GRSG 35th Conference 2024 Abstract

Title: Hyperspectral Mine Face Monitoring in the Republic of Cyprus – Comparing reflectance data from platforms ground-based, UAV Nadir and Oblique scanning

Author: Justus Constantin Hildebrand

Organisation: HySpex by NEO (Norsk Elektro Optik)

Justus Constantin Hildebrand1, Friederike M. Koerting1, Daniel Schläpfer2, Nicole Koellner3, Rikke Vestergaard1, Ioulia Georgiadou Gavrilovic4
1 NORSK ELEKTRO OPTIKK AS, HYSPEX, NORWAY
2 RESE APPLICATIONS LLC, SWITZERLAND
3 GEOFORSCHUNGSZENTRUM POTSDAM, GERMANY
4 GEOLOGICAL SURVEY DEPARTMENT, NICOSIA, CYPRUS

The procurement of essential metals and minerals continues to rely on the exploitation of new resources. The growing demand for these resources inevitably has consequences for mining, mineral processing, and the environment. If the increasing demand for raw materials and the impact of expanded mining operations are not addressed more effectively, it could slow global progress towards a cleaner energy future. The M4Mining project (www.m4mining.eu) aims to promote sustainable mining practices through integrated remote sensing data from uncrewed aerial vehicles (UAV) and satellites, with the intention of facilitating the monitoring of these masses. The overarching goal is to develop comprehensive remote sensing solutions for mining and tailing sites.

In this study, we present hyperspectral imagery (HSI) collected in April 2024 from the case study site at Memi Mine. The Memi Mine is a legacy pyrite mine site located in the Republic of Cyprus. It contains processed wastes from the production of sulfur concentrate, which is derived from iron pyrite ore mining at the Memi and Alestos mines. HSI data were collected with a Mjolnir VS-620 consisting of a Visible to near- and shortwave infrared (VNIR-SWIR) camera with a combined range of 400-2500 nm. The camera was mounted on different platforms to obtain scans of the tailing slopes. The camera was first mounted on a tripod resulting in ground-based 90-degree vertical scans of the tailing slopes. Subsequently, UAV imagery was collected in two modes, as nadir scans (0 degrees, top-down view) and as oblique scans (45 degree angle, sideways view).

The datasets were acquired from the same area, at comparable distances between the camera and the tailing slope, with an approximate range of 80-120m. The objective is to investigate the differences between the datasets and the retrieved reflectance values for the various platforms, viewing geometries and effects on the correction of the atmosphere intersecting between the sensor and the surface. The UAV-based HSI is processed using industry-standard software PARGE and DROACOR, and the ground-based HSI is corrected based on the calibrated sensor radiances in combination with radiometric modelling in the DROACOR software, which has recently been adapted specifically for horizontal view geometries.

In-situ reference panel and empirical line corrected data serve as references for validation of the retrieved reflectance quantities. As part of the M4Mining project, the drone-based HSI is no longer displayed solely in 2D, but on 3D surfaces based on co-acquired UAV-based LiDAR data. Additionally, the VNIR and SWIR instruments are integrated into a single full-spectrum hyperspectral point cloud. Data from different angles, such as 0, 45 and 90 degrees, will be combined.

The observed differences between the various acquisition geometries are investigated with special attention to the used atmospheric correction approaches. The reflectance-corrected HSI is assessed and compared by applying different classification methods for minerals known to occur in areas affected by acid mine drainage and secondary iron mineral concentrations. That is the Spectral Angle Mapper (SAM), USGS PRISM Material Identification and Characterization Algorithm (MICA) and a machine learning method for the identification of secondary iron minerals based on laboratory scans of hand specimens.

Ground control points (GCPs) that were established are visible in the HSI, thus aiding the validation of the data products. These points were sampled spectrally using a portable field spectrometer, which collected data in the wavelength range of 500-2500 nm, and then sampled physically. The geochemical and mineralogical characterization of the samples still remains, which will provide further information for the interpretation and validation of the collected data and mineral classifications.