S34I Project: Secure and Sustainable Supply of Raw Materials for

EU Industry

Ana Cláudia Teodoro

1 a

, Joana Cardoso-Fernandes

1 b

, Mihaela Gheorghe

2 c

Francesco Falabella

3 d

, Fabiana Calò

3 e

, Antonio Pepe

3 f

, Delira Hanelli

4 g

Andreas Knobloch

4 h

, Roberto De La Rosa

4 i

, Fahimeh Farahnakian

5,6 j

Georgios Periklis Georgalas

7 k

, Enoc Sanz-Ablanedo

8 l

, Vaughan Williams

, Krištof Oštir

10 m

Instituto de Ciências da Terra, Departamento de Geociências, Ambiente e Ordenamento do Território, Faculdade de

Ciências, Universidade do Porto, Porto, Portugal

GMV Innovating Solutions SRL, Bucharest, Romania

CNR - National Research Council of Italy, Institute for the Electromagnetic Sensing of the Environment, Napoli, Italy

Beak Consultants GmbH, Freiberg, Germany

Geological Survey of Finland (GTK), Espoo, Finland

Department of Computing, University of Turku, Turku, Finland

Department of General Geology, Geological Mapping & Applications, Hellenic Survey of Geology & Mineral Exploration,

Acharnae, Attica, Greece

Grupo de Investigación en Geomática e Ingeniería Cartográfica (GEOINCA), Universidad de León, León, Spain

Aurum Exploration Ltd, Ireland

University of Ljubljana, Faculty of Civil and Geodetic Engineering, Ljubljana, Slovenia

Keywords: Exploration, Extraction, Environmental Monitoring, Remote Sensing, Artificial Intelligence.

Abstract: The Secure and Sustainable Supply of Raw Materials for EU Industry – S34I project is researching and

innovating new data-driven methods to analyze Earth Observation (EO) data, supporting systematic mineral

exploration and continuous monitoring of extraction, closure, and post-closure activities to increase European

autonomy regarding raw materials (RM) resources, and to use EO not only for the management of technical

and environmental issues for a green transition but also to support public awareness, mining's social

acceptance, and better legislation. S34I uses data from satellites, airborne, unmanned aerial vehicles, ground-

based sensors, underwater hyperspectral imaging and conventional in-situ techniques/methods and fieldwork.

The S34I project is supporting the technical experiments and pilot validations/demonstrations for the six pilot

use cases and at different phases of the mining life-cycle to address the challenges of the topic: Onshore

exploration (Aramo in Spain); Shallow water exploration (Ria de Vigo in Spain); Extraction (Gummern in

Austria); and Closure/post-closure (Lausitz in Germany, Aijala and Outokumpu in Finland). The S34I project

involves 19 partners from 12 European countries. The project started in January 2023 and ends in June 2025.

https://orcid.org/0000-0002-8043-6431

https://orcid.org/0000-0001-8265-3897

https://orcid.org/0000-0001-8171-6411

https://orcid.org/0000-0002-3698-908X

https://orcid.org/0000-0002-0174-5894

https://orcid.org/0000-0002-7843-3565

https://orcid.org/0009-0002-0012-1729

https://orcid.org/0000-0001-7515-001X

https://orcid.org/0000-0002-3004-7104

https://orcid.org/0000-0002-7672-9346

https://orcid.org/0009-0003-1180-2519

https://orcid.org/0000-0001-9975-5726

https://orcid.org/0000-0002-4887-7798

Teodoro, A. C., Cardoso-Fernandes, J., Gheorghe, M., Falabella, F., Calò, F., Pepe, A., Hanelli, D., Knobloch, A., De La Rosa, R., Farahnakian, F., Georgalas, G. P., Sanz-Ablanedo, E.,

Williams, V. and Oštir, K.

S34I Project: Secure and Sustainable Supply of Raw Materials for EU Industry.

DOI: 10.5220/0013470500003935

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 11th International Conference on Geographical Information Systems Theory, Applications and Management (GISTAM 2025), pages 277-285

ISBN: 978-989-758-741-2; ISSN: 2184-500X

277

1 INTRODUCTION

Sustainable mining practices, i.e., the minimization of

environmental damage, social responsibility,

reduction of ecological footprints, rehabilitating

mined land, managing waste effectively, and

conserving biodiversity, are a very hot topic and have

been a concern for many years (Laurence, 2011; Chen

et al., 2024).

