The Perseverance rover landed in Jezero crater on Mars, which once contained a lake of liquid water. We report the rock properties encountered by Perseverance during a 10-kilometer traverse extending over 400 meters in elevation, from beneath Jezero's western sedimentary fan to the upper crater rim. These rocks consist of coarse-grained olivine, magnesium and iron carbonates, silica, and phyllosilicates, including some of the oldest materials exposed within Jezero. We infer that these rocks formed by olivine accumulation in an igneous system of layered intrusions, followed by exposure to water and carbon dioxide, which caused extensive carbonation of the silicate minerals. Aqueous alteration was more pronounced at lower elevations. Higher-elevation exposures on the crater rim appear similar to olivine-rich rocks distributed over the wider Nili Fossae region.

Multiple orbital studies have highlighted the importance of Fe/Mg phyllosilicate minerals on Mars, especially in Noachian terrains (e.g. Poulet et al., 2006; Mangold et al., 2007;), even showing that they are the dominant hydrous mineral family on Mars (Carter et al., 2013). Although widespread in exposures of the oldest terrains on Mars, it is hard to constrain either their composition or formation process based on orbital data only (Ehlmann et al., 2011; Carter et al., 2015). The best matches for the orbital spectra were proposed to be smectite (nontronite, saponite) and/or vermiculite (Carter et al., 2013). In the Nili Fossae region specifically, the Noachian basement was shown to be bear widespread signatures of Fe/Mg smectites (Goudge et al., 2015).After exploring diverse geological units inside Jezero Crater (Nili Fossae, Mars), and going over the rim of the crater, the Perseverance rover has reached a unit informally called Krokodillen, at the base of the outer part of the rim. It is thought to be part of the Noachian crust that was locally uplifted by the emplacement of Jezero Crater (Sun & Stack et al., 2020). Dark looking from orbit, it is surrounded on the North, West and South by ridges and an exposure of the regional olivine-rich unit, understood to be younger.We will present the data acquired on rocks of the Krokodillen area with the SuperCam instrument (ref Maurice et al., SSR 2021; Wiens et al., SSR 2021). Overall structureless, the rocks of Krokodillen are generally fine grained, with locally some millimetric granules. The chemical composition characterized with LIBS shows a relatively homogeneous composition intermediate between the average basaltic crust of Mars and orthopyroxene. This is likely an average, close to the bulk composition, due to the mixing of multiple fine-grain mineral phases within the footprint of the LIBS analysis. Visible and near infrared (VISIR) reflectance spectroscopy data show strong and ubiquitous signatures of Fe-Mg phyllosilicates, closely matching those observed from orbit in the broader Nili Fossae region.We propose that the rocks of Krokodillen are representative of the Noachian clay-bearing rocks characterized from orbit, specifically the (ridged) Altered Basement mapped by Goudge et al., (2015) in the Jezero watershed. In that case, the in situ measurements from the Mars 2020 mission provide the first in situ constraints on the composition, aqueous alteration and emplacement mechanism of these rocks.

We implement a local gradient-based scheme to fit Laser-Induced Break-down Spectroscopy (LIBS) spectra to the emission of a uniform, isothermal,stationary plasma in local thermal equilibrium. The gradients of the lossfunction are obtained by automatic differentiation. We demonstrate the self-consistency and robustness of our fitting scheme through systematic analysisof synthetic spectra and find that, for a wide range of plasma parameters,there only exists one minimum of the loss function, enabling reliable recoveryof the true parameters.

Since the Mars Science Laboratory landed in 2012, the ChemCam instrument aboard the rover has collected in-situ laser-induced breakdown spectroscopy (LIBS) data and context images along more than 35 km of the Gale Crater traverse, providing valuable observations including diagenetic features such as light-toned veins. These veins are of particular scientific interest because they are interpreted as indicators of past fluid circulation on Mars and provide insights into the evolution of habitability on Mars. Their identification, however, currently relies on manual visual inspection of Remote Micro Imager (RMI) images, a process that is time-consuming and sensitive to differences in human interpretation. To address this issue, in this paper we introduce a novel pixel-level labeled, multimodal dataset of ChemCam observations specifically tailored for vein detection, along with customized U-Net models to integrate both textural (RMI) and chemical (LIBS) modalities. To further ensure trustworthy scientific use, we incorporate the Learn-Then-Test (LTT) framework to provide statistical control of the false discovery rate without requiring model retraining. The experimental results demonstrate that the proposed customized U-Net models trained on the developed dataset, combined with risk-controlled prediction, increases the efficiency of pixel-level vein identification through automation and produces statistically reliable predictions for multimodal ChemCam data.

