Article

Comparative Performance of Sentinel-2 and Landsat-9 Data for

Raw Materials’ Exploration Onshore and in Coastal Areas
Morgana Carvalho 1 , Joana Cardoso-Fernandes 1 , Francisco Javier González 2 and Ana Claudia Teodoro 1, *

1 Institute of Earth Sciences, Faculty of Sciences, University of Porto, Rua Campo Alegre s/n,
4169-007 Porto, Portugal; morgana.carvalho@fc.up.pt (M.C.); joana.fernandes@fc.up.pt (J.C.-F.)
2 Marine Geology Resources and Extreme Environments, Geological Survey of Spain (IGME-CSIC),
C/Ríos Rosas 23, 28003 Madrid, Spain; fj.gonzalez@igme.es
* Correspondence: amteodor@fc.up.pt

Abstract: The demand for Critical Raw Materials (CRM) is increasing due to the need
to decarbonize economies and transition to a sustainable low-carbon future achieving
climate goals. To address this, the European Union is investing in the discovery of new
mineral deposits within its territory. The S34I project (Secure and Sustainable Supply of
Raw Materials for EU Industry) is developing Earth observation (EO) methods to support
this goal. This study compares the performance of two satellites, Sentinel-2 and Landsat-9,
for mineral exploration in two geologically distinct areas in northern Spain. The first
area, Ria de Vigo, contains marine placer deposits of heavy minerals, while the second,
Aramo, hosts Co-Ni epithermal deposits. These sites provide exceptional case studies to
improve EO-based methods for CRM exploration onshore and coastal regions, focusing on
deposits often overlooked in remote sensing studies. Standard remote sensing methods
such as RGB combinations, Principal Component Analysis (PCA), and band ratios were
adapted and compared for both satellites. The results showed similar performance in the
Ria de Vigo area, but Sentinel-2 performed better in Aramo, identifying a higher number of
zones of mineral alterations. The study highlights the advantages of Sentinel-2’s higher
spatial resolution, especially for mapping smaller or more scattered mineral deposits.
These findings suggest that Sentinel-2 could play a larger role in mineral exploration. This
research provides valuable insights into using EO data for diverse CRM deposits.
Academic Editor: Dino Ienco

Received: 2 December 2024 Keywords: earth observation; critical raw materials; marine placer deposits; cobalt; nickel;
Revised: 8 January 2025 mineral exploration; remote sensing
Accepted: 13 January 2025
Published: 16 January 2025

Citation: Carvalho, M.;

Cardoso-Fernandes, J.; González, F.J.; 1. Introduction
Teodoro, A.C. Comparative
Performance of Sentinel-2 and
Intending to address the European Union’s (EU) dependency on imported Critical Raw
Landsat-9 Data for Raw Materials’ Materials (CRM), the Critical Raw Materials Act was proposed in March of 2023, responding
Exploration Onshore and in Coastal to the “Green Industrial Plan”. The act aims to support the green transition, anticipating
Areas. Remote Sens. 2025, 17, 305. and mitigating supply risks, and reducing the underutilization of CRM deposits located
https://doi.org/10.3390/rs17020305
in European soil [1]. This act emphasizes the strategic role of advanced remote sensing
Copyright: © 2025 by the authors. technologies in identifying and prioritizing CRM deposits within European territories. By
Licensee MDPI, Basel, Switzerland. promoting investments in Earth observation tools, we can achieve a more accurate and
This article is an open access article
sustainable assessment of mineral deposits.
distributed under the terms and
CRMs are used for a number of application areas, including solar panels, wind turbines,
conditions of the Creative Commons
Attribution (CC BY) license
and electric vehicle batteries and motors, which are critical to replace fossil-based energy
(https://creativecommons.org/ and fuels in support of the EU’s green transition and sustainability goals [2,3]. CRMs are
licenses/by/4.0/). also used to store energy in batteries and infrastructure for cleaner technologies [2]. Metals

Remote Sens. 2025, 17, 305 https://doi.org/10.3390/rs17020305

Remote Sens. 2025, 17, 305 2 of 19

like Lithium, Cobalt (Co), and Nickel (Ni) are used in varying proportions in lithium-ion
batteries, while rare earth elements (REEs) are used in permanent magnets for electric
motors and wind turbines. Co and Ni demand is expected to be almost 20 times higher
in 2040 compared to 2020 [4]. Moreover, CRM extraction, processing, and manufacturing
can also contribute to employment opportunities, fostering economic development and
social sustainability [5]. Therefore, responsible exploration, extraction, and use of CRMs
can contribute to various Sustainable Development Goals (SDGs): SDG 7, SDG 13, SDG 8,
SDG 9, and SDG 11, among others [6]. Consequently, the Critical Raw Materials Act is a
policy instrument to bolster the European green transition [3].
The present study is part of the S34I project (https://s34i.eu/; accessed 26 November
2024), with the objective to explore and prototype new methods to analyze Earth Obser-
vation (EO) data, targeting mineral deposits not classically explored in remote sensing
studies, contributing to the secure and sustainable supply of CRM in Europe [7].
EO methods involve remote sensing technologies that make obtaining data from an
area with short revisiting periods of time possible. EO data/methods have applications in
diverse fields, with a highlight on land monitoring and land cover classification and many
studies focusing on lithological/alteration mapping [8–12]. Currently, the most available
data from multispectral satellites comes from the Landsat series and Sentinel-2. The two
satellites share similarities in spatial, spectral, and angular specifications [8]. Currently,
NASA (National Space Administration) is working to integrate the data from those satellites
in the HSL (Harmonized Landsat Sentinel-2) project. The integration is possible due to the
similarities between those satellites, which can be useful when higher temporal resolution,
cloud, and shadow-free data are required [8].
Sentinel-2 is a Copernicus program satellite coordinated by the EU and the European
Space Agency (ESA). This mission comprises two satellites, Sentinel-2A and Sentinel-2B,
launched in 2015 and 2017, respectively [13]. Landsat-9, launched in 2021, is the more
recent satellite launched as part of the Landsat program (Landsat-1 was launched in 1972)
managed by NASA and the United States Geological Survey (USGS) [14]. Landsat-9 is a
continuation of the Landsat program, reducing the revisiting time of Landsat-8/9 from
16 to 8 days [14]. Landsat-9 has 11 bands, of which 2 are thermal bands. Most of its spectral
bands have a 30-m resolution, but the satellite includes a panchromatic band with a 15-m
resolution. The Sentinel-2 revisit time is 10 days at the equator, which can be reduced to
5 days with both Sentinel-2A and Sentinel-2B [13]. The Sentinel-2 satellite has a higher
spatial resolution, with 13 bands ranging between 10, 20, and 60 m resolution. Both satellites
have similar spectral band configurations since they were designed for land and vegetation
observation and capture data in visible, near-infrared (NIR), and shortwave infrared (SWIR)
wavelengths. Sentinel-2 has additional red-edge bands, covering 705 nm (band 5), 740 nm
(band 6), and (band 7) 783 nm. Those satellites operate in sun-synchronous orbits and
Landsat-9 has a swath of 185 km, while Sentinel-2 has a wider swath of 290 km [9,13,14].
Sentinel-2 and Landsat-8 satellites have been explored in several studies for mineral
exploration. Adiri et al. [15] wrote a review about the use of the two satellites in the topic,
pointing out that while Sentinel-2 provides a better spatial resolution, Landsat has a more
extended history of use due to the program’s longevity. Sentinel-2 has been highly effective,
particularly in detecting iron oxides and hydroxyl-bearing minerals [15].
Landsat data have been broadly employed to map clays, carbonates, hydroxyls, ferric
iron and ferrous minerals, and hydrothermal alterations, using traditional remote sensing
techniques [16] and, more recently, diverse machine learning models [17–19]. The two satel-
lites have been compared for mapping sandbars and water on the Lower Targus River,
Portugal [20], and the study demonstrated that both satellites have similar capacities, which
Remote Sens. 2025, 17, 305 3 of 19

amplifies the possibility of intercalating the data if an analysis with shorter revisiting time
is required.
This study applies the traditional methodology of band ratios, RGB (Red, Green, Blue)
compositions, and Principal Component Analysis (PCA) to two different areas with a
unique geological context, Ria de Vigo, which contains marine placer deposits, and Aramo,
which hosts Co-Ni deposits. Those methodologies were developed either for the Landsat
series and adapted in the scope of this study to Sentinel-2 for comparison or were applied to
Sentinel-2 and adapted to Landsat-9 bands. The application of remote sensing techniques,
such as band ratios, allows for the quick identification of zones with specific mineralogical
proprieties [21], as well as PCA and RGB compositions, which are applied in diverse studies
and are an effective tool to map alteration zones, mineral zones, lithological differences
and snow/ice, bare soil, and vegetation land cover mappings [11,22–24]. A preliminary
study for the two areas explored Sentinel-1 and Sentinel-2 capabilities using unsupervised
learning, band ratios, RGB compositions, and PCA in a broad approach [25]. In the present
study, we reapplied the combinations that resulted better and adapted them to Landsat-9
for comparison. We also explored new band ratios that were applied to both areas. Our
main objective was to understand how comparable those satellites, Sentinel-2 and Landsat-
8/9, are when employed to specific deposit types, heavy mineral marine placer and Co-Ni
epithermal mineralization, which are not usually targeted in remote sensing studies but
have growing importance for the green transition/economy and are expected to be in an
increasing demand.