The Critical Raw Materials Act (CRMA),

proposed by the European Commission, seeks to

address the challenges faced by the European Union

(EU) in securing a stable and sustainable supply of

critical raw materials (European Commission, 2023).

Its primary goal is to reduce the EU's reliance on

external sources for these materials while ensuring

their availability to support strategic sectors,

including those vital for decarbonization and

advancing green technologies (Hool et al., 2024). The

Act sets several benchmarks by 2030 along the

strategic raw materials value chain and for the

diversification of the EU supplies: (i) at least 10% of

the EU's annual consumption for extraction; (ii) at

least 40% of the EU's annual consumption for

processing; and (iii) at least 25% of the EU's annual

consumption for recycling no more than 65% of the

EU's annual consumption from a single third country.

Satellite data is crucial in sustainable mining

practices by offering tools and insights to minimize

environmental impact, optimize operations, and

ensure compliance with sustainability standards

(Farahnakian et al., 2024; Li et al., 2023; Persello et

al., 2022).

Regarding the applications of satellite data in

sustainable mining, using these data and image-

processing techniques reduces the environmental

impact by identifying high-potential areas remotely

(Rajan Girija & Mayappan, 2019). Remote sensing

has the ability to help exploration companies explore

much larger and often more remote and inaccessible

areas whilst at the same time focusing time and costs

on identifying specific target areas for further testing

(Pour et al., 2019; Beiranvand Pour et al., 2018). This

will lead to more efficient timelines for discovery.

Mineral mapping can analyze mine waste (tailings)

for recoverable resources, turning waste into reusable

materials and reducing the need for fresh extraction

(Gulicovski et al., 2024; Rodríguez-Hernández et al.,

2019; Zoran et al., 2009).

The recent advances in Artificial Intelligence (AI)

algorithms and Earth Observation (EO) free data are

aspects that broadly support several Sustainable

Development Goals and promote sustainable mining

(Chen et al., 2024; Persello et al., 2022).

The S34I project has explored new data-driven

methods to analyze EO data for systematic mineral

exploration, continuous extraction monitoring,

closure and post-closure activities, increasing

European autonomy regarding raw materials

(Farahnakian et al., 2024; Carvalho et al., 2025). The

consortium is composed of 19 partners from 12

countries (11 from the EU, plus Norway) (see Figure

1). The partners are 50% from academia and national

research centres and 50% from private

companies/industries. The project started in January

of 2023 and will end in June 2025.

Figure 1: S34I project partners.

The main objective of this work is to present the

project's main objectives and some preliminary

results. The project tackles the entire mining life cycle

(exploration, extraction, mine closure) by focusing on

six distinct pilot sites. Each mining phase and pilot

presents different challenges that were addressed

using EO data and techniques. This work will

summarize the S34I approach to address these

challenges. First, all datasets used in S34I will be

listed according to the mining phase. Then, an

overview of the methods developed/adapted in the

scope of S34I will be given. Preliminary results will

be presented and discussed for selected methods.

These methods and results will serve as a base to

develop specific services to address the identified

challenges.

2 DATA

S34I utilized data from Copernicus missions and

Copernicus Contributing Missions (CCM) obtained

from the European Space Agency and other satellite

sensors, while additional platforms, which included

airborne systems, unmanned aerial vehicles (UAVs),

ground-based methods, in-situ techniques, and

S34I 2025 - Special Session on S34I - From the Sky to the Soil

278

fieldwork, were employed for calibration, validation,

or to complement the satellite data. The data used

varied depending on the mining phase and pilot site:

▪ Exploration Phase - Onshore Pilot (Aramo

Mine, Spain): Sentinel-1, Sentinel-2, Landsat-

9, Hyperspectral Precursor of the Application

Mission (PRISMA), Advanced Land

Observing Satellite (ALOS) Phased Array type

L-band Synthetic Aperture Radar (PALSAR)-

2, Constellation of Small Satellites for

Mediterranean basin Observation (COSMO-

SkyMed), airborne Light Detection and

Ranging (LiDAR), hyperspectral data, and

ground spectral libraries of rocks and soils.