The analysis of the mineralogical composition of rocks within the Jezero crater, Mars, enables the reconstruction of the aqueous activity history of the site, formed during the planet's early epochs. Numerous secondary minerals resulting from aqueous alteration, as well as some primary minerals, are observable through near-infrared reflectance spectroscopy, as performed by the IRS/SuperCam instrument onboard the Perseverance rover. The characterization of these minerals, their distribution within geological units, as well as the composition of their assemblages, is crucial for deciphering the chemical, climatic, and geological history of the Jezero crater. In order to systematically study the growing number of IR spectra collected by the instrument, we have developed an automated method for extracting the characteristics of absorption bands, based on their modeling using a combination of Gaussian curves (MultiGM). The application of this method to the entire IRS dataset reveals a significant diversity of minerals distributed throughout the rover traverse, with a near- systematic presence of phyllosilicates in bedrocks, as well as increasingly frequent occurrences of carbonates on the delta, particularly within the Margin unit. These carbonates exhibit variable 2.5 mu m band positions attributed to different Fe and Mg compositions. The delta front presents the sole unit generally enriched in Fe/Mg sulfates (Yori Pass/Hogwallow Flats) accompanied by occurrences of bassanite (hydrated CaSO4), whereas the presence of sulfates in other delta units and the crater floor is more sporadic and corresponds to fracture/vesicle fills. Finally, almost all the rocks analyzed with IRS/SuperCam are notably dominated by phyllosilicates signatures (Fe/Mg-smectites, serpentine), excepted for a few outcrops caping the Delta. This phyllosilicate presence witnesses the extended surficial (smectites) and hydrothermal (serpentine) aqueous alteration of Jezero's rocks, either in situ or carried from the watershed.

The martian soil is of particular interest as it can help us understand the different processes that have occurred on Mars by studying the chemistry and mineralogy of its constituents as a function of grain size. The finegrained martian soil is thought to be homogeneous across the planet and thus to represent a global component. In this study we report on the soil targets analysed by the SuperCam instrument aboard the Perseverance rover, which is currently exploring Jezero crater. A total of 343 targets were analysed. Their grain size distribution confirms the sparsity of 250-900 Acirc;mu m particles in the martian soil, although both smaller and larger grains are present. We found that the local components, due to erosion of the local bedrock, are present not only in the very coarse grains or larger gravels of the soil, but also in the very fine ones (<250 Acirc;mu m). We detected some very coarse grains enriched in olivine, pyroxene and carbonate in both the crater floor and the delta front locations, whereas phyllosilicate-rich grains have been encountered only in the delta front. We have compared the Jezero fine-grained soil targets with those of Gale crater using ChemCam data. We found that those at Jezero show no evidence of Mg sulfates, in contrast to the observation at Gale. In addition, the fine-grained soil at Jezero is more hydrated than that at Gale, probably due to its higher specific surface area.

The spectroscopic technique of laser-induced breakdown spectroscopy (LIBS) is a powerful method to perform rapid chemical analysis of geologic samples with short measurement times and no need for sample preparation. After the ChemCam instrument aboard NASA's MSL rover proved its suitability for space missions that explore planetary surfaces in 2012, the interest in LIBS instruments as payloads has grown and several subsequent missions have successfully used this technique since. The characteristics of a LIBS plasma depend on experimental and environmental parameters as well as on sample properties, including atmospheric conditions, laser irradiance and sample lithology. Consequently, LIBS instruments need to be designed and optimized specifically for each use case to maximize their science output. To aid in the development of new LIBS instruments for space exploration, we investigate the influence of atmospheric conditions, laser irradiance and sample lithology on the lifetime, size and emission of laser-induced plasmas. In our measurements, we use a plasma imaging setup with high temporal resolution of down to 2 ns to investigate the evolution of the plasma from its ignition to its decay. We present a comparable data set recorded at terrestrial, Martian and airless atmospheric conditions, covering irradiances between 0.79 GW/mm2 and 1.43 GW/mm2 and samples with diverse properties, namely basalt and soapstone, as well as the lunar regolith simulants LHS-1 and LMS-1. Our measurements show the strong influence of atmospheric conditions on the plasma size and emission, while the lithologies and laser irradiances covered in this work playa minor role. This shows that instruments designed to work at certain atmospheric conditions can be used fora range of laser parameters and sample properties. Furthermore, we demonstrate that the decay of the plasma emission and the expansion of the plasma plume parallel to the sample surface can be described well by a power law and a drag model, respectively.