2. Study Areas
2.1. Ria de Vigo
The Rias Baixas region, on the coast of Galicia, northwestern Spain, where Ria de Vigo
is located (Figure 1), consists of a group of four funnel-shaped estuaries, oriented NE-SW,
with several occurrences of heavy mineral marine placer deposits [26,27]. Marine placer
deposits occur primarily along the coastline or in shallow waters of the Ria de Vigo estuary.
Placers of heavy minerals, in general, are derived from the eroded bedrock, mostly from
metamorphic and igneous rocks, transported to the coastal area by rivers [28] and, due
to their characteristic densities, they are deposited by the action of waves, currents, and
tides in the sandy beach systems [26–29]. In Ria de Vigo, the waves mainly influence the
selection of sediments [26,28–30], disfavoring the decantation of finer sediments in shallow
areas that tend to concentrate coarser fractions. The tide influence in the area is restricted
to shallower zones, and the coarser material transported by rivers tends to remain close to
the river mouth Vilas, et al. [26]. The heavy minerals placers found in the region include
deposits of zircon, garnet, monazite, tourmaline, pyroxene, ilmenite, magnetite, hematite,
goethite, and lithium minerals like spodumene associated with pegmatites [26,31–33].
Challenges in detecting the marine placers of heavy minerals in the region are related
to the small size of the deposits [34], seasonal variations in their distribution, and spatial
and spectral resolution limitations. Band ratios were selected to detect heavy minerals
associated with metamorphic and intrusive adjacent rocks [35] (hydrothermal alteration
band ratio), garnets (iron band ratios), and clay minerals that can be associated with
placers deposits

2.2. Aramo
The Aramo study area is located in the Sierra del Aramo, in northern Spain (Figure 2),
within the Cantabrian Zone of the Iberian Massif [36]. The area hosts Co, Cu, and Ni
mineralization linked to the Late-Variscan Aramo Fault. The geological setting features a
Remote Sens. 2025, 17, 305 4 of 19

Remote Sens. 2025, 17, x FOR PEER REVIEW 4 of 20

stacking of Namurian limestones (Upper Carboniferous) in thrust sheets, creating a rugged
morphologically
Figure 1. The Ria dediverse landscape
Vigo study with karst
area location features
in northern [37,38].
Spain and details of the beaches analyzed
in this study.

2.2. Aramo
The Aramo study area is located in the Sierra del Aramo, in northern Spain (Figure
2), within the Cantabrian Zone of the Iberian Massif [36]. The area hosts Co, Cu, and Ni
Spain
mineralization linked to the Late-Variscan Aramo Fault. The geological setting features a
stacking of Namurian limestones (Upper Carboniferous) in thrust sheets, creating a rug-
ged morphologically diverse landscape with karst features [37,38].
The Aramo Unit consists of Devonian to Carboniferous rocks, including dolomites,
limestones, sandstones, and shales, with major fault and thrust systems shaping the land-
scape [38]. The Cu-Co-Ni mineralization is hosted in Carboniferous limestones and is pri-
marily linked to epithermal deposits in karst cavities [39,40]. The mineral assemblage in-
cludes chalcopyrite, cobaltite, pyrite, and various secondary minerals, such as chalcocite
and covellite, formed by supergene oxidation. The Aramo mine is characterized by Cu-
Ni-Co sulfides and arsenides formed in a hydrothermal system associated with Late-Var-
iscan tectonic activity [37,39]. Dolomitization is the main alteration process around the
fault zones, with three primary hypogenic mineralization stages, followed by significant
Figure
supergene The
Figure1.1.The Riade
Ria deVigo
Vigostudy
weathering. studymineralization
The arealocation
area locationin
innorthern
northern
occurs Spain
inSpain and
and
veins details
details
and lenses ofthe
of thebeaches
alongbeaches analyzed
E-W analyzed
and NE-
in
in this
this study.
study.
SW faults, often accompanied by secondary porosity from karstification [40].

2.2. Aramo
The Aramo study area is located in the Sierra del Aramo, in northern Spain (Figure
2), within the Cantabrian Zone of the Iberian Massif [36]. The area hosts Co, Cu, and Ni
mineralization linked to the Late-Variscan Aramo Fault. The geological setting features a
stacking of Namurian limestones (Upper Carboniferous) in thrust sheets, creating a rug-
ged morphologically diverse landscape with karst features [37,38].
The Aramo Unit consists of Devonian to Carboniferous rocks, including dolomites,
limestones, sandstones, and shales, with major fault and thrust systems shaping the land-
scape [38]. The Cu-Co-Ni mineralization is hosted in Carboniferous limestones and is pri-
marily linked to epithermal deposits in karst cavities [39,40]. The mineral assemblage in-
cludes chalcopyrite, cobaltite, pyrite, and various secondary minerals, such as chalcocite
and covellite, formed by supergene oxidation. The Aramo mine is characterized by Cu-
Ni-Co sulfides and arsenides formed in a hydrothermal system associated with Late-Var-
iscan tectonic activity [37,39]. Dolomitization is the main alteration process around the
fault zones,
Figure withstudy
2. Aramo threearea
primary hypogenic
location mineralization
in Spain, Aramo stages,
plateau detail andfollowed byscheme
geological significant
of the
supergene weathering. The mineralization occurs in veins and lenses along E-W and NE-
Sierra del Aramo. The legend corresponds to 1. Dolomites, limestones, sandstones, shales, marlstones,
microconglomerates,
SW and siltstones
faults, often accompanied by(Devonian).
secondary2.porosity
Quartzitefrom
sandstones, marlstones,
karstification [40].and limestones
(Upper Devonian–Carboniferous). 3. Massive limestones (Namurian Carboniferous). 4. Shales,
sandstones, limestones, quartzite conglomerates, and coal (Moscovian Carboniferous). 5. Mass
movement deposits, debris deposits, and alluvial deposits (Quaternary). 6. Overthrusts. 7. Faults [38].
Datum WGS 84, UTM zone 30N, scale 1:95,000.

The Aramo Unit consists of Devonian to Carboniferous rocks, including dolomites,

limestones, sandstones, and shales, with major fault and thrust systems shaping the land-
scape [38]. The Cu-Co-Ni mineralization is hosted in Carboniferous limestones and is
primarily linked to epithermal deposits in karst cavities [39,40]. The mineral assemblage
includes chalcopyrite, cobaltite, pyrite, and various secondary minerals, such as chalcocite
and covellite, formed by supergene oxidation. The Aramo mine is characterized by Cu-Ni-
Co sulfides and arsenides formed in a hydrothermal system associated with Late-Variscan
 Figure 2. Aramo study area location in Spain, Aramo plateau detail and geological scheme of the
Remote Sens. 2025, 17, 305 5 of 19
Sierra del Aramo. The legend corresponds to 1. Dolomites, limestones, sandstones, shales, marl-
stones, microconglomerates, and siltstones (Devonian). 2. Quartzite sandstones, marlstones, and
limestones (Upper Devonian–Carboniferous). 3. Massive limestones (Namurian Carboniferous). 4.
tectonic activity [37,39]. Dolomitization is the main alteration process around the fault
Shales, sandstones, limestones, quartzite conglomerates, and coal (Moscovian Carboniferous). 5.
zones, with three primary hypogenic mineralization stages, followed by significant super-
Mass movement deposits, debris deposits, and alluvial deposits (Quaternary). 6. Overthrusts. 7.
gene weathering. The mineralization occurs in veins and lenses along E-W and NE-SW
Faults [38]. Datum WGS 84, UTM zone 30N, scale 1:95,000.
faults, often accompanied by secondary porosity from karstification [40].

3. Materials
3. Materials and
and Methods
Methods
The methodology
The methodology applied
applied involved
involved selecting
selecting suitable
suitable images
images fromfrom Senitnel-2
Senitnel-2 and
and
Landsat-9, selecting
Landsat-9, selecting bands
bands of
of interest,
interest,and
andcreating
creatinga asubset
subsetforfor
thethe
interest area.
interest TheThe
area. dif-
ferent remote sensing methods, band rationing, PCA, and RGB compositions
different remote sensing methods, band rationing, PCA, and RGB compositions were used, were used,
and aathreshold
and thresholdwas
wasapplied
appliedtotothe
theband
bandratio
ratioresults.
results. Then,
Then, the
the results
results were
were critically
critically
analyzed and interpreted. Figure 3 explains the steps followed in the
analyzed and interpreted. Figure 3 explains the steps followed in the study. study.

Figure3.3.Flowchart
Figure Flowchartfor
forthe
thesteps
stepsapplied
appliedin
inthe
themethodology.
methodology.

