GEOSCAN, résultats de la recherche


TitreAn emerging paradigm for surficial geological mapping of Arctic Canada at the Geological Survey of Canada
AuteurRussell, H A J; Broscoe, D; Giroux, D; Grunsky, E; Harris, J; Kerr, D; Lesemann, J; Parkinson, W; Richardson, M; Sharpe, D R
SourceScientific Program: AQSSS-CSSS Joint meeting; 2012 p. 132
LiensOnline - En ligne
Séries alt.Secteur des sciences de la Terre, Contribution externe 20120070
RéunionAQSSS-SCSS Joint Meeting; Quebec; CA; juin 3-8, 2012
Mediapapier; en ligne; numérique
Sujetstechniques de cartographie; télédétection; levés géologiques; satellite LANDSAT; modélisation numérique de terrain; classification des formes de relief
ProgrammeBases de données couvrant les trois territoires (la télécartographie prédictive), GEM : La géocartographie de l'énergie et des minéraux
Résumé(disponible en anglais seulement)
A challenge in remote predictive mapping (RPM) of glaciated landscapes is recognition of the series of steps in the traditional cogitative terrain analysis process and encapsulating them within computational workflows based on image analysis and statistical modelling. Within the SMART (Systematic Mapping of Arctic Canada by Remote Techniques) project of Geo-Mapping for Energy and Minerals Program (GEM), a methodology and data handling framework is being developed to improve mapping productivity.
SMART mapping is a challenge that involves : i)development of a science language for glaciated terrain, ii) integration of expert knowledge and legacy datasets, iii) parsing knowledge into machine operable components (morphology, texture, shape etc.), iv) classification of attributes, v) evaluation of various geoscience data types (i.e. remotely sensed , topographic and various calculated derivative images) for surficial mapping, and vi) statistical analysis, modelling and expert systems integration of the diverse landscape attributes within a geoscience data stack. Morphology, for example, is being extracted through analysis of DEM data and derivatives. This work forms the basis for specific landform analysis (e.g. eskers) and as a component of the data stack. Material (texture, lithology) types are primarily being captured using remotely sensed data (LANDSAT, Radar) in concert with pixel-to-pixel-based classification algorithms. Lake shape and landforms are being analyzed using form statistics and object-orientated, landscape-segmentation techniques. Spatial association of various landform metrics provides a challenge that is being undertaken using density functions and integration of specific expert interpreted data layers (drumlin, esker, etc.).
Integration of this diverse suite of data layers is being completed using several techniques, including : statistical approaches, decision trees, fuzzy sets, and expert system approaches. The requirement to revisit the classification lexicon of glacial landscapes, landforms and geological legends will result in improved perspectives and understanding of the signatures and processes of the glaciated landscape of northern Canada derived from remote predictive mapping.