The Mars 2020 Perseverance rover introduced Raman spectroscopy to in situ planetary exploration for the first time when it landed in Jezero crater on Mars in February 2021. The SuperCam instrument onboard Perseverance is a multi-analytical tool capable of acquiring time-resolved Raman data from Martian targets at standoff distances of a few meters. This is a particularly challenging task due to the operational constraints, the harsh conditions on the Martian surface, and especially the very fine-grained nature of the Martian soil. To address these challenges, the SuperCam Raman team has invested significant effort into optimizing both the acquisition and post-processing of Raman data collected on Mars, as detailed in this work. Additionally, this paper reviews and discusses the detections made by SuperCam Raman during the first 1,000 sols (almost 3 Earth years) of the Mars 2020 mission. During this period, SuperCam Raman data provided key insights into the mineralogy of Jezero throughout the Crater, Delta, and Margin Campaigns. Key detections include olivine, carbonates, perchlorates, and sulfates (such as anhydrite), identified in both abraded patches and natural surfaces. The high specificity of Raman spectroscopy enables the unequivocal identification of these minerals, allowing for rapid and direct interpretation of Jezero's mineralogy, especially when combined with other techniques from SuperCam or others on the rover. Furthermore, this paper compiles the spectra acquired from the SuperCam Calibration Target samples on Mars, including studies on the degradation of the Ertalyte (PET), an organic polymer sample and analyses of diamond, apatite, and other reference materials.

On Earth, silica-rich phases from opal to quartz are important indicators and tracers of geological processes. Hydrated silica, such as opal, is a particularly good matrix for the preservation of molecular and macroscopic biosignatures. Cherts, a type of silica-dominated rocks, provide a unique archive of ancient terrestrial life while quartz is the emblematic mineral of the Earth's continental crust. On Mars, hydrated silica has been detected in several locations based on remote sensing and rover-based studies. In the present article we report on the detection of cobbles made of hydrated silica (opal or chalcedony), as well as well-crystallized quartz. These detections were made with the SuperCam instrument onboard Perseverance (Mars 2020 mission), using a combination of LIBS, infrared and Raman spectroscopy. Quartz-dominated stones are detected unambiguously for the first time on the Martian surface, and based on grain size and crystallinity are proposed to be of hydrothermal origin. Although these rocks were all found as float, we propose that these detections are part of a common hydrothermal system, and represent different depths / temperatures of precipitation. This attests that hydrothermal processes were active in and around Jezero crater, possibly triggered by the Jezero crater-forming impact. These silica-rich rocks, in particular opaline silica, are very promising targets for sampling and return to Earth given their high biosignature preservation potential.

The Perseverance rover is exploring Jezero crater on Mars, one of its goals being to collect samples to be returned to Earth to search for organic remains of ancient Martian life. However, the organic content of these rocks has likely suffered from the radiation environment on the surface of Mars to an extent yet to be quantified. For the first time, a 1000 sols long ageing experiment was conducted at the surface of Mars, i.e. under actual Martian conditions, relying on the 100 % organic Ertalyte target carried by Perseverance. White at landing, the Ertalyte target has turned brown with time, while its Raman signal changed, with a modification of the background (its maximum has shifted from 1500 to 2000 cm-1) and a reduction of the contribution of the Raman signal of Ertalyte (by a factor of 5 over the first 500 sols). Given the intrinsic resistance of the Ertalyte to UV exposure, which is not anticipated for most Martian organic materials, these results suggest that exposure at the surface of Mars will make the detection of Martian organic molecules challenging.

