Next Article in Journal
Polarization Scattering Regions: A Useful Tool for Polarization Characteristic Description
Previous Article in Journal
Unmanned Aerial Vehicles for Real-Time Vegetation Monitoring in Antarctica: A Review
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Remote Sensing
Volume 17
Issue 2
10.3390/rs17020305
remotesensing-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Comparative Performance of Sentinel-2 and Landsat-9 Data for Raw Materials’ Exploration Onshore and in Coastal Areas

by
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
2
Marine Geology Resources and Extreme Environments, Geological Survey of Spain (IGME-CSIC), C/Ríos Rosas 23, 28003 Madrid, Spain
*
Author to whom correspondence should be addressed.
Remote Sens. 2025, 17(2), 305; https://doi.org/10.3390/rs17020305
Submission received: 2 December 2024 / Revised: 8 January 2025 / Accepted: 13 January 2025 / Published: 16 January 2025
(This article belongs to the Section Remote Sensing in Geology, Geomorphology and Hydrology)
Browse Figures
Versions Notes

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.

Graphical Abstract

1. Introduction

Intending to address the European Union’s (EU) dependency on imported Critical Raw Materials (CRM), the Critical Raw Materials Act was proposed in March of 2023, responding to the “Green Industrial Plan”. The act aims to support the green transition, anticipating and mitigating supply risks, and reducing the underutilization of CRM deposits located in European soil [1]. This act emphasizes the strategic role of advanced remote sensing technologies in identifying and prioritizing CRM deposits within European territories. By promoting investments in Earth observation tools, we can achieve a more accurate and sustainable assessment of mineral deposits.
CRMs are used for a number of application areas, including solar panels, wind turbines, and electric vehicle batteries and motors, which are critical to replace fossil-based energy and fuels in support of the EU’s green transition and sustainability goals [2,3]. CRMs are also used to store energy in batteries and infrastructure for cleaner technologies [2]. Metals 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 Observation (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 [8,9,10,11,12]. 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 [17,18,19]. The two satellites 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 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 [11,22,23,24]. 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 [26,27,28,29]. In Ria de Vigo, the waves mainly influence the selection of sediments [26,28,29,30], 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 [26,31,32,33].
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 stacking of Namurian limestones (Upper Carboniferous) in thrust sheets, creating a rugged 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 landscape [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 tectonic activity [37,39]. Dolomitization is the main alteration process around the fault zones, with three primary hypogenic mineralization stages, followed by significant supergene weathering. The mineralization occurs in veins and lenses along E-W and NE-SW faults, often accompanied by secondary porosity from karstification [40].

3. Materials and Methods

The methodology applied involved selecting suitable images from Senitnel-2 and Landsat-9, selecting bands of interest, and creating a subset for the interest area. The different remote sensing methods, band rationing, PCA, and RGB compositions were used, and a threshold was applied to the band ratio results. Then, the results were critically analyzed and interpreted. Figure 3 explains the steps followed in the study.

3.1. Data Acquisition

The Landsat-9 and Sentinel-2 images utilized in this study were acquired at level 2 of processing, which means the images were already atmospherically corrected and surface reflectance was already calculated, allowing us to use the images directly for analysis. Landsat images are free to download and were obtained from earthexplorer.usgs.gov (accessed on 12 October 2022). The Landsat-9 image for the Aramo area (LC09_L2SP_203030_20231007_20231008_02_T1) was acquired on 1 October 2023 and for the Ria de Vigo area (LC09_L2SP_204031_20230216_20230310_02_T1) in 16 February 2023. 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 the Aramo area was acquired on 28 June 2023 (S2B_MSIL2A_20230628T112119_N0509_R037_T29TQH_20230628T124435) and for Ria de Vigo (S2A_MSIL2A_20230216T113321_N0509_R080_T29TNG_20230216T151057) on 16 February 2023.

3.2. Sentinel-2 and Landsat-9 Bands

Sentinel-2 captures images with a spatial resolution variating from 10 to 60 m, across 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 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.
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, 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].
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 applied 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.
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 than using a larger number of bands. Additionally, we tested PCA combinations shown to effectively map hydrothermal alterations across multiple sensors [42].

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 representing 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
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 (Figure 4a,d and Figure 5a,d for ferric iron, and Figure 4b,e and Figure 5b,e for iron oxides).
The hydrothermal alteration band ratio produced mixed results: it worked well with Sentinel-2 (Figure 4c and Figure 5c) but performed poorly with Landsat-9 (Figure 4f and Figure 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.
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 differentiate features on the beach face effectively.

