|
Several studies on remote sensing and mineral exploration have been developed or improved in the last decade. However, the low spatial resolution of satellites is a recurring problem in many cases where the mineral or rock is much smaller than the pixel size of the satellite images. This phenomenon, called sub-pixel occurrence, generates an extremely mixed pixel that difficult the performance of conventional classification algorithms. Satellites with high spatial resolution, such as the Worldview-3 (2 m Ground Sample Distance), have become an essential tool for mineral exploration studies. In addition to its high spatial resolution, the Worldview-3 has 16 bands, eight in the Visible and Near-infrared (VNIR) and eight in the Shortwave Infrared (SWIR) region, which further increases its potential for mineral exploration. This study applies a spectral unmixing-based method, using the Spectral Hourglass Wizard Workflow (SHW), to extract and select pegmatites endmembers from the WorldView-3 images. Further, these endmembers were used for a subpixel classification, performed through Mixture Tuned Matched Filtering (MTMF), to map possible pegmatite outcrops in the Tysfjord pegmatite field, Norway. After classification a high Digital Terrain Model (DTM) hillshade acquired from LiDAR technology and geological data were used to eliminate false positives. The subpixel classification results were compared with geological data and 17 points of interest for pegmatite exploration were selected for further field validation. This work shows the potential of the Worldview-3 high-resolution images processed with a spectral unmixing-based method for mineral exploration in an area of subpixel occurrence. The results are encouraging and show great value for the scientific field of pegmatite exploration.
|