3.1.
3.1.Data
DataAcquisition
Acquisition
The
The Landsat-9 and
Landsat-9 and Sentinel-2 images utilized
Sentinel-2 images utilized in
in this
this study
studywere
wereacquired
acquiredatatlevel
level2
2 of processing, which means the images were already atmospherically corrected and
of processing, which means the images were already atmospherically corrected and sur-
surface reflectance was already calculated, allowing us to use the images directly
face reflectance was already calculated, allowing us to use the images directly for analysis.
for analysis. Landsat images are free to download and were obtained from earthex-
Landsat images are free to download and were obtained from earthexplorer.usgs.gov (ac-
plorer.usgs.gov (accessed on 12 October 2022). The Landsat-9 image for the Aramo
cessed on 12 October 2022). The Landsat-9 image for the Aramo area
area (LC09_L2SP_203030_20231007_20231008_02_T1) was acquired on 1 October 2023
(LC09_L2SP_203030_20231007_20231008_02_T1) was acquired on 1 October 2023 and for
and for the Ria de Vigo area (LC09_L2SP_204031_20230216_20230310_02_T1) in 16 Febru-
the Ria de Vigo area (LC09_L2SP_204031_20230216_20230310_02_T1) in 16 February 2023.
ary 2023. The Sentinel-2 images also are available for free and were downloaded from
The Sentinel-2 images also are available for free and were downloaded from
browser.dataspace.copernicus.eu (accessed on 12 October 2022). The Sentinel-2 image for
browser.dataspace.copernicus.eu (accessed on 12 October 2022). The Sentinel-2 image for
the Aramo area was acquired on 28 June 2023 (S2B_MSIL2A_20230628T112119_N0509_R037_
the Aramo area was acquired on 28 June 2023
T29TQH_20230628T124435) and for Ria de Vigo (S2A_MSIL2A_20230216T113321_N0509_
(S2B_MSIL2A_20230628T112119_N0509_R037_T29TQH_20230628T124435) and for Ria
R080_T29TNG_20230216T151057) on 16 February 2023.
de Vigo (S2A_MSIL2A_20230216T113321_N0509_R080_T29TNG_20230216T151057) on
16 February
3.2. Sentinel-22023.
and Landsat-9 Bands
Sentinel-2 captures images with a spatial resolution variating from 10 to 60 m, across
3.2. Sentinel-2 and Landsat-9 Bands
13 spectral bands, ranging from visible to SWIR [13]. Bands in the visible and near-
infrared (NIR) regions achieve a 10 m resolution that allows for the finer delineation of
smaller geological targets. The Landsat-9 satellite has a spectral resolution of 11 bands,
encompassing the visible to thermal infrared, and acquires images at the spatial resolution
Remote Sens. 2025, 17, 305 6 of 19

of 30 m in the spectral bands, 15 m in the panchromatic band, and 100 m in the infrared
bands [14]. The thermal bands (Bands 10 and 11) available on Landsat-9 offer the potential
to detect heat-emitting mineral deposits. However, these were not applied in the current
study due to the absence of equivalent bands on Sentinel-2 for comparison. Table 1
compares the bands of the two satellites.

Table 1. Sentinel-2 and Landsat 8/9 comparative bands.

Sentinel-2 * Landsat-8/9
Wavelength Wavelength
Resolution (m) Resolution (m)
Range (nm) Range (nm)
Band 1 60 433–453 30 435–451
Band 2 10 458–523 30 452–512
Band 3 10 543–578 30 533–590
Band 4 10 650–680 30 636–673
Band 5 20 698–713 30 851–879
Band 6 20 733–748 30 1.566–1.651
Band 7 20 773–793 30 2.107–2.294
Band 8 10 785–900 15 503–676
Band 8A 20 855–875 - -
Band 9 60 935–955 30 1.363–1.384
Band 10 60 1.360–1.390 100 10.600–11.190
Band 11 20 1.565–1.655 100 11.500–12.510
Band 12 20 2.100–2.280 - -
* Wavelength ranges for Sentinel-2A. For information on Sentinel-2B, please visit https://sentiwiki.copernicus.
eu/web/s2-mission (accessed on 26 November 2024).

Some Sentinel-2 bands were not applied in the present study, such as B10, which
is designed to detect cirrus clouds, B1, which detects coastal aerosol, and B9, for water
vapor [41]. Similarly, some Landsat-9 were also excluded, namely the thermal infrared
bands (B10 and B11), B1, focused on aerosol, and B9, for detecting cirrus clouds. In the
following applications of band ratios, RGB combinations, and PCA, we considered the most
similar wavelength ranges between the two satellites to make the comparisons. As explored
in the literature, excluding those bands has no impact on mineral alteration studies [41] as
they do not address mineralogical or lithological properties.

3.3. Band Ratio Calculation

Band ratios capable of highlighting possible areas of occurrence of different minerals
were applied according to the equivalent bands of Sentinel-2 and Landsat-9 using a Python
script and the libraries rasterio 1.3.9 and numpy 1.24.4. As described by Sabins [21], ratio
images are created by dividing the pixel values from one band by another band. The images
should be already corrected atmospherically. The different responses of minerals in the
different bands are highlighted when this procedure is applied, corresponding to the highest
ratio values in the histogram, allowing the identification of mineral targets [21]. The band
ratios computed (Table 2) included the following. The iron oxides band ratio, for mineral
signatures in the visible and VINR regions (hematite, goethite, jarosite) [12,16]; Ferric iron
ratio, applied to highlight iron oxides and sensitive to differences in the vegetation cover
and surface materials [21]; Hydrothermal alteration Sabin’s ratio, combination of band
ratios that highlights mineral proprieties observed in hydrothermally altered zones (clays,
silicates and iron minerals) [11]; Hydrothermal alteration* ratio replacing the blue band
for the green band as an experiment to improve the sensibility to the local minerals; Silica
index band ratio [10] can identify silica-rich formations, like quartz veins, that are linked
with the mineralization in the study are; Clay minerals band ratio, uses the SWIR region,
Remote Sens. 2025, 17, 305 7 of 19

that is sensitive to hydroxyl-bearing minerals [42] and the NIR band enhance the separation
of rocks and vegetated areas and discriminates clays [43,44]; Ferrous Iron band ratio for
areas with ferrous iron minerals [45]; Iron oxide index ratio to emphasize areas with strong
differences in the blue and red, highlighting iron oxide minerals [46].

Table 2. Band ratios were selected for the study.

Target Sentinel-2 Landsat-9

Ferric Iron [21] B4/B3 B4/B3
Ferrous Iron [46] (B3 + B12)/(B4 + B8) (B3 + B7)/(B4 + B5)
Hydrothermal alteration a [11] B4/B2 + B6/B11 + B11/12 B4/B2 + B6/B5 + B6/B7
Hydrothermal alteration b B4/B3 + B6/B11 + B11/B12 B4/B3 + B6/B5 + B6/B7
Iron oxide index [12,16] (B4 − B2)/(B4 + B2) (B4 − B2)/(B4 + B2)
Silica index [10] B11/B4 B5/B4
Clay/hydroxyl minerals [42,44] B11/B12 B6/B7
a Sabin’s hydrothermal band ratio. b alternative hydrothermal band ratio.

Other band ratios relevant to the Aramo area include Sentinel-2 combinations for
gossan (B11/B4), ferric oxide (B11/B8a), and ferrous silicate (B12/B11) [42]. These were ap-
plied in a previous study using Sentinel-1 and Sentinel-2 data but did not yield significantly
distinct results [25].

3.4. RGB Compositions

RGB compositions are often used in lithological and alteration mapping to create
false-color images highlighting different geological properties [23]. In this study, we
applied the RGB combinations that produced the best results for the Aramo area in our
previous assessment [25] using Sentinel data to compare with Landsat-9. Additionally, we
tested an RGB combination designed to detect limestones, proposed by [47] for Landsat-
8, and adapted it for Sentinel-2. We also applied an RGB combination for lithological
discrimination [48]. The RGB combinations applied are in Table 3.

Table 3. RGB combinations were selected for the study.

Target Sentinel-2 Landsat-9

Iron/oxides and clay minerals [25] B2, B3, B12 B2, B3, B7
Hydrothermal alteration 1 [25] B8, B11, B12 B5, B6, B7
Hydrothermal alteration 2 [25] B8, B12, B3 B5, B7, B3
Limestone [47] B12/B8, B11/B4, B4/B2 B7/B5, B6/B4, B4/B2
Lithological discrimination [48] B12, B3, B5 B7, B3, B5

In the same previous study, RGB compositions did not produce favorable results for
the Vigo area. Instead, the best results were achieved using PCA applied to Sentinel-2
bands (2 and 4; 2, 4, 8 and 11; 3 and 4). For this study, we tested the same PCA combinations
using the equivalent bands from the Landsat-9 satellite.

3.5. Principal Component Analysis (PCA)

PCA was applied to the selected bands (Table 4). The equivalent number of bands was
used for the number of PCs using ArcGIS PRO version 3.0. The bands were chosen based
on their effectiveness in identifying iron minerals and hydrothermally altered areas. These
combinations were selected from a previous analysis of the study areas using Sentinel
data [25], which identified the most effective bands for each region. In general, we observed
that PCA performed better when using fewer bands with the most significant information
Remote Sens. 2025, 17, 305 8 of 19

than using a larger number of bands. Additionally, we tested PCA combinations shown to
effectively map hydrothermal alterations across multiple sensors [42].