4.2. Aramo

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 limited results with Landsat-9, detecting only ferric iron (Figure 6a) and ferrous iron (Figure 6b) anomalies. Sentinel-2, however, identified additional features, including iron oxides (Figure 6e) and hydrothermal alterations (Figure 6f). The clay minerals band ratio did not generate meaningful results with either satellite.
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 anomalies 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.
Related to the RGB combinations, Sentinel-2 RGB 2-3-12 effectively highlights areas 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 corresponding Landsat-9 RGB 2-3-7 (Figure 7b) yields a similar result, primarily separating vegetation from exposed rock in purple tones. RGB combinations designed for hydrothermal 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 5-6-7 and 5-7-3) produce similar contrasts between zones (Figure 7d).
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), 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-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 of bands in PC1 (Table 5) can be attributed to the overall scene brightness or albedo [49]. In PC2, the pair of bands displays moderate to strong loadings with opposite signs for both satellites. For Sentinel-2, hydroxyl minerals will be mapped in high pixel values in PC2 due to the positive input of band 11 (reflective band) and the negative input of band 12 (absorption band). In the 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 in PC2. This can be corrected through simple mathematical band operations, specifically by multiplying PC2 of Landsat-9 by -1 and ensuring the target areas correspond to high pixel values [49].
PCA applied to Sentinel-2 bands 11 and 12 (Figure 8b) highlighted similar areas to the ones using the Sabins hydrothermal band ratio (Figure 6f) and ferrous iron (Figure 6e) for the same satellite. The results are also aligned with the one obtained with the RGB combination for hydrothermal alterations (Figure 7c). However, the equivalent combination for Landsat-9 (Figure 8a) does not yield similar results.
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 compositions, 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 significant 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 documented [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 learning 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 supervised machine learning approach aligned with the results obtained with the application 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 machine learning algorithms, as those approaches improved the detection and quantification of mineral placers in recent studies [52,53,54]. 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 [51,55,56,57].
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 beyond the European framework, as they are already broadly applied to diverse geographic contexts [15,22]. The application of these methodologies reduces reliance on invasive exploration 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 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 characteristics 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 hyperspectral 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 download. 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 downloaded from https://browser.dataspace.copernicus.eu/ (accessed on 21 November 2023). The derived/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).