▪ Exploration Phase - Shallow Waters Pilot (Rias

Baixas, Spain): Sentinel-1, Sentinel-2,

Landsat-9, WorldView-2/-3, EnMap,

Underwater Hyperspectral Imaging (UHI),

complementary spectral libraries, and pre-

existing or newly acquired geological data.

▪ Extraction Phase (Gummern, Austria):

Pléiades Neo tri-stereo, WorldView-2,

Sentinel-1, Sentinel-2, COSMO-SkyMed,

UAV data, and ground Global Navigation

Satellite Systems (GNSS) stations.

▪ Closure and post-closure Phase (Lausitz and

Outokumpu): Sentinel-2, PRISMA,

WorldView-3, UAV data, and geochemical

water data; and Sentinel-1, Sentinel-2, and

COSMO-SkyMed data (Aijala).

It should be noted that the S34I consortium

worked together with the holders of the rights for

exploration in Aramo and for exploitation in

Gummern.

3 METHODS

The data and methodology employed in the S34I

project depend on the pilot case and the mining phase

addressed. However, the principal outcomes of the

S34I project focus on processing Copernicus and

CCM data.

Several approaches were applied and developed,

from traditional methods to new ensemble machine

learning (ML) algorithms. Figure 2 presents the

methods developed in the primary pilot area. The

techniques have been developed in a specific location

but later implemented in other pilot areas.

The methods were developed according to the

type of EO data exploited (satellite and other data), as

shown in Figure 3.

Figure 2: Developed methods according to the primary pilot

area.

Figure 3: Developed methods according to the data.

3.1 Exploration Phase

For the onshore pilot (Aramo, Spain), various

analytical methods were employed, including RGB

combinations, band ratios, Principal Component

Analysis (PCA), K-means clustering, end-member

extraction, minimum wavelength mapping, Spectral

Angle Mapper (SAM), Self-Organizing Maps

(SOM), and Artificial Neural Networks (ANNs). A

novel ensemble AI method was developed by

integrating Support Vector Machines (SVM),

Random Forest (RF), and ANNs. Additionally, a

specialized AI algorithm was designed for automated

pre-processing of hyperspectral airborne data,

requiring minimal ground truth input. This was aided

by pre-existing geochemical datasets, which

minimized the requirement for a large part of follow-

up ground truthing.

In the shallow water exploration pilot (Rias

Baixas, Spain), RGB combinations, band ratios,

PCA, K-means clustering, spectral unmixing, and

Object-Based Image Analysis (OBIA) enhanced

feature detection and classification.

S34I Project: Secure and Sustainable Supply of Raw Materials for EU Industry

279

3.2 Extraction Phase

In the extraction pilot (Gummern, Austria), high-

resolution Digital Elevation Models (DEMs) were

generated, and UAV photogrammetry was effectively

performed using Structure from Motion (SfM) for

detailed surface modelling.

Interferometric Synthetic Aperture Radar

(InSAR) and change detection techniques, including

the Normalized Decorrelation Change Index (NDCI),

were applied to Synthetic Aperture Radar (SAR) data

for detailed surface analysis. Additionally, advanced

AI models—Residual-in-Residual Dense Block

(RRDB), Super-Resolution U-Net (SRUN), and

Optical-Guided Super-Resolution Network

(OGSRN)—were implemented to enhance the

resolution and quality of SAR imagery.

A Low-cost GNSS Monitoring System (LGMS)

using low-cost GNSS receivers to monitor

displacements with high precision was implemented.

The LGMS receives and stores GNSS observations

continuously, which are later post-processed to

estimate daily displacements of the monitoring

locations.

3.3 Closure and Post-Closure Phase

For closed mines affected by Acid Mine Drainage

(AMD) (Lausitz, Germany and Outokumpu,

Finland), unsupervised learning methods such as

SOM and K-means clustering were utilized for

pattern recognition and data analysis. Supervised

classification techniques were applied to improve

predictive accuracy, including ANN, logistic

regression, RF, and K-nearest Neighbors (KNN).

Additionally, image enhancement and change

detection techniques were implemented at the Aijala

pilot (Finland).