Meteorite paleomagnetic studies indicate planetesimal generated magnetic fields, but spacecraft magnetic measurements have yet to identify asteroidal natural remanent magnetization (NRM). This apparent discrepancy is of particular interest in the context of the NASA Psyche mission, which will search for evidence of past magnetic activity of the metal-rich asteroid (16) Psyche. Here, we aim to test whether the NRM of meteorites inevitably drops below detectable values as specimen size increases, which could explain why asteroidal NRMs could never be detected. We focus on iron meteorites as possible analogs to (16) Psyche's constituent material. To do so, we measure the remanent magnetic field and estimate the NRM of samples of four iron meteorites with volumes between mm3 and m3. We find that their estimated NRMs decrease with increasing sample size but appear to plateau. These data are compatible with the idea that the bulk NRM of increasingly large objects becomes dominated by the fraction of this NRM produced by assemblages of magnetic minerals sharing a common magnetization direction. Moreover, all m3-sized meteorites carry NRMs that are two orders of magnitude above the detectability limit of the Psyche Magnetometer, three of which are possibly pre-terrestrial. These data, acquired on some of the largest masses of iron meteorites available on Earth, support the range of plausible NRM values for km-size regions of (16) Psyche, used to establish the spacecraft Magnetometer's performance requirements. Nevertheless, large-scale events such as brecciation of the asteroid following magnetization acquisition could always lower the asteroid's NRM below the detectability limit.

Introduction: When the MSL rover Curiosity landed on Mars in 2012, it successfully demonstrated the use of Laser-Induced Breakdown Spectroscopy (LIBS) for in-situ geochemical investigations of planetary bodies for the first time [1]. Since then, the importance of LIBS for Solar System exploration has only grown, with several subsequent missions employing LIBS systems as part of their sensor suite [2, 3]. LIBS measurements are performed by focusing a pulsed laser beam on the surface of a sample to generate a local plasma. The emission of this plasma is then analyzed to investigate the sample composition. Since the lifetime, size and emission of the laser-induced plasma depend strongly on experimental parameters [4, 5], LIBS instruments for in-situ analysis of planetary bodies have to be designed for specific environments to generate high-quality data. Important parameters include the ambient atmosphere, the laser irradiance and the sample lithologyHere, we present measurements of the plasma emission at different atmospheric conditions and laser irradiances for four samples with different lithologies to aid in the development of new LIBS instruments for in-situ geochemical research of planetary bodies.Experimental Setup: To investigate the lifetime, size and emission of laser-induced plasmas, we use a plasma imaging system that employs a gated ICCD sensor to achieve a temporal resolution of down to 2 ns. The ICCD is most sensitive between about 300 nm and 900 nm. The plasma is generated by a 1064 nm Nd:YAG laser with a pulse duration of 8.1 ns and a spot diameter of about 40 µm.Figure 1: Overview of the experimental setup.Different atmospheric conditions can be simulated using a vacuum chamber in which the plasma is ignited, see Fig. 1. In this study, we focus on atmospheric conditions on Earth, Mars and airless planetary bodies such as the Moon.Methodology: For all investigated samples, measurements were performed at terrestrial, Martian and airless conditions. The laser pulse energy was varied between 11.86 mJ, 8.75 mJ and 6.56 mJ, which corresponds to average irradiances of about 1300 MW/mm², 990 MW/mm² and 750 MW/mm². Four samples with different lithologies were investigated: LMS1 (pressed pellet), LHS1 (pressed pellet), soapstone (talc) and basalt (cut rock). For this abstract, we limit the discussion to data from the basalt sample.Each plasma image was recorded from a single laser shot with ICCD gate times between 2 ns and 200 ns. All images were pre-processed by subtracting a dark image, applying a uniform filter to reduce the noise level and removing residual constant offsets. The total plasma emission was calculated as the sum over each image. To account for the different ICCD gate times, the total emission of each image was normalized with the gate time. The lifetime of the plasma was computed as the time that covers 90 % of the total detected plasma emission. The full width at half maximum sizes of the plasmas were calculated as the largest horizontal distance between points that show half of the maximum plasma emission of the image.Results: Fig. 2 shows the expansion of the plasma plume from the basalt sample at the three investigated atmospheric conditions for a laser energy of 8.75 mJ. All plasma images are normalized to their respective maximum. At 100 ns after plasma ignition, two plasma plumes are visible for the measurement performed at terrestrial conditions. This secondary breakdown is likely initiated on residual ejecta from the ablation process.Figure 2:Time series of plasma images at the three investigated atmospheric conditions. Each plasma image is normalized to its respective maximum signal. Note the different time scales.Overall, a change in the extent of the plasma and its dynamics is visible between different atmospheric conditions. The plasma’s lifetime is longest at terrestrial atmospheric conditions and shortest at airless conditions. At terrestrial conditions, the plasma is most confined. It is larger at Martian atmospheric conditions and at airless conditions, the plasma extends freely into the vacuum, which leads to a small and bright plasma core close to the sample surface.Figure 3: Comparison of the development of the total emitted plasma radiation, the plasma’s horizontal and the plasma lifetime.. Note the different scales of the y-axes.In Fig. 3, the total emission of the plasma is compared for the three atmospheric conditions and laser energies. Fainter colors represent lower laser energy. The data is normalized to the strongest emission at terrestrial conditions. The vertical lines represent the computed plasma lifetime. On the secondary y-axis, the is plotted. At terrestrial atmospheric conditions, the plasma is brightest at 10 ns after plasma ignition. The plasma reaches its maximum horizontal extend of about 1.3 mm around 3 µs after ignition. 90 % of the emission are contained within the first 2.5 µs. For Martian conditions, the peak emission occurs after about 20 ns and is about 20 % of the peak emission at terrestrial conditions. The plasma lifetime is reduced to about 1.5 µs and its maximum extend of about 3 mm is reached around 500 ns after ignition. Finally, for airless conditions the peak plasma emission is reduced to about 7% of the peak emission at terrestrial conditions with a lifetime of about 80 ns. The plasma reaches its maximum extent of about 2 mm after around 40 ns.Conclusion: The presented data show the different behaviors of plasmas ignited at different atmospheric conditions. The maximum extent varies between about 1.3 mm for terrestrial conditions, 3 mm for Martian conditions and 2 mm for airless conditions. The plasma is brightest at Terrestrial atmospheric conditions. At Martian and airless conditions, the plasma’s peak emission is about 20% and 7% of that at terrestrial conditions, respectively. Part of this work was first presented at the 55th LPSC conference [6].References: [1] Maurice, S. et al. (2016) J. Anal. At. Spectrom., 31(4), 863-889. [2] Maurice, S. et al. (2021) Space Sci. Rev., 217, 1-108. [3] Wan, X. (2021) At. Spectrosc., 42(6), 294-298. [4] Singh, J. P. and Thakur, S. N. (Eds.). (2020). Laser-induced breakdown spectroscopy. Elsevier. [5] Lasue, J. et al. (2012) J. Geophys. Res. Planets, 117, (E1). [6] Seel, F. et al. (2024) 55th LPSC, 1418.