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. [Google Scholar] [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. [Google Scholar] [CrossRef]
  3. Srivastava, N. Strengthening European Energy Security and Resilience through Minerals. Eur. Energy Environ. Law Rev. 2024, 33, 35–45. [Google Scholar] [CrossRef]
  4. Komnitsas, K.; Lazos, I.; Eerola, T. Energy Transition Metals: Future Demand and Low-Carbon Processing Technologies. Mater. Proc. 2023, 15, 56. [Google Scholar] [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. [Google Scholar]
  6. Atlas, A. Mapping Mining to the Sustainable; World Economic Forum: Cologny/Geneva, Switzerland, 2016. [Google Scholar]
  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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [CrossRef]
  21. Sabins, F.F. Remote sensing for mineral exploration. Ore Geol. Rev. 1999, 14, 157–183. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar]
  24. Imbroane, M.A.; Melenti, C.; Gorgan, D. Mineral explorations by Landsat image ratios. In Proceedings of the Ninth International 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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [CrossRef]
  29. Cronan, D.S. Handbook of Marine Mineral Deposits, 1st ed.; CRC Press: New York, NY, USA, 2000. [Google Scholar]
  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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [CrossRef]
  36. Aller, J. La estructura geológica de la Sierra del Aramo (Zona Cantábrica, NO de España). Trab. De Geol. 1983, 19, 3–15. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar]
  41. Hawryło, P.; Wężyk, P. Predicting Growing Stock Volume of Scots Pine Stands Using Sentinel-2 Satellite Imagery and Airborne Image-Derived Point Clouds. Forests 2018, 9, 274. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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).
  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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [CrossRef]
  49. Loughlin, W. Principal component analysis for alteration mapping. Photogramm. Eng. Remote Sens. 1991, 57, 1163–1169. [Google Scholar]
  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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [CrossRef]
  55. Azzalini, A.; Cardoso-Fernandes, J.; Carvalho, M.; Williams, V.; Lima, A.; Teodoro, A.C. Multitemporal LiDAR-based terrain 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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [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. [Google Scholar] [CrossRef]
Remotesensing 17 00305 g001
Figure 1. The Ria de Vigo study area location in northern Spain and details of the beaches analyzed in this study.
Figure 1. The Ria de Vigo study area location in northern Spain and details of the beaches analyzed in this study.
Remotesensing 17 00305 g001
Remotesensing 17 00305 g002
Figure 2. Aramo study area location in Spain, Aramo plateau detail and geological scheme of the Sierra del Aramo. The legend corresponds to 1. Dolomites, limestones, sandstones, shales, marlstones, microconglomerates, and siltstones (Devonian). 2. Quartzite sandstones, marlstones, 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.
Figure 2. Aramo study area location in Spain, Aramo plateau detail and geological scheme of the Sierra del Aramo. The legend corresponds to 1. Dolomites, limestones, sandstones, shales, marlstones, microconglomerates, and siltstones (Devonian). 2. Quartzite sandstones, marlstones, 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.
Remotesensing 17 00305 g002
Remotesensing 17 00305 g003
Figure 3. Flowchart for the steps applied in the methodology.
Figure 3. Flowchart for the steps applied in the methodology.
Remotesensing 17 00305 g003
Remotesensing 17 00305 g004
Figure 4. Band ratios for the Barra Beach. Sentinel-2: (a) ferric iron (purple); (b) iron oxides (pink); (c) hydrothermal alteration a (red). Landsat-9: (d) ferric iron (purple); (e) iron oxides (pink); (f) hydrothermal alteration b (red). Datum WGS 84 UTM zone 29 N, scale 1:15,000. a and b according to Table 2.
Figure 4. Band ratios for the Barra Beach. Sentinel-2: (a) ferric iron (purple); (b) iron oxides (pink); (c) hydrothermal alteration a (red). Landsat-9: (d) ferric iron (purple); (e) iron oxides (pink); (f) hydrothermal alteration b (red). Datum WGS 84 UTM zone 29 N, scale 1:15,000. a and b according to Table 2.
Remotesensing 17 00305 g004
Remotesensing 17 00305 g005
Figure 5. Band ratio results for Sentinel-2 for Vao beach: (a) ferric iron (purple); (b) iron oxides (pink), (c) hydrothermal alteration a (red). Landsat-9 for Vao beach (d) ferric iron (purple); (e) iron oxides (pink), (f) hydrothermal alteration b (red). Datum WGS 84 UTM Zone 29 N, scale 1:7,500. a and b according to Table 2.
Figure 5. Band ratio results for Sentinel-2 for Vao beach: (a) ferric iron (purple); (b) iron oxides (pink), (c) hydrothermal alteration a (red). Landsat-9 for Vao beach (d) ferric iron (purple); (e) iron oxides (pink), (f) hydrothermal alteration b (red). Datum WGS 84 UTM Zone 29 N, scale 1:7,500. a and b according to Table 2.