4 RESULTS AND DISCUSSION

In this section, some preliminary results will be

presented and discussed.

4.1 Exploration Phase

In the onshore pilot study, SOM was applied for data

exploration to analyze geochemical sample points. A

K-means clustering algorithm was also applied to the

SOM output to categorize the data into distinct

clusters. This combination allows for identifying

meaningful groupings and enhances the

interpretability of spatial and spectral patterns.

Ultimately, it improves understanding of Cobalt (Co)

distribution captured by PRISMA and LiDAR data

(Figure 4).

Figure 4: SOM results from geochemical sample points of

PRISMA: (a-c) Example variable SOMs related to the band

information (band 1, 40 and 106).

S34I 2025 - Special Session on S34I - From the Sky to the Soil

280

A new ensemble AI method was developed by

integrating SVM, RF, and ANN to exploit different

satellite-based datasets. Emphasis was given to

Copernicus data (Sentinel-2), Landsat-9 and

PRISMA, combining multispectral and hyperspectral

images. This classifier ensemble method employed a

Soft Voting strategy. The ensemble classifier shows

variable performance with different input data.

PRISMA data allows for better class separability, but

the model tends to overfit, indicating the need for

better regularisation (Figure 5).

Figure 5: Examples of prediction map of hydrothermal

alteration area (coloured pixels) generated from PRISMA

dataset on the onshore exploration phase.

A spectral library was created, including 311

samples of outcropping rocks, 371 samples of soils,

and 208 samples from the old mine in Aramo. The

mineralogical associations corresponding to the

outcropping rocks and soils in the Aramo Plateau,

their spectral signatures in the SWIR region and their

relationship with the contents of Co and other

elements (Nickel (Ni) and Copper (Cu)) were

determined. Although no clear spectral signature for

Co was identified, the study successfully defined nine

distinct spectral signatures from the 11 mineralogical

associations. This could be mainly attributed to the

fact that the deposit is mainly enriched in Cu, but with

associated Cu and Ni. Additionally, statistical

correlations between mineralogical associations and

geochemical data revealed that specific associations

showed a higher probability of containing elevated

Co, Ni, and Cu levels.

A method was also developed for pre-processing

airborne hyperspectral data. This involves converting

digital counts from hyperspectral sensors into

radiance values through several key steps. Geometric

correction ensures that image pixels correspond to

their correct geographic locations, while atmospheric

correction mitigates the effects of the atmosphere,

such as absorption and scattering, on the measured

radiance. Additionally, spectral calibration ensures

precise wavelength alignment to match known

spectral features. A semi-automated workflow was

implemented for the large-scale interpretation of

hyperspectral data, integrating satellite, airborne

platforms and UAVs, combined with ground

spectroradiometer measurements to automate

extracting meaningful features for geological

interpretation.

A methodology for utilizing EO data from

airborne and/or UAVs to develop predictive mineral

maps for the Aramo plateau during the exploration

mining phase was also developed.

Regarding the shallow waters pilot, a methodology

was developed for identifying placer deposits using an

OBIA and high-resolution satellite data. This analysis

revealed distinct signatures for each material within the

spectral resolution of the WorldView-3 satellite data,

leading to the creation of a new band ratio for

prospective deposit categorization (VNIR1-VNIR4

band ratio to identify Ilmenite) and one for determining

the geological background from vegetation (VNIR6-

VNIR8). Three different ML approaches are developed

and compared with each other: a single-level model

using the SVM algorithm, a multi-level model using

the KNN algorithm, and a dynamic classification

developed with a decision tree model. The outcome of

implementing these methods is the production of maps

illustrating the distribution of placer deposits within the

coastal area of Vigo, which can be seen in Figure 6.

Figure 6: Examples of prediction map generated using

Single-level OBIA on the WorldView-3 dataset on the

shallow waters pilot.

S34I Project: Secure and Sustainable Supply of Raw Materials for EU Industry

281

We also applied an innovative UHI for shallow

water exploration to identify potential areas for the

occurrence of CRMs. UHI data was acquired with

field work and refers to three different sea zones in

Ria de Vigo, along with seabed samples from the

same areas. The methodology centres on the Mixture

Tuned Matched Filtering (MTMF) algorithm to map

seabed targets by matching known spectral signatures

with UHI data. It also identified sediments with

spectral signatures consistent with placer deposits

found on Ria de Vigo beaches.

Finally, we integrate traditional geological

methodologies with EO techniques for delineating

CRM prospective areas for placer deposits.

4.2 Extraction Phase

The experimental results include images at a global

scale, as collected in the Sentinel-1 and -2 datasets.

All models' performance was evaluated using PSNR

(Peak Signal-to-Noise Ratio) and SSIM (Structural

Similarity Index Measure). The Sentinel-2 enhanced

model (OGSRN) performs less than the Dense

Residual-in-Residual Dense Block (RRDB) model,

based on Sentinel-1 only. Overall, it demonstrated

how deep learning models can improve the resolution

of SAR data, showing an increase of 18% for the

PSNR score and about 2% for the SSIM score

compared to the corresponding baseline bicubic

scores.

The Multigrid InSAR technique provides accurate

measurements, with millimetre accuracy, of the

ground deformations along the satellite radar line-of-

sight (LOS) direction. Combining multi-band and

multi-orbit SAR data, the obtained LOS displacement

measurements can be profitably exploited to compute

three-dimensional (3D) ground movement (Figure 7).

Deformation is due to soil compaction, so the material

added to the waste dump naturally keeps compacting.

The implementation of LGMSs at Gummern

Mine includes two stable reference points (F1 and F3)

placed on the stable ground and three observation

points (ST1, ST2, and ST3), realized with metal

poles, mounted in concrete pillars (Figure 8).

Based on the one-year testing period, it can be

stated that LGMS performed well and detected slow

movement with sub-centimetre accuracy. The results

indicated that horizontal and vertical displacements

of 10 and 20 mm occurred in ST1, while larger

displacements were noticed in ST2, which moved 25

mm horizontally and 40 mm vertically.

Figure 7: Gummern ground deformation map (a) and time

series obtained from COSMO-SkyMed (b).

Figure 8: Locations of measuring stations.

An innovative methodology was also developed

based on satellite images to continuously monitor the

life and evolution of mining waste deposits. Initially,

land surveying is conducted to establish ground

control points (GCPs). Following this, UAV flights

are undertaken to produce high-resolution DEMs.

The data acquisition process involves SfM

processing. The next stage involves acquiring new

satellite image datasets for specific epochs. For each

epoch, a different DEM and orthophoto are obtained.

Comparative analysis of successive DEMs allows

calculating geometric or volumetric changes in the

waste dumps over time. The Pléiades Neo tri-stereo

dataset was the first satellite imagery used in the

study, captured in October 2023, approximately one

month after the UAV flight. These images already

showed notable changes in the waste dump. To

S34I 2025 - Special Session on S34I - From the Sky to the Soil

282

compare the UAV-derived high-resolution DEM (HR

DEM) with the Pléiades Neo-derived DEM, the UAV

HR DEM was resampled to match the resolution of

the Pléiades Neo DEM (30 cm). WorldView-2

images, taken in April 2024, six months after the

Pléiades Neo tri-stereo images, provided an

opportunity to study changes over a longer period.

Figure 9 illustrates altitude differences between the

DEM from WorldView-2 (12. 4. 2024) and the DEM

from Pléiades Neo (1. 10. 2023). The volume changes

in the areas where waste dumping occurred are

summarized, with a total discharge volume of

101,872 m³.

Figure 9: Volumes of waste dumps between 12. 4. 2024

(DEM from Pléiades NEO) and 1. 10. 2023 (DEM from

WorldView 2).

4.3 Closure and Post-Closure Phase

Several ML algorithms were adapted to enhance pixel

classification of AMD from Sentinel 2 imagery

(Lausitz and Outokumpu) and also using space- and

airborne- multispectral and hyperspectral imagery.

SOM was also utilized to visualize and cluster high-

dimensional data to interpret complex spatial data for

AMD mapping. The study also evaluates the potential

of spaceborne hyperspectral imagery for AMD

mapping.

We also proposed ML algorithms, including RF,

KNN, Logistic Regression (LR), and MLP. They are

used to perform a pixel-based classification of the

images into AMD or non-AMD classes, as well as to

assess the severity of AMD by quantitative mapping

of AMD constituents, such as iron concentration and

pH values. The prediction map for three lakes in the

Outokumpu area is shown in Figure 10. The

visualization indicates that the RF model accurately

classified the pixels, as the lakes were primarily

contaminated by AMD rather than coastal areas.

Additionally, the SOM method was also used to

visualize and cluster high-dimensional data,

simplifying the understanding and interpretation of

complex spatial data for AMD mapping. The output

of the SOM method is grid data, such as heatmaps or

U-matrix plots, which provide insights into the

clustering and organization of the data within the grid.

The main objective was the identification of AMD-

affected areas. These experiments were conducted in

three lakes located in Outokumpu, Finland.

Figure 10: Prediction map of the best model (RF) for three

study lakes and their water samples in Outokumpu.

Finally, we perform cross-sensor analysis over

water bodies to harmonize Worldview-3 and UAV

multispectral datasets to Sentinel-2. Given the free

availability of Sentinel-2 data and the typically high

costs of high-resolution commercial EO datasets, we

propose a methodology where MLP is trained using

Sentinel-2 data over a large area in conjunction with

extensive geochemical monitoring data and the

established dependencies are applied to commercial

high-resolution datasets for targeted identification of

AMD in specific areas. This approach enables a cost-

efficient combination of free-of-charge and

commercial EO datasets for AMD mapping.

Despite the cessation of mining activities, a

sudden ground collapse in February 2017 near the

Aijala refinery highlighted ongoing environmental

risks, demonstrating the need to monitor post-closure

mining sites continuously. SAR image coherence

measures how similar two radar images of the same

area are, taken at different times. For the temporal

decorrelation analysis, we proposed calculating a new

index, the Normalized Decorrelation Change Index

(NDCI), presented in Figure 11. The method's

effectiveness was validated by comparing results with

S34I Project: Secure and Sustainable Supply of Raw Materials for EU Industry

283

Sentinel 2A images and relevant background

information about the Aijala site.

Figure 11: NDCI depicting ongoing high disturbances in the

Aijala post-closure site between 2015-2017.

5 CONCLUSIONS

In this work, we present the preliminary results of the

S34I HORIZON project. At the moment, the methods

are being validated and verified in the pilots and also

with the end-users.

Based on the methods developed under the scope

of S34I, we prototyped EO-based services that cater

to the specific needs of mining stakeholders. These

services aim to address three key areas:

1. RM Deposits Mapping: This involves using

EO data to identify and map potential mineral

deposits, both on land and in shallow waters.

2. Early Warnings: The focus here is on

developing EO-based systems that can provide early

warnings of potential hazards at mining sites, such as

ground instability.

3. Environmental Monitoring: This

encompasses the use of EO data to monitor the

environmental impact of mining activities, such as the

detection of AMD.

In the future, these services will be available for

the stakeholders and end-users.

The result of this project will be an important step

forward in monitoring all phases of the mining cycle

using EO data, contributing to more sustainable

mining practices.

ACKNOWLEDGEMENTS

This study is funded by the European Union under

grant agreement no. 101091616

(https://doi.org/10.3030/101091616), project S34I –

Secure and Sustainable Supply of Raw Materials for

EU Industry.

REFERENCES

Beiranvand Pour, A., Park, T.-Y. S., Park, Y., Hong, J. K.,

Zoheir, B., Pradhan, B., Ayoobi, I., & Hashim, M.

(2018). Application of Multi-Sensor Satellite Data for

Exploration of Zn–Pb Sulfide Mineralization in the

Franklinian Basin, North Greenland. Remote

Sensing, 10(8), 1186.

https://doi.org/10.3390/rs10081186.

Cardoso-Fernandes, J. (2024). S34I. D4.1: Catalogue of

EO-based Prototyped Services for CRM and non-CRM

Deposits Mapping. S34I Horizon Europe GA

101091616. Available at:

https://cordis.europa.eu/project/id/101091616/results

(accessed on 15 January 2025).

Carvalho, M., Cardoso-Fernandes, J., González, F. J., &

Teodoro, A. C. (2025). Comparative Performance of

Sentinel-2 and Landsat-9 Data for Raw Materials'

Exploration Onshore and in Coastal Areas. Remote

Sensing, 17(2). doi: 10.3390/rs17020305.