Table 4. Bands used for PCA application and respective study areas.

Study Area Sentinel-2 Landsat-9 Target Number of PCs

Aramo [25] B11, B12 B6, B7 Hydroxyl minerals 2
Aramo/Vigo [25,42] B2, B4 B2, B4 Ferric iron 2
Vigo [25] B2, B4, B8, B11 B2, B4, B5, B6 Hydroxyl minerals 4
Aramo/Vigo [42] B11, B4 B6, B4 Gossan 2
Aramo/Vigo [42] B11, B8 B6, B5 Ferric oxides 2

3.6. Threshold Calculation

A threshold calculation method was applied to identify the anomalies within the
results obtained by the band ratio application. We considered a confidence interval of
95%, meaning that 95% of the data after normalization are expected to be below this
threshold. The 5% of data above this threshold are expected to indicate anomalies repre-
senting areas of interest that may exhibit geological contrasts, hydrothermal alteration, or
mineralized zones.
After computing the band ratios, the data were normalized between 0 and 1 for
comparability of scales. The threshold at a confidence of 95% was applied to the band
ratio results following equation 1, as described in [16]. The threshold (TH) is based on the
mean (M) and standard deviation (SD) of the dataset, as follows:

TH = M + 2 × SD for a 95% confidence interval (1)

With the application of the threshold, the data points exciding the calculated value
(top 5%) were considered anomalies and were plotted for visualization.

4. Results
This section includes the best results obtained with the application of the band ratio
and PCA and RGB techniques for the Ria de Vigo and Aramo study areas.

4.1. Ria de Vigo

Results for Ria de Vigo are presented for the beaches of Barra (Figure 4) and Vao
(Figure 5), which have known occurrences of heavy minerals (latter) or the potential for
finding placer deposits (first). The band ratio analysis produced positive results for iron
oxides, ferrous iron, and hydrothermal alterations. The ferrous iron band ratio showed
high values near the coastline on the water surface but not on the beach areas. The clay
minerals band ratio did not yield any response for the beaches. Both satellites performed
similarly in identifying ferric iron and iron oxides (Figures 4a,d and 5a,d for ferric iron, and
Figures 4b,e and 5b,e for iron oxides).
The hydrothermal alteration band ratio produced mixed results: it worked well with
Sentinel-2 (Figures 4c and 5c) but performed poorly with Landsat-9 (Figures 4f and 5f).
The variance observed across resolutions on the maps was primarily due to the different
resolutions of the two satellites, which are 20 m for Sentinel-2 and 30 m for Landsat-9.
Remote Sens. 2025, 17, x FOR PEER REVIEW 9 of 20
Remote Sens. 2025, 17, 305 9 of 19

(a) (b)

(c) (d)

(e) (f)

Figure 4. Band
Figure ratiosratios
4. Band for the
forBarra Beach.Beach.
the Barra Sentinel-2: (a) ferric(a)
Sentinel-2: iron (purple);
ferric (b) iron oxides
iron (purple); (pink);
(b) iron oxides
a (red). Landsat-9: (d) ferric iron (purple); (e) iron oxides (pink);
(c) (pink); (c) hydrothermal
hydrothermal alteration alteration
a (red). Landsat-9: (d) ferric iron (purple); (e) iron oxides (pink); (f) hy-
(f) hydrothermal alteration b (red). Datum WGS 84 UTM zone 29 N, scale 1:15,000. a b according
drothermal alteration b (red). Datum WGS 84 UTM zone 29 N, scale 1:15,000. a and b and according to
to Table
Table 2. 2.

PCA combinations were not effective in distinguishing specific zones within the beach
The hydrothermal alteration band ratio produced mixed results: it worked well with
areas of Ria de Vigo. However, they successfully differentiated the sandy beach zones
Sentinel-2 (Figures 4c and 5c) but performed poorly with Landsat-9 (Figures 4f and 5f).
from vegetation and mussel farms, as noted in a previous study [19]. Similarly, some RGB
The variance observed across resolutions on the maps was primarily due to the different
combinations from [25] highlighted vegetation and mussel farms but failed to differentiate
resolutions of the two satellites, which are 20 m for Sentinel-2 and 30 m for Landsat-9.
features on the beach face effectively.
PCA combinations were not effective in distinguishing specific zones within the
beach areas of Ria de Vigo. However, they successfully differentiated the sandy beach
zones from vegetation and mussel farms, as noted in a previous study [19]. Similarly,
 some RGB combinations from [25] highlighted vegetation and mussel farms but failed to
Remote Sens. 2025, 17, 305 10 of 19
differentiate features on the beach face effectively.

(a) (b)

(c) (d)

(e) (f)

Figure
Figure5. 5.
Band
Bandratio
ratioresults
resultsfor
for Sentinel-2 for Vao
Sentinel-2 for Vaobeach:
beach:(a)(a)ferric
ferric iron
iron (purple);
(purple); (b) (b)
ironiron oxides
oxides (pink),
(c) hydrothermal alteration a (red). Landsat-9 for Vao beach (d) ferric iron iron
(purple); (e) iron oxides
(pink), (c) hydrothermal alteration a (red). Landsat-9 for Vao beach (d) ferric (purple); (e) iron
(pink), (f) hydrothermal alteration b (red). Datum WGS 84 UTM Zone 29 N, a anda b
oxides (pink), (f) hydrothermal alteration b (red). Datum WGS 84 UTM Zone 29scale 1:7,500.
N, scale 1:7,500.
andaccording
b accordingto Table 2. 2.
to Table

4.2. Aramo
4.2. Aramo
The same band ratios applied to Ria de Vigo were also used in the Aramo area, except
The same band ratios applied to Ria de Vigo were also used in the Aramo area, except
for the silica index, which was applied only in Aramo. In general, the band ratios produced
for the silica index, which was applied only in Aramo. In general, the band ratios pro-
limited results with Landsat-9, detecting only ferric iron (Figure 6a) and ferrous iron
duced limited results with Landsat-9, detecting only ferric iron (Figure 6a) and ferrous
(Figure 6b) anomalies. Sentinel-2, however, identified additional features, including iron
iron (Figure 6b) anomalies. Sentinel-2, however, identified additional features, including
oxides (Figure 6e) and hydrothermal alterations (Figure 6f). The clay minerals band ratio
did not generate meaningful results with either satellite.
Remote Sens. 2025, 17, x FOR PEER REVIEW 11 of 20

Remote Sens. 2025, 17, 305 11 of 19

iron oxides (Figure 6e) and hydrothermal alterations (Figure 6f). The clay minerals band
ratio did not generate meaningful results with either satellite.

(a) (b)

(c) (d)

(e) (f)
Landsat-9band
Figure6.6.Landsat-9
Figure bandratios
ratioshighlighting
highlighting (a)
(a) ferric
ferric iron
iron (pink); (b) iron oxides (orange).
(orange). Sentinel-2
Sentinel-
band ratios highlighting
2 band ratios highlighting (c) ferric iron (pink); (d) iron oxides (orange), (e) ferrous
ferric iron (pink); (d) iron oxides (orange), (e) ferrous iron iron (purple)
(f) hydrothermal a
ratio (dark a
(purple) (f) hydrothermal ratio pink).
a (darkDatum WGS 84,
pink). Datum WGSUTM 84,zone
UTM 30zone
N, scale
30 N,1:95,000. according
scale 1:95,000. a ac-

to Table 2.
cording to Table 2.

Most detected alterations are associated with iron minerals, highlighting the same
Most detected alterations are associated with iron minerals, highlighting the same
areas of the Aramo plateau when using Sentinel-2 data. Landsat-9 identified fewer anoma-
areas of the Aramo plateau when using Sentinel-2 data. Landsat-9 identified fewer anom-
lies compared to Sentinel-2. The silica index band ratio did not detect quartz vein-related
alies compared to Sentinel-2. The silica index band ratio did not detect quartz vein-related
alterations as expected but instead highlighted topographic peaks and vegetated areas for
both satellites.
 with potential occurrences of iron oxides and clay minerals in purple (Figure 7a). This
combination also differentiates vegetated areas from non-vegetated areas well. The corre-
sponding Landsat-9 RGB 2-3-7 (Figure 7b) yields a similar result, primarily separating
Remote Sens. 2025, 17, 305 vegetation from exposed rock in purple tones. RGB combinations designed for hydrother-12 of 19
mal alteration using Sentinel-2 (bands 8, 11, and 12; bands 8, 12, and 3) highlight the same
regions in greenish-blue tones (Figure 7c). The equivalent Landsat-9 combinations (bands
Related
5-6-7 and 5-7-3)toproduce
the RGBsimilar
combinations, Sentinel-2
contrasts between RGB(Figure
zones 2-3-127d).
effectively highlights areas
withFor
potential
limestone detection, the RGB combination from [47] showspurple
occurrences of iron oxides and clay minerals in (Figure
limestone 7a). This
in orange
combination
with Sentinel-2also differentiates
(Figure 7e) and in vegetated
yellow withareas from non-vegetated
Landsat-9 areas well.
(Figure 7f). Vegetation appearsThe cor-
responding
in bright green Landsat-9 RGB However,
in both cases. 2-3-7 (Figurethe 7b) yields
results are aless
similar
clear result,
than inprimarily separating
[47] because, in
their case, it was
vegetation fromapplied
exposed to rock
an arid
in area, and
purple in theRGB
tones. present study, it wasdesigned
combinations applied to
fora much
hydrother-
denser vegetated
mal alteration area.Sentinel-2
using Similarly, (bands
the RGB8,combination from [48]
11, and 12; bands forand
8, 12, lithological discrim-
3) highlight the same
ination (Figure 7e,f) primarily distinguishes the exposed Aramo plateau
regions in greenish-blue tones (Figure 7c). The equivalent Landsat-9 combinations (bands from the sur-
rounding
5-6-7 andvegetation but does
5-7-3) produce not provide
similar contrastsdetailed
between differentiation.
zones (Figure 7d).