Remotesensing 17 00305 g005
Remotesensing 17 00305 g006
Figure 6. Landsat-9 band ratios highlighting (a) ferric iron (pink); (b) iron oxides (orange). Sentinel-2 band ratios highlighting (c) ferric iron (pink); (d) iron oxides (orange), (e) ferrous iron (purple) (f) hydrothermal ratio a (dark pink). Datum WGS 84, UTM zone 30 N, scale 1:95,000. a according to Table 2.
Figure 6. Landsat-9 band ratios highlighting (a) ferric iron (pink); (b) iron oxides (orange). Sentinel-2 band ratios highlighting (c) ferric iron (pink); (d) iron oxides (orange), (e) ferrous iron (purple) (f) hydrothermal ratio a (dark pink). Datum WGS 84, UTM zone 30 N, scale 1:95,000. a according to Table 2.
Remotesensing 17 00305 g006
Remotesensing 17 00305 g007aRemotesensing 17 00305 g007b
Figure 7. Results for Aramo using Sentinel-2 RGB combination with bands: 2-3-12 (a), Landsat-9 RGB combination with bands: 2-3-7 (b). Sentinel-2 RGB combination with bands: 8-11-12 (c), Landsat-9 RGB combination with bands: 5-6-7 (d). Sentinel-2 RGB combination with bands: 12/8, 11/4, 4/2 (e), Landsat-9 RGB combination with bands: 7/5, 6/4, 4/2 (f). Datum WGS 84, UTM zone 30 N, scale 1:95,000.
Figure 7. Results for Aramo using Sentinel-2 RGB combination with bands: 2-3-12 (a), Landsat-9 RGB combination with bands: 2-3-7 (b). Sentinel-2 RGB combination with bands: 8-11-12 (c), Landsat-9 RGB combination with bands: 5-6-7 (d). Sentinel-2 RGB combination with bands: 12/8, 11/4, 4/2 (e), Landsat-9 RGB combination with bands: 7/5, 6/4, 4/2 (f). Datum WGS 84, UTM zone 30 N, scale 1:95,000.
Remotesensing 17 00305 g007aRemotesensing 17 00305 g007b
Remotesensing 17 00305 g008
Figure 8. (a) Landsat-9 PCA result with bands 6 and 7, PC2. (b) Sentinel-2 PCA result with bands 11 and 12, PC2. Red pixels indicate hydroxyl mineral targets. Datum WGS 84, UTM zone 30 N, scale 1:95,000.
Figure 8. (a) Landsat-9 PCA result with bands 6 and 7, PC2. (b) Sentinel-2 PCA result with bands 11 and 12, PC2. Red pixels indicate hydroxyl mineral targets. Datum WGS 84, UTM zone 30 N, scale 1:95,000.
Remotesensing 17 00305 g008
Table 1. Sentinel-2 and Landsat 8/9 comparative bands.
Table 1. Sentinel-2 and Landsat 8/9 comparative bands.
Sentinel-2 *Landsat-8/9
Resolution (m) Wavelength Range (nm) Resolution (m) Wavelength 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).
Table 2. Band ratios were selected for the study.
Table 2. Band ratios were selected for the study.
TargetSentinel-2Landsat-9
Ferric Iron [21]B4/B3B4/B3
Ferrous Iron [46](B3 + B12)/(B4 + B8)(B3 + B7)/(B4 + B5)
Hydrothermal alteration a [11]B4/B2 + B6/B11 + B11/12B4/B2 + B6/B5 + B6/B7
Hydrothermal alteration bB4/B3 + B6/B11 + B11/B12B4/B3 + B6/B5 + B6/B7
Iron oxide index [12,16](B4 − B2)/(B4 + B2)(B4 − B2)/(B4 + B2)
Silica index [10]B11/B4B5/B4
Clay/hydroxyl minerals [42,44]B11/B12B6/B7
a Sabin’s hydrothermal band ratio. b alternative hydrothermal band ratio.
Table 3. RGB combinations were selected for the study.
Table 3. RGB combinations were selected for the study.
TargetSentinel-2Landsat-9
Iron/oxides and clay minerals [25]B2, B3, B12B2, B3, B7
Hydrothermal alteration 1 [25]B8, B11, B12B5, B6, B7
Hydrothermal alteration 2 [25]B8, B12, B3B5, B7, B3
Limestone [47]B12/B8, B11/B4, B4/B2B7/B5, B6/B4, B4/B2
Lithological discrimination [48]B12, B3, B5B7, B3, B5
Table 4. Bands used for PCA application and respective study areas.
Table 4. Bands used for PCA application and respective study areas.
Study AreaSentinel-2Landsat-9TargetNumber of PCs
Aramo [25]B11, B12B6, B7Hydroxyl minerals2
Aramo/Vigo [25,42]B2, B4B2, B4Ferric iron2
Vigo [25]B2, B4, B8, B11B2, B4, B5, B6Hydroxyl minerals4
Aramo/Vigo [42]B11, B4B6, B4Gossan2
Aramo/Vigo [42]B11, B8B6, B5Ferric oxides2
Table 5. PCA Eigenvalues for the PCA results are presented in the paper.
Table 5. PCA Eigenvalues for the PCA results are presented in the paper.
Eigenvalues
BandPC1PC2
Landsat-9B60.83718−0.54692
B70.546920.83718
Sentinel-2B110.717580.69647
B120.69647−0.71758
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.

Share and Cite

MDPI and ACS Style

Carvalho, M.; Cardoso-Fernandes, J.; González, F.J.; Teodoro, A.C. Comparative Performance of Sentinel-2 and Landsat-9 Data for Raw Materials’ Exploration Onshore and in Coastal Areas. Remote Sens. 2025, 17, 305. https://doi.org/10.3390/rs17020305

AMA Style

Carvalho M, Cardoso-Fernandes J, González FJ, Teodoro AC. Comparative Performance of Sentinel-2 and Landsat-9 Data for Raw Materials’ Exploration Onshore and in Coastal Areas. Remote Sensing. 2025; 17(2):305. https://doi.org/10.3390/rs17020305

Chicago/Turabian Style

Carvalho, Morgana, Joana Cardoso-Fernandes, Francisco Javier González, and Ana Claudia Teodoro. 2025. "Comparative Performance of Sentinel-2 and Landsat-9 Data for Raw Materials’ Exploration Onshore and in Coastal Areas" Remote Sensing 17, no. 2: 305. https://doi.org/10.3390/rs17020305

APA Style

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), 305. https://doi.org/10.3390/rs17020305

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop