer


Dyk ned i forskningsemnerne om 'Predictive gold potential in entire southern West Greenland assessed by geological experience, artificial neural network and self-organizing maps analysis'. Sammen danner de et unikt fingeraftryk.