(a) (b)

Remote Sens. 2025, 17, x FOR PEER REVIEW 13 of 20

(c) (d)

(e) (f)
Resultsfor
Figure7.7.Results
Figure forAramo
Aramousing
usingSentinel-2
Sentinel-2RGB
RGBcombination
combinationwithwithbands:
bands:2-3-12
2-3-12(a),
(a),Landsat-9
Landsat-9 RGB
combination
RGB withwith
combination bands: 2-3-7
bands: (b).(b).
2-3-7 Sentinel-2 RGB
Sentinel-2 RGBcombination
combinationwith bands:
with 8-11-12
bands: (c),(c),
8-11-12 Landsat-9
Land- RGB
sat-9 RGB combination with bands: 5-6-7 (d). Sentinel-2 RGB combination with bands: 12/8, 11/4, Landsat-9
combination with bands: 5-6-7 (d). Sentinel-2 RGB combination with bands: 12/8, 11/4, 4/2 (e), 4/2
RGB
(e), combination
Landsat-9 with bands: 7/5,
RGB combination with 6/4, 4/27/5,
bands: (f).6/4,
Datum WGS
4/2 (f). 84, UTM
Datum WGSzone 30 N,zone
84, UTM scale301:95,000.
N, scale
1:95,000.

Multiple band ratios from a previous study [25] were not replicated here because
they did not reveal distinct features in the Aramo plateau. Additionally, the limited ex-
posed area and the presence of dense vegetation reduce the effectiveness of RGB combi-
Remote Sens. 2025, 17, 305 13 of 19

For limestone detection, the RGB combination from [47] shows limestone in orange
with Sentinel-2 (Figure 7e) and in yellow with Landsat-9 (Figure 7f). Vegetation appears
in bright green in both cases. However, the results are less clear than in [47] because, in
their case, it was applied to an arid area, and in the present study, it was applied to a
much denser vegetated area. Similarly, the RGB combination from [48] for lithological
discrimination (Figure 7e,f) primarily distinguishes the exposed Aramo plateau from the
surrounding vegetation but does not provide detailed differentiation.
Multiple band ratios from a previous study [25] were not replicated here because they
did not reveal distinct features in the Aramo plateau. Additionally, the limited exposed
area and the presence of dense vegetation reduce the effectiveness of RGB combinations for
lithological differentiation.
Selective PCA was used to isolate target spectral signatures of alteration minerals.
The bands in each subset were selected based on the spectral properties of the target
minerals. A careful analysis of eigenvalues and eigenvectors was conducted following the
“Crosta technique” by evaluating the eigenvectors’ magnitude and sign (positive/negative),
Remote Sens. 2025, 17, x FOR PEER REVIEW 14 of 20
considering the spectral properties of the target minerals [49]. The eigenvalues for the PCs
used in Figure 8 to target hydroxyl minerals are displayed in Table 5. Hydroxyl minerals
have an important absorption covered by band 7 for Landsat-9 and band 12 for Sentinel-
Table 5. PCA Eigenvalues for the PCA results are presented in the paper.
2 while displaying high reflectance within bands 6 and 11 of Landsat-9 and Sentinel-2,
respectively [21,50]. The strong correlation between the pair Eigenvalues
of bands in PC1 (Table 5)
Band PC1
can be attributed to the overall scene brightness or albedo [49]. In PC2, the pair PC2 of bands
moderate to strong loadingsB6
displays Landsat-9 0.83718
with opposite signs for both satellites.−0.54692
For Sentinel-2,
B7 0.54692 0.83718
hydroxyl minerals will be mapped in high pixel values in PC2 due to the positive input
of band 11 (reflective band) and the B1e1
negative input of0.71758
band 12 (absorption0.69647
band). In the
Sentinel-2
B12 0.69647 −0.71758
case of Landsat-9, due to the negative input of band 6 (reflective band) and the positive
input of band 7 (absorption band), hydroxyl minerals will be mapped in low pixel values
Selective
in PC2. PCA
This can be applied
correctedtothrough
the other bandmathematical
simple subsets has band
not provided significative
operations, specificallyre-
by
sults, mostly highlighting some vegetation variation, while selective PCA combination
multiplying PC2 of Landsat-9 by -1 and ensuring the target areas correspond to high pixel for
ferric
valuesoxides
[49]. has not produced statistically significant PCs (eigenvalues < 0.3).

(a) (b)
(a)Landsat-9
Figure8.8.(a)
Figure Landsat-9PCA
PCA result
result with
with bands
bands 6 and
6 and 7, PC2.
7, PC2. (b)(b) Sentinel-2
Sentinel-2 PCAPCA result
result withwith bands
bands 11
11 and 12, PC2. Red pixels indicate hydroxyl mineral targets. Datum WGS 84, UTM
and 12, PC2. Red pixels indicate hydroxyl mineral targets. Datum WGS 84, UTM zone 30 N, scale zone 30 N,
scale 1:95,000.
1:95,000.

PCA applied to Sentinel-2 bands 11 and 12 (Figure 8b) highlighted similar areas to the
5. Discussion
ones using the Sabins hydrothermal band ratio (Figure 6f) and ferrous iron (Figure 6e) for
In thissatellite.
the same study, we Theexplored thealso
results are capabilities of two
aligned with satellites,
the Sentinel-2
one obtained with and Landsat-9,
the RGB combi-
in the application of classical remote sensing techniques, band ratios, PCA, and RGB com-
positions, for targeting CRM potential areas using two distinct geological settings as study
cases. In the Ria de Vigo study area, the results obtained from both satellites, Landsat-9
and Sentinel-2, were quite similar. The primary variation stemmed from differences in
spatial resolution, with Sentinel-2 having a higher resolution, providing a more detailed
Remote Sens. 2025, 17, 305 14 of 19

nation for hydrothermal alterations (Figure 7c). However, the equivalent combination for
Landsat-9 (Figure 8a) does not yield similar results.

Table 5. PCA Eigenvalues for the PCA results are presented in the paper.

Eigenvalues
Band PC1 PC2
B6 0.83718 −0.54692
Landsat-9 B7 0.54692 0.83718
B11 0.71758 0.69647
Sentinel-2 B12 0.69647 −0.71758

Selective PCA applied to the other band subsets has not provided significative results,
mostly highlighting some vegetation variation, while selective PCA combination for ferric
oxides has not produced statistically significant PCs (eigenvalues < 0.3).

5. Discussion
In this study, we explored the capabilities of two satellites, Sentinel-2 and Landsat-9,
in the application of classical remote sensing techniques, band ratios, PCA, and RGB com-
positions, for targeting CRM potential areas using two distinct geological settings as study
cases. In the Ria de Vigo study area, the results obtained from both satellites, Landsat-9 and
Sentinel-2, were quite similar. The primary variation stemmed from differences in spatial
resolution, with Sentinel-2 having a higher resolution, providing a more detailed analysis
of certain features. An alternative band ratio for hydrothermal alteration proposed in this
study was proved effective when applied to Landsat-9 data but yielded less reliable results
for Sentinel-2. This difference may be tied to each satellite’s distinct spectral sensitivity,
capturing similar but different wavelength ranges. However, the overall capability of
detecting mineral anomalies in the Ria de Vigo region was more limited compared to
the Aramo area. The difficulty in obtaining significant results when using PCA or RGB
combinations in Ria de Vigo reflects the challenge in detecting mineral placer deposits
using remote sensing methods. These limitations were already observed in our previous
study exploring Sentinel data [25].
In the Aramo region, the clay and hydrothermal band ratios did not produce sig-
nificant results. This outcome could be attributed to the geological characteristics of the
area. Previous fieldwork has indicated that clay minerals in the region exhibit limited
and localized expressions, which may not be detectable at the satellite’s resolution and
spectral ranges. Notwithstanding, the presence of goethite, hematite, and calcite was docu-
mented [34], suggesting that while alteration minerals are present, their spectral signals
might overlap or be reduced in satellite data.
As also previously noted by [15], Sentinel-2 offers better spatial resolution compared
to Landsat-9. This advantage likely contributed to the more satisfactory results in certain
areas, particularly in Aramo. A comparative study [48] reached a similar conclusion where
Sentinel-2 outperformed Landsat-8 in mineral prospectivity analyses using machine learn-
ing techniques. The finer spatial resolution of Sentinel-2 allows for better discrimination
of small-scale alteration zones and mineral distributions, which is crucial for regions with
subtle geological expressions. Previous studies already noticed that Sentinel-2 can provide
greater precision in detecting subtle spectral variations, given its higher signal-to-noise
ratio (SNR) [9,13].
In a previous exploratory study involving the Aramo area, decision tree algorithms
were applied to Landsat-9 data to identify alteration zones [51]. The results of that super-
vised machine learning approach aligned with the results obtained with the application
Remote Sens. 2025, 17, 305 15 of 19

of PCA to Sentinel-2 data in the current study. This agreement between methodologies
suggests a consistency in detecting alteration zones, and the combination of dimensionality
reduction techniques, such as PCA and ICA (Independent Component Analysis), is a
common approach to mineral exploration. These findings reinforce that a harmonized
approach to integrating Sentinel-2 and Landsad-8/9 data is favorable, and the data from
both systems should be integrated into CRM exploration workflows, as machine learning
can be trained on datasets incorporating both satellites to obtain better predictive insights.

