Multiscale Remote Sensing and Ground Truth Dataset for Mine Waste Characterisation at Memi Mine, Republic of Cyprus (m4mining project)

This report documents the preparation, curation, and publication of a comprehensive multi-source geospatial dataset acquired as part of the Horizon Europe m4mining project. The publication aims to make high-quality hyperspectral imaging, Light Detection and Ranging (LiDAR), field spectroscopy, mineralogical, and geochemical data publicly available to support research into mine waste characterisation, environmental monitoring, and the development of advanced remote sensing methodologies. The dataset has been curated to ensure traceability, reproducibility, and long-term usability by the scientific community.

The published data originate from a field campaign conducted in April 2024 at the legacy Memi Mine in the Republic of Cyprus, a volcanogenic massive sulphide (VMS) deposit within the Troodos Ophiolite Complex. The site was selected as one of the m4mining case studies because it provides a representative environment for evaluating hyperspectral remote sensing technologies applied to weathered mine wastes. Data acquisition combined UAV-mounted hyperspectral imaging in the visible to near infrared and shortwave infrared (VNIR-SWIR) and LiDAR with ground-based handheld spectroradiometer measurements via the analytical spectral device (ASD) fieldspec instrument (VNIR-SWIR).

The report describes the complete data acquisition methodology and processing workflow used to generate georeferenced, terrain-corrected hyperspectral reflectance products. Processing was undertaken using an integrated workflow involving HySpex RAD, Applanix POSPac MMS, HySpex NAV, INTACOR, PARGE, and DROACOR software, resulting in high-resolution reflectance products accompanied by detailed processing metadata, digital surface models (DSMs), navigation information, and quality documentation. Particular emphasis is placed on processing traceability by documenting all relevant processing parameters, software versions, metadata, and workflow steps required to reproduce the published datasets. Ground truthing formed a critical component of the campaign through systematic field spectroscopy, laboratory mineralogical and geochemical analyses, and detailed documentation of sample locations and photographs. Samples were analysed using laboratory X-ray diffraction (XRD), X-ray fluorescence (XRF) and Loss On Ignition (LOI) determination. The dataset includes site sample photography, and precisely surveyed ground control points (GCPs), providing a comprehensive reference dataset across multiple spatial scales. This provides users with a robust framework for validating hyperspectral interpretations, developing mineral classification algorithms, and evaluating relationships between remotely sensed observations and laboratory-derived mineralogical and geochemical properties.

The resulting publication represents one of the most comprehensive openly available hyperspectral mine waste datasets and was collected in the context of the m4mining project using hyperspectral hardware by HySpex (Mjolnir VS-620). By combining UAV hyperspectral imagery, LiDAR-derived terrain information, field spectroscopy, laboratory analyses, and comprehensive metadata, the dataset provides a valuable resource for developing and benchmarking new approaches to mineral mapping, mine waste characterisation, environmental monitoring, and data fusion. The published dataset is intended to facilitate future collaborative research, algorithm development, and reproducible scientific investigations of hyperspectral remote sensing for the mining sector.