Since 2012, NASA’s Curiosity rover has been looking for evidence of previous habitability on Mars [1,2]. For this purpose, the rover is equipped with a variety of scientific instruments. One of them is ChemCam (Chemistry and Camera), which consists of two parts: LIBS (Laser Induced Breakdown Spectrometer) and RMI (Remote Micro Imager). LIBS provides information on the elemental composition of targets and RMI takes context images [3]. More than 4000 targets have been measured by ChemCam since it landed [4]. One of the common questions asked when analyzing ChemCam targets is when similar targets were observed. The size of the data makes manual labeling tedious, and automating the process would reduce the human workload. In this work, we will explore machine learning methods to improve the classification process of RMI images in terms of rock texture. In previous works, we used unsupervised classification, k-means clustering, to derive potential labels for the images [5]. We came down to nine classes: smooth, low nodular, high nodular, fractured, veins, layered, pebbles, soil, and drill shown in Fig 1. We labeled 100 images per class and used transfer learning, an already pretrained model VGG16, to make the size of the training set sufficient for convolutional neural networks [6]. The experiments showed us that more than one label applies to most ChemCam targets. This led us to multilabel classification. Although we added labels to the targets and fine-tuned models, we were not able to reach more than 80% accuracy. The model kept confusing labels, and one approach to investigate potential reasons is to employ Explainable Artificial Intelligence (XAI) methods to understand what was learned by the model. There are various methods to visualize learned patterns of each layer in deep neural networks. In order to address the so-called “black box”, we employed Guided Backpropagation [6]. The technique is a combination of backpropagation and the deconvolutional network. By setting negative gradients to zero it highlights the most activated pixels by each layer. An example of using guided backpropagation on the 10th layer of VGG16 is illustrated in Fig 2. This method allowed us to understand which textural features were learned by the model and which needed more refining.Additionally, XAI methods are able to estimate the importance of features in the training set. Some characteristics of the image may not influence the decision-making process of the model and can be a source of extra information making the model complex and heavy. Considering the limited capacity of resources of onboard computing in in-situ missions, making models lighter is one of the priorities. Shapley values from game theory evaluates the contribution of each feature in the model that can be translated into feature importance [7]. In this work, we evaluate the feature importance of our dataset in terms of model accuracy and remove unnecessary weight from the model.Overall, we explore multilabel classification of ChemCam targets based on their textures captured by RMIs . The automatization of labeling allows efficient interpretation of rocks and identification of regions with similar targets. We apply XAI methods such as guided backpropagation to improve the accuracy of the classification by visualizing the learned patterns. Additionally, by estimating feature importance using Shapley values, we make the model lighter for potential in-situ operation. Figure 1 Representative images of the clusters and corresponding labels.Figure 2 On the left ChemCam target "Rooibank" sol 1266, displaying veins. On the right Guided backpropagation is applied to the target, highlighting veins learned by the 10th layer of VGG16.Refs:[1] J. P. Grotzinger, J. Crisp, A. R. Vasavada, and R. P Anderson. Mars Science Laboratory Mission and Science Investigation. Space Sci Rev, 2012.[2] A. R. Vasavada. Mission Overview and Scientific Contributions from the Mars Science Laboratory Curiosity Rover After Eight Years of Surface Operations. Space Sci Rev, 2022.[3] R. C. Wiens, S. Maurice, B. Barraclough, M. Saccoccio, and W. C. Barkley. The ChemCam Instrument Suite on the Mars Science Laboratory (MSL) Rover: Body Unit and Combined System Tests. Space Sci Rev, 2012[4] O. Gasnault et al. Exploring the sulfate-bearing unit: Recent ChemCam results at Gale crater, Mars. EPSC, 2024.[5] A. Lomashvili et al. Rock classification via transfer learning in the scope of ChemCam RMI image data. LPSC, 2023.[6] J. T. Springenberg, A. Dosovitskiy, T. Brox, and M. Riedmiller. Striving for Simplicity: The All Convolutional Net. ICLR 2015.[7] S. Lundberg and S. I. Lee. A Unified Approach to Interpreting Model Predictions. NIPS 2017.