6. Perspectives
The application of EO methods in diverse areas, including onshore and beach zones,
has demonstrated its potential but has also highlighted significant challenges. In the Ria
de Vigo study area, which hosts heavy mineral placer deposits, traditional remote sensing
techniques seem to be of low effectiveness for detecting this type of deposit. Despite
this, band ratio combinations were promising, particularly in identifying iron oxides and
hydrothermal alterations. These preliminary results suggest that further exploration of
temporal analyses focused on band ratio performance could enhance detection capabilities
in similar settings.
Future studies should incorporate methodologies, including spectral libraries and ma-
chine learning algorithms, as those approaches improved the detection and quantification
of mineral placers in recent studies [52–54]. Incorporating spectral libraries for specific
mineral assemblages in placer deposits could enable more precise identification of target
features [52,54]. Moreover, machine learning can identify subtle patterns and relationships
that may not be evident through traditional methods [53]. For the Aramo area, machine
learning approaches have started to be developed, but future studies must continue to
explore the detection of alteration zones integrating data such as radar and hyperspectral
resolution satellite data [51,55–57].
Concerning the topic of CRM exploration in coastal areas, the next steps include
exploring methods to detect mineral placers in shallow water zones. This was not in
the scope of the current study. However, developing methodologies to detect mineral
placers in such environments should be a priority for future research. We consider that an
important step in this process will be acquiring a comprehensive knowledge of the mineral
placer characteristics in the study areas, as well as the geological and environmental factors
influencing their formation and distribution, and the methodologies to correct the water
column influence should be employed. Correcting the water column’s influence is essential
to achieve reliable detection. This involves addressing atmospheric disturbances, water
absorption, and scattering effects that distort optical bands. These phenomena significantly
limit the penetration of electromagnetic signals, with only shorter wavelengths reaching the
seabed at shallow depths [58]. Correction models must account for the optical complexity
of coastal waters, including variable turbidity, suspended sediments, and chlorophyll
concentrations [59].
The methodologies explored in the present study have potential for application be-
yond the European framework, as they are already broadly applied to diverse geographic
contexts [15,22]. The application of these methodologies reduces reliance on invasive ex-
ploration techniques, aligning with sustainable development goals. Future research should
focus on refining water column correction models, integrating these with machine learning
techniques for CRM detection.

7. Conclusions
This study highlights the potential and limitations of using Sentinel-2 and Landsat-9
satellites for targeting alteration areas related to CRM in two geologically distinct areas: the
Remote Sens. 2025, 17, 305 16 of 19

Ria de Vigo region, containing heavy mineral marine placer deposits, and the Aramo area,
which hosts epithermal Co-Ni deposits. This is the first comparative study of those satellites
focusing on these specific deposit types and provides a novel application of tailored band
ratios or PCA combinations for CRM exploration. The findings demonstrate that while
traditional remote sensing techniques like band ratios, PCA, and RGB compositions can be
effective for identifying mineralogical and alteration zones, their success is dependent on
the geological context, the specific mineral targets, and the spectral and spatial characteris-
tics of the satellites. The contrast between the geological features and the targets related
to the mineral alteration is a key factor for the effectiveness of the methodologies and
deposits that exhibit more subtle or small occurrences present additional challenges. Areas
with the presence of dense vegetation or with the influence of tide and atmospheric effects
related to coastal areas are also limiting factors. Future studies in this area should include
the integration of radar data, which can allow the identification of the structural features
related to the mineralization and provide insights on the terrain [55], as well as hyperspec-
tral data, that can detect more subtle differences than multispectral differences, despite
implying additional processing to data, and local ground spectral library, an approach
that has been successfully applied in zones with similar lithologies [60,61]. Sentinel-2
consistently outperformed Landsat-9, allowing more detailed mapping due to its higher
spatial resolution and more spectral bands, especially in the Aramo study area, being more
indicated for local scale studies, while Landsat is more favorable for historical analysis
and regional scale. The Ria de Vigo region presented significant challenges for remote
sensing-based detection and developing methodologies to address these specific challenges,
such as advanced water column correction models and spectral libraries, is essential for
future research. This study provides ground for further advancements in remote sensing
methodologies focused on non-traditional mineral deposits like Co-Ni and marine placers.
Such technologies can contribute to a more sustainable and secure supply of CRMs in the
EU, supporting the green transition and energetic resilience in the European continent. The
development of remote sensing methods can also contribute to the sustainability of mining
activities and environmental monitoring [62,63]. Future directions for this study include
integrating machine learning and hyperspectral data, as well as finer spatial resolution.

Author Contributions: Conceptualization, M.C., A.C.T., and J.C.-F.; investigation M.C.; methodology,
M.C.; software, M.C.; writing—original draft preparation, M.C.; formal analysis M.C.; visualization
M.C.; writing—review and editing, M.C., J.C.-F., F.J.G., and A.C.T.; supervision, A.C.T.; funding
acquisition, A.C.T.; project administration A.C.T. All authors have read and agreed to the published
version of the manuscript.

Funding: This study is funded by the European Union under grant agreement no. 101091616 (https:
//doi.org/10.3030/101091616 (accessed on 19 November 2024)) project S34I—Secure and Sustainable
Supply of Raw Materials for EU Industry, under topic HORIZON-CL4-2022-RESILIENCE-01-08 - Earth
observation technologies for the mining life cycle in support of EU autonomy and transition to a climate-
neutral economy (RIA). This work is also supported by national funds through FCT—Fundação para a
Ciência e Tecnologia, I.P., in the framework of the UIDB/04683 and UIDP/04683—Instituto de Ciências
da Terra programs.

Data Availability Statement: The satellite images used in the present study are available for down-
load. Landsat images are free to download and were obtained from https://earthexplorer.usgs.gov/
(accessed on 21 November 2023). The Sentinel- 2 images are available for free and were down-
loaded from https://browser.dataspace.copernicus.eu/ (accessed on 21 November 2023). The de-
rived/processed data generated in the scope of this study and the S34I project are openly available in
Zenodo at: https://doi.org/10.5281/zenodo.14652200 (accessed on 21 November 2023).
Remote Sens. 2025, 17, 305 17 of 19

Acknowledgments: Morgana Carvalho is pursuing a PhD degree supported by FCT (Grant:

2024.03620.BD).

Conflicts of Interest: The authors declare no conflicts of interest. The funders had no role in the design
of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript;
or in the decision to publish the results. Views and opinions expressed are, however, those of the
author(s) only and do not necessarily reflect those of the European Union or HADEA. Neither the
European Union nor the granting authority can be held responsible for them.