Chen, L., Xie, Y., Wang, Y., Ge, S., Wang, F.Y (2024).

Sustainable Mining in the Era of Artificial Intelligence.

IEEE/CAA Journal of Automatica Sinica, 11(1), 1-4,

doi.org/10.1109/JAS.2023.124182.

European Commission (2023). European Critical Raw

Materials Act. European Commission - Press release.

https://ec.europa.eu/commission/presscorner/detail/en/

ip_23_1661. Accessed 14 January 2025.

Farahnakian, F., Luodes, N., & Karlsson, T. (2024).

Machine Learning Algorithms for Acid Mine Drainage

Mapping Using Sentinel-2 and Worldview-3. Remote

Sensing, 16(24), 4680.

https://doi.org/10.3390/rs16244680.

Gulicovski, J., Stojmenović, M., Rosić, M., Vasić, A.,

Ristović, I., Janković-Častvan, I., & Kragović, M.

(2024). Mineralogical Characterization of the Grot

Lead and Zinc Mine Tailings from Aspects of Their

Further Use as Raw Material. Applied Sciences, 14(3),

1167. https://doi.org/10.3390/app14031167.

Hool, A., Helbig, C., Wierink, G. (2024). Challenges and

opportunities of the European Critical Raw Materials

Act. Mineral Economics, 37(3), 661-668,

doi.org/10.1007/s13563-023-00394-y.

Laurence, D. (2011). Establishing a sustainable mining

operation: an overview. Journal of Cleaner Production,

19(2), 278-284, doi.org/10.1016/j.jclepro.2010.08.019.

Li, S., Wang, R., Wang, L., Liu, S., Ye, J., Xu, H., & Niu,

R. (2023). An Approach for Monitoring Shallow

Surface Outcrop Mining Activities Based on

Multisource Satellite Remote Sensing Data. Remote

Sensing, 15(16), 4062.

https://doi.org/10.3390/rs15164062.

Oštir, K. et al. (2024). S34I. D3.3: Catalogue of Innovative

New Methods to Analyze Earth Observation Data. S34I

S34I 2025 - Special Session on S34I - From the Sky to the Soil

284

Horizon Europe GA 101091616. Available at:

https://cordis.europa.eu/project/id/101091616/results

(accessed on 15 January 2025).

Persello, C., Wegner, J.D., Hänsch, R., Tuia, D., Ghamisi,

P., Koeva, M., Camps-Valls, G. (2022). Deep Learning

and Earth Observation to Support the Sustainable

Development Goals: Current approaches, open

challenges, and future opportunities. IEEE Geoscience

and Remote Sensing Magazine, 10(2), 172-200,

doi.org/10.1109/MGRS.2021.3136100, 2022.

Pour, A. B., Hashim, M., Hong, J. K., & Park, Y. (2019).

Lithological and alteration mineral mapping in poorly

exposed lithologies using Landsat-8 and ASTER

satellite data: North-eastern Graham Land, Antarctic

Peninsula. Ore Geology Reviews, 108, 112-133. doi:

https://doi.org/10.1016/j.oregeorev.2017.07.018.

Rajan Girija, R., & Mayappan, S. (2019). Mapping of

mineral resources and lithological units: a review of

remote sensing techniques. International Journal of

Image and Data Fusion, 10(2), 79–106.

https://doi.org/10.1080/19479832.2019.1589585

Rodríguez-Hernández, A., Briones-Gallardo, R., Razo, I.,

Noyola-Medrano, C., & Lázaro, I. (2016). Processing

Methodology Based on ASTER Data for Mapping

Mine Waste Dumps in a Semiarid Polysulphide Mine

District. Canadian Journal of Remote Sensing, 42(6),

643–655.

https://doi.org/10.1080/07038992.2016.1197037.

Zoran, M. A., Savastru, R. S., Savastru, D. M., Miclos, S.

I., & Tautan, M. N. M. (2009). Remote sensing

techniques for mining waste characterization. Proc.

SPIE, Remote Sensing for Environmental Monitoring,

GIS Applications, and Geology IX, 74780Q. doi:

10.1117/12.830513.

S34I Project: Secure and Sustainable Supply of Raw Materials for EU Industry

285