Laser-induced breakdown spectroscopy (LIBS) and Raman spectroscopy are still rather new techniques for in-situ exploration of extraterrestrial planetary surfaces but have shown their suitability and great potential in several successful robotic missions already. Next to serving primary scientific applications, both methods can also be used in the context of in-situ resource utilization (ISRU) such as scouting for wanted substances and the surveillance of extraction processes. Here, we present two laboratory studies conducted in the context of ISRU with a focus on the chain from prospecting to extracting oxygen from lunar regolith. For LIBS, with optimized data processing and combined with state-of-the-art multivariate data analysis approaches, we show the potential of the technique for identifying samples with increased ilmenite content and for elemental quantification. The measurements were done using lunar regolith simulant and low pressures simulating vacuum on atmosphereless bodies such as the Moon. With Raman spectroscopy, we analyzed lunar regolith simulant samples that underwent electrochemical alteration for oxygen extraction and production of metal alloys demonstrating the potential of Raman spectroscopy for ISRU process monitoring. We also discuss the results in a broader context, evaluating the potential of both methods for other aspects of ISRU support.

The ChemCam instrument on the Curiosity rover performs geochemical analyses of rocks on Mars using Laser-Induced Breakdown Spectroscopy (LIBS). The shockwaves generated during the LIBS measurements sometimes shift dust from the surface of the target. The study of the Martian dust phenomena in the scope of the ChemCam instrument has the potential to provide insight into the planet's geology and aid calibration methods for data processing. In this study, we develop a pipeline, named Dust Displacement Detection (DDD), for automatic detection of dust displacement on LIBS targets based on the image dataset acquired by ChemCam. To this end, we introduce a data preprocessing methodology and test two-stage models with a pretrained model in the first stage for feature extraction and a Random Forest classifier or a Support Vector Machine as a binary classifier in the second stage. The best performing model was found to consist of the first 10 layers of VGG16 and a Random Forest classifier, achieving 92% accuracy. Additionally, we use Explainable AI (XAI) methods such as Shapley values and guided backpropagation for model optimization. The experiments show potential for model optimization, and the application examples presented encourage discussion of machine learning in the field of Martian dust research.