References
1. Hool, A.; Helbig, C.; Wierink, G. Challenges and opportunities of the European critical raw materials act. Miner. Econ. 2023,
37, 661–668. [CrossRef]
2. Kamran, M.; Raugei, M.; Hutchinson, A. Critical elements for a successful energy transition: A systematic review. Renew. Sustain.
Energy Transit. 2023, 4, 100068. [CrossRef]
3. Srivastava, N. Strengthening European Energy Security and Resilience through Minerals. Eur. Energy Environ. Law Rev. 2024,
33, 35–45. [CrossRef]
4. Komnitsas, K.; Lazos, I.; Eerola, T. Energy Transition Metals: Future Demand and Low-Carbon Processing Technologies. Mater.
Proc. 2023, 15, 56. [CrossRef]
5. Tercero Espinoza, L.; Kroll, H.; Stijepic, D.; Bettin, S.; Favreuille, S.; Mesbahi, Z.; Udrea, T.; Forsberg, E.-M.; Pauna, V.; Baxter, J.
The Role of Research and Innovation in Ensuring a Safe and Sustainable Supply of Critical Raw Materials in the EU; European Parliament:
Brussels, Belgium, 2024.
6. Atlas, A. Mapping Mining to the Sustainable; World Economic Forum: Cologny/Geneva, Switzerland, 2016.
7. Persello, C.; Wegner, J.D.; Hänsch, R.; Tuia, D.; Ghamisi, P.; Koeva, M.; Camps-Valls, G. Deep Learning and Earth Observation to
Support the Sustainable Development Goals: Current approaches, open challenges, and future opportunities. IEEE Geosci. Remote
Sens. Mag. 2022, 10, 172–200. [CrossRef]
8. Berra, E.F.; Fontana, D.C.; Yin, F.; Breunig, F.M. Harmonized Landsat and Sentinel-2 Data with Google Earth Engine. Remote Sens.
2024, 16, 2695. [CrossRef]
9. Claverie, M.; Ju, J.; Masek, J.G.; Dungan, J.L.; Vermote, E.F.; Roger, J.-C.; Skakun, S.V.; Justice, C. The Harmonized Landsat and
Sentinel-2 surface reflectance data set. Remote Sens. Environ. 2018, 219, 145–161. [CrossRef]
10. Aisabokhae, J.; Osazuwa, I. Radiometric mapping and spectral based classification of rocks using remote sensing data analysis:
The Precambrian basement complex, NW Nigeria. Remote Sens. Appl. Soc. Environ. 2021, 21, 100447. [CrossRef]
11. Osinowo, O.O.; Gomy, A.; Isseini, M. Mapping hydrothermal alteration mineral deposits from Landsat 8 satellite data in Pala,
Mayo Kebbi Region, Southwestern Chad. Sci. Afr. 2021, 11, e00687. [CrossRef]
12. Liu, F.; Wu, X.; Sun, H.; Guo, Y. Alteration Information Extraction by Applying Synthesis Processing Techniques to Landsat ETM+
Data: Case Study of Zhaoyuan Gold Mines, Shandong Province, China. J. China Univ. Geosci. 2007, 18, 72–76. [CrossRef]
13. Drusch, M.; Del Bello, U.; Carlier, S.; Colin, O.; Fernandez, V.; Gascon, F.; Hoersch, B.; Isola, C.; Laberinti, P.; Martimort, P.;
et al. Sentinel-2: ESA’s Optical High-Resolution Mission for GMES Operational Services. Remote Sens. Environ. 2012, 120, 25–36.
[CrossRef]
14. Lulla, K.; Nellis, M.D.; Rundquist, B.; Srivastava, P.K.; Szabo, S. Mission to earth: LANDSAT 9 will continue to view the world.
Geocarto Int. 2021, 36, 2261–2263. [CrossRef]
15. Adiri, Z.; Lhissou, R.; El Harti, A.; Jellouli, A.; Chakouri, M. Recent advances in the use of public domain satellite imagery for
mineral exploration: A review of Landsat-8 and Sentinel-2 applications. Ore Geol. Rev. 2020, 117, 103332. [CrossRef]
16. Nait Amara, B.; Aissa, D.E.; Maouche, S.; Braham, M.; Machane, D.; Guessoum, N. Hydrothermal alteration mapping and
structural features in the Guelma basin (Northeastern Algeria): Contribution of Landsat-8 data. Arab. J. Geosci. 2019, 12, 94.
[CrossRef]
17. Nafigin, I.O.; Ishmukhametova, V.T.; Ustinov, S.A.; Minaev, V.A.; Petrov, V.A. Geological and Mineralogical Mapping Based on
Statistical Methods of Remote Sensing Data Processing of Landsat-8: A Case Study in the Southeastern Transbaikalia, Russia.
Sustainability 2022, 14, 9242. [CrossRef]
18. Nwaila, G.T.; Zhang, S.E.; Bourdeau, J.E.; Ghorbani, Y.; Carranza, E.J.M. Artificial intelligence-based anomaly detection of the
Assen iron deposit in South Africa using remote sensing data from the Landsat-8 Operational Land Imager. Artif. Intell. Geosci.
2022, 3, 71–85. [CrossRef]
19. Shevyrev, S.; Carranza, E.J.M. Application of maximum entropy for mineral prospectivity mapping in heavily vegetated areas of
Greater Kurile Chain with Landsat 8 data. Ore Geol. Rev. 2022, 142, 104758. [CrossRef]
20. Gerardo, R.; de Lima, I.P. Comparing the Capability of Sentinel-2 and Landsat 9 Imagery for Mapping Water and Sandbars in the
River Bed of the Lower Tagus River (Portugal). Remote Sens. 2023, 15, 1927. [CrossRef]
Remote Sens. 2025, 17, 305 18 of 19

21. Sabins, F.F. Remote sensing for mineral exploration. Ore Geol. Rev. 1999, 14, 157–183. [CrossRef]
22. Pour, A.B.; Park, Y.; Park, T.-Y.S.; Hong, J.K.; Hashim, M.; Woo, J.; Ayoobi, I. Regional geology mapping using satellite-based
remote sensing approach in Northern Victoria Land, Antarctica. Polar Sci. 2018, 16, 23–46. [CrossRef]
23. Han, T.; Nelson, J. Mapping hydrothermally altered rocks with Landsat 8 imagery: A case study in the KSM and Snow field
zones, northwestern British Columbia. Br. Columbia Geol. Surv. Pap. 2015, 1, 103–112.
24. Imbroane, M.A.; Melenti, C.; Gorgan, D. Mineral explorations by Landsat image ratios. In Proceedings of the Ninth Inter-
national Symposium on Symbolic and Numeric Algorithms for Scientific Computing (SYNASC 2007), Timisoara, Romania,
26–29 September 2007; IEEE: New York, NY, USA; pp. 335–340. [CrossRef]
25. Cardoso-Fernandes, J.; Carvalho, M.; Azzalini, A.; Rodrigues, G.; Monteiro, G.; Lima, A.; Teodoro, A.C. Sentinel data for critical
raw materials (CRM) exploration: First results of the S34I project. In Proceedings of the SPIE Remote Sensing, Amsterdam, The
Netherlands, 19 October 2023; SPIE: San Francisco, CA, USA, 2023. [CrossRef]
26. Vilas, F.; Bernabeu, A.; Méndez, G. Sediment distribution pattern in the Rias Baixas (NW Spain): Main facies and hydrodynamic
dependence. J. Mar. Syst. 2005, 54, 261–276. [CrossRef]
27. González, F.J.; Medialdea, T.; Schiellerup, H.; Zananiri, I.; Ferreira, P.; Somoza, L.; Monteys, X.; Alcorn, T.; Marino, E.; Lobato,
A.B.; et al. MINDeSEA: Exploring seabed mineral deposits in European seas, metallogeny and geological potential for strategic
and critical raw materials. Geol. Soc. Lond. Spec. Publ. 2023, 526, 289–317. [CrossRef]
28. Hou, B.; Keeling, J.; Van Gosen, B.S. Geological and Exploration Models of Beach Placer Deposits, Integrated from Case-Studies
of Southern Australia. Ore Geol. Rev. 2017, 80, 437–459. [CrossRef]
29. Cronan, D.S. Handbook of Marine Mineral Deposits, 1st ed.; CRC Press: New York, NY, USA, 2000.
30. Bernabeu, A.M.; Lersundi-Kanpistegi, A.V.; Vilas, F. Gradation from oceanic to estuarine beaches in a ría environment: A case
study in the Ría de Vigo. Estuar. Coast. Shelf Sci. 2012, 102–103, 60–69. [CrossRef]
31. Alvarez, I.; Decastro, M.; Gomez-Gesteira, M.; Prego, R. Inter-and intra-annual analysis of the salinity and temperature evolution
in the Galician Rías Baixas–ocean boundary (northwest Spain). J. Geophys. Res. Ocean. 2005, 110, C04008. [CrossRef]
32. Richard Gent, M.; Menéndez Alvarez, M.; José María García Iglesias, J.; TaraÑo Álvarez, J. Offshore Occurrences of Heavy-Mineral
Placers, Northwest Galicia, Spain. Mar. Georesources Geotechnol. 2005, 23, 39–59. [CrossRef]
33. Ng-Cutipa, W.L.; González, F.J.; Lobato, A.; Zananiri, I.; Teodoro, A. Titanium, Zirconium and Rare Earth Element Placer Deposits
in Coastal Environments of Rías Baixas (Galicia, NW Spain). In Proceedings of the Underwater Mineral Conference, Rotterdam,
The Netherlands, 1–6 October 2023; International Marine Minerals Society: Petersburg, FL, USA; pp. 80–83. [CrossRef]
34. Ng-Cutipa, W.L.; Lobato, A.; González, F.; Georgalas, G.; Zananiri, I.; Cardoso-Fernandes, J.; Carvalho, M.; Azzalini, A.; Araújo,
B.L.; Teodoro, A.C. Airborne images integration for a first cartographic approximation of critical raw-materials-rich placers on
Santa Marta Beach (Ría de Vigo, NW Spain). In Proceedings of the Remote Sensing, Edinburgh, UK, 13 November 2024; SPIE: San
Francisco, CA, USA. [CrossRef]
35. Prego, R.; Caetano, M.; Vale, C.; Marmolejo-Rodríguez, J. Rare earth elements in sediments of the Vigo Ria, NW Iberian Peninsula.
Cont. Shelf Res. 2009, 29, 896–902. [CrossRef]
36. Aller, J. La estructura geológica de la Sierra del Aramo (Zona Cantábrica, NO de España). Trab. De Geol. 1983, 19, 3–15. [CrossRef]
37. Paniagua, A.; Loredo, J.; Garcia Iglesias, J. Epithermal (Cu-Co-Ni) mineralization in the Aramo mine (Cantabrian Mountains,
Spain): Correlation between paragenetic and fluid inclusion data. Bull. De Minéralogie 1988, 111, 383–391. [CrossRef]
38. Beato Bergua, S.; Poblete Piedrabuena, M.Á.; Rodríguez Pérez, C.; Marino Alfonso, J.L. Geomorphology of the Sierra del Aramo
(Asturian Central Massif, Cantabrian Mountains, NW Spain). J. Maps 2019, 15, 590–600. [CrossRef]
39. Álvarez, R.; Ordóñez, A.; Pérez, A.; De Miguel, E.; Charlesworth, S. Mineralogical and environmental features of the asturian
copper mining district (Spain): A review. Eng. Geol. 2018, 243, 206–217. [CrossRef]
40. Alonso Sánchez, V. Caracterización Estructural de la Mineralización y Alteración de Cu-Co-Ni Asociada a Rocas Carbonatadas,
Mina de Aramo, Oroclinal Cantábrico, Asturias, Norte de España. Master’s Thesis, Universidad Internacional de Andalucía,
Seville, Andalusia, Spain, 2020.
41. Hawryło, P.; W˛eżyk, P. Predicting Growing Stock Volume of Scots Pine Stands Using Sentinel-2 Satellite Imagery and Airborne
Image-Derived Point Clouds. Forests 2018, 9, 274. [CrossRef]
42. Shebl, A.; Abdellatif, M.; Elkhateeb, S.O.; Csámer, Á. Multisource Data Analysis for Gold Potentiality Mapping of Atalla Area
and Its Environs, Central Eastern Desert, Egypt. Minerals 2021, 11, 641. [CrossRef]
43. Clark, R.N.; Swayze, G.A.; Livo, K.E.; Kokaly, R.F.; Sutley, S.J.; Dalton, J.B.; McDougal, R.R.; Gent, C.A. Imaging spectroscopy:
Earth and planetary remote sensing with the USGS Tetracorder and expert systems. J. Geophys. Res. Planets 2003, 108, 7–12.
[CrossRef]
44. Hoatson, D. Exploration strategies for Precambrian layered mafic—Ultramafic intrusions in the East Kimberley. AGSO Res. Newsl.
2000, 32, 7–12. Available online: https://pid.geoscience.gov.au/dataset/ga/90545 (accessed on 15 October 2024).
Remote Sens. 2025, 17, 305 19 of 19

45. Mars, J.C.; Rowan, L.C. Regional mapping of phyllic-and argillic-altered rocks in the Zagros magmatic arc, Iran, using Advanced
Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data and logical operator algorithms. Geosphere 2006,
2, 161–186. [CrossRef]
46. Rowan, L.C.; Mars, J.C. Lithologic mapping in the Mountain Pass, California area using Advanced Spaceborne Thermal Emission
and Reflection Radiometer (ASTER) data. Remote Sens. Environ. 2003, 84, 350–366. [CrossRef]
47. Courba, S.; Youssef, H.; Jamal, A.; Abdessalam, O.; Mohamed, E.A.; Larbi, B.; Assia, I.; Zineb, A.; Slimane, S.; Lahcen, O.; et al.
Litho-structural and hydrothermal alteration mapping for mineral prospection in the Maider basin of Morocco based on remote
sensing and field investigations. Remote Sens. Appl. Soc. Environ. 2023, 31, 100980. [CrossRef]
48. Abd El-Wahed, M.; Zoheir, B.; Pour, A.B.; Kamh, S. Shear-Related Gold Ores in the Wadi Hodein Shear Belt, South Eastern Desert
of Egypt: Analysis of Remote Sensing, Field and Structural Data. Minerals 2021, 11, 474. [CrossRef]
49. Loughlin, W. Principal component analysis for alteration mapping. Photogramm. Eng. Remote Sens. 1991, 57, 1163–1169.
50. Ali, A.; Pour, A. Lithological mapping and hydrothermal alteration using Landsat 8 data: A case study in ariab mining district,
red sea hills, Sudan. Int. J. Basic Appl. Sci. 2014, 3, 199–208. [CrossRef]
51. Carvalho, M.; Cardoso-Fernandes, J.; Azzalini, A.; Williams, V.; Lima, A.; Teodoro, A.C. Decision tree algorithms applied to detect
mineral alteration related to cobalt-nickel mineralisation in Asturias: Insights of S34I project. In Proceedings of the SPIE Remote
Sensing, Edinburgh, UK, 13 November 2024; SPIE: San Francisco, CA, USA; Volume 13197. [CrossRef]
52. Prates Hallal, G.; de Almeida Espinoza, J.M.; Veettil, B.K.; Porcher, C.C.; Oliveira Righi da Silva, M.; Beatriz Alves Rolim, S.
Mapping heavy mineral deposits on the coast of the state of Rio Grande do Sul (Brazil) using orbital and proximal remote sensing.
PLoS ONE 2024, 19, e0309043. [CrossRef] [PubMed]
53. Rejith, R.G.; Sundararajan, M.; Gnanappazham, L.; Seenipandi, K.; Ramaswamy, S. GIS-based machine learning algorithms for
mapping beach placer deposits in the southwest coast of India using Landsat-8 OLI images. J. Appl. Remote Sens. 2021, 16, 012011.
[CrossRef]
54. Rejith, R.G.; Sundararajan, M.; Gnanappazham, L.; Loveson, V.J. Satellite-based spectral mapping (ASTER and landsat data) of
mineralogical signatures of beach sediments: A precursor insight. Geocarto Int. 2022, 37, 2580–2603. [CrossRef]
55. Azzalini, A.; Cardoso-Fernandes, J.; Carvalho, M.; Williams, V.; Lima, A.; Teodoro, A.C. Multitemporal LiDAR-based ter-
rain anomaly detection of Karstic environments in the Asturian Central Massif (Cantabrian Mountains, Northwest Spain).
In Proceedings of the Remote Sensing, Edinburgh, UK, 13 November 2024; SPIE: San Francisco, CA, USA; Volume 13197.
[CrossRef]
56. Araújo, B.; Cardoso-Fernandes, J.; Azzalini, A.; Carvalho, M.; Lima, A.; González, F.; Lobato, A.; Ng-Cutipa, W.; Teodoro,
A.C. Automatic lineament extraction, using Sentinel-1 data, in the Ria de Vigo, Spain. In Proceedings of the Remote Sensing,
Edinburgh, UK, 13 November 2024; SPIE: San Francisco, CA, USA; Volume 13197. [CrossRef]
57. Carvalho, M.; Azzalini, A.; Cardoso-Fernandes, J.; Santos, P.; Lima, A.; Teodoro, A.C. Multi-Sensor Approach for Cobalt
Exploration in Asturias (Spain) Using Machine Learning Algorithms. In Proceedings of the IGARSS 2024—2024 IEEE International
Geoscience and Remote Sensing Symposium, Athens, Greece, 7–12 July 2024; pp. 2122–2126. [CrossRef]
58. Vahtmäe, E.; Kutser, T.; Paavel, B. Performance and Applicability of Water Column Correction Models in Optically Complex
Coastal Waters. Remote Sens. 2020, 12, 1861. [CrossRef]
59. Mederos-Barrera, A.; Marcello, J.; Eugenio, F.; Hernández, E. Seagrass mapping using high resolution multispectral satellite
imagery: A comparison of water column correction models. Int. J. Appl. Earth Obs. Geoinf. 2022, 113, 102990. [CrossRef]
60. Beckert, J.; Vandeginste, V.; McKean, T.J.; Alroichdi, A.; John, C.M. Ground-based hyperspectral imaging as a tool to identify
different carbonate phases in natural cliffs. Int. J. Remote Sens. 2018, 39, 4088–4114. [CrossRef]
61. Rajendran, S.; Nasir, S.; El-Ghali, M.A.K.; Alzebdah, K.; Salim Al-Rajhi, A.; Al-Battashi, M. Spectral Signature Characterization
and Remote Mapping of Oman Exotic Limestones for Industrial Rock Resource Assessment. Geosciences 2018, 8, 145. [CrossRef]
62. Falabella, F.; Pepe, A.; Oštir, K.; Calò, F. Use of Interferometric Synthetic Aperture Radar Techniques as a remote tool for Mineral
Extraction Sites Monitoring. In Proceedings of the EGU General Assembly 2024, Vienna, Austria, 14–19 April 2024. Copernicus
Meetings. [CrossRef]
63. Farahnakian, F.; Luodes, N.; Karlsson, T. Machine Learning Algorithms for Acid Mine Drainage Mapping Using Sentinel-2 and
Worldview-3. Remote Sens. 2024, 16, 4680. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.