Microbial life inhabiting subseafloor sediments plays an important role in Earth’s carbon cycle. However, the impact of geodynamic processes on the distributions and carbon-cycling activities of subseafloor life remains poorly constrained. We explore a submarine mud volcano of the Nankai accretionary complex by drilling down to 200 m below the summit. Stable isotopic compositions of water and carbon compounds, including clumped methane isotopologues, suggest that ~90% of methane is microbially produced at 16° to 30°C and 300 to 900 m below seafloor, corresponding to the basin bottom, where fluids in the accretionary prism are supplied via megasplay faults. Radiotracer experiments showed that relatively small microbial populations in deep mud volcano sediments (10² to 10³ cells cm⁻³) include highly active hydrogenotrophic methanogens and acetogens. Our findings indicate that subduction-associated fluid migration has stimulated microbial activity in the mud reservoir and that mud volcanoes may contribute more substantially to the methane budget than previously estimated.

The optimization of blasting operations greatly benefits from the prediction of rock fragmentation. The main factors that affect fragmentation are rock mass characteristics, blast geometry, and explosive properties. This paper is a step towards the implementation of machine learning and deep learning algorithms for predicting the extent of fragmentation (in percentage) in open pit mining. While various parameters can affect rock fragmentation, this study considers ten among them (i.e., spacing, drill hole diameter, burden, average bench height, powder factor, number of holes, charge per delay, uniaxial compressive strength, specific drilling, and stemming) to train and test the models. However, due to a weak correlation with rock fragmentation, drill diameter, average bench height, compressive strength, stemming, and charge per delay are eliminated to reduce model complexity. A total of 219 data sets having five input features including the number of holes, spacing, burden, specific drilling, and powder factor are used to develop the models. Machine learning models (random forest regression, support vector regression, and XG boost), as well as a deep learning model (neural network regression), are applied to develop a practical way that can optimize the prediction of fragmentation. This study employs performance measures such as R-squared, RMSE, MSE, MAPE, and MAE. The optimization of the model revealed promising results, indicating that the architecture 5-64-32-16-1 exhibits strong performance. Specifically, the model achieved mean squared error (MSE) values of 41.32 and 28.59 on the training and test datasets, respectively. The R² value for both training and test is 0.83. RFR is also performing well compared to SVR and XG boost with MSE values of 12.37 and 9.89 on training and testing data, respectively.in both sets, the R² value is 94%. Based on permutation importance and shapely plot values, it is observed that the powder factor has the highest impact, while the burden has the lowest impact on fragmentation.

Carbonatites undergo various magmatic-hydrothermal processes during their evolution that are important for the enrichment of rare earth elements (REE). This geochemical, petrographic, and multi-isotope study on the Kangankunde carbonatite, the largest light REE resource in the Chilwa Alkaline Province in Malawi, clarifies the critical stages of REE mineralization in this deposit. The δ⁵⁶Fe values of most of the carbonatite lies within the magmatic field despite variations in the proportions of monazite, ankerite, and ferroan dolomite. Exsolution of a hydrothermal fluid from the carbonatite melts is evident based on the higher δ⁵⁶Fe of the fenites, as well as the textural and compositional zoning in monazite. Field and petrographic observations, combined with geochemical data (REE patterns, and Fe, C, and O isotopes), suggest that the key stage of REE mineralization in the Kangankunde carbonatite was the late magmatic stage with an influence of carbothermal fluids i.e. magmatic–hydrothermal stage, when large (~200 µm), well-developed monazite crystals grew. The C and O isotope compositions of the carbonatite suggest a post-magmatic alteration by hydrothermal fluids, probably after the main REE mineralization stage, as the alteration occurs throughout the carbonatite but particularly in the dark carbonatites.

Heavy minerals, such as Rutile, Ilmenite and Zircon, and other essential trace elements are important in modern technology development. The integration of hyperspectral imaging and artificial intelligence presents a promising approach for the accurate identification of heavy minerals, especially Rare Earth Element (REE)–bearing phases such as Monazite. This study evaluates three AI classifiers, Support Vector Machine (SVM), Neural Networks (NNs) and Convolutional Neural Networks (CNNs), for their performance in classifying ten different minerals distributed across six grain size groups ranging from 125 μm to over 300 μm. The analysis focuses on how grain size affects spectral reflectance and classification accuracy. Among the tested models, SVM consistently outperformed NN and CNN, achieving the highest precision, recall and spectral similarity, particularly within the 150–300 μm grain size range. CNN showed the lowest performance and frequently misclassified spectrally similar minerals, such as Zircon and Rutile, likely due to its 1D architecture and limited spatial representation. Monazite, notable for its strong Nd³⁺ absorption features, was accurately identified across applicable grain sizes, highlighting its reliability for REE detection. Spectral Angle Mapper (SAM) analysis confirmed that SVM and NN maintained better spectral similarity than CNN. In general, the results highlight the significant influence of grain size, spectral similarity and dataset size on classification accuracy and the overall effectiveness of AI models in hyperspectral mineral analysis.

Mining methods selection (MMS) is one of the most critical and complex decision-making tasks in mine planning. The selection of underground mining methods is considered to be the most problematic due to the complexity associated with the orebody geometry, geology, and geotechnical properties. This study integrated artificial intelligence and machine learning in the MMS process by introducing the recommendation systems (RS) approach in MMS through the nonnegative matrix factorization (NMF) algorithm. As such, the weighted nonnegative matrix factorization (WNMF) algorithm is applied to build a model for underground MMS. The study's input dataset is based on thirty mining projects' historical data. In the experiments, we evaluate the capability of the WNMF to predict underground mining methods using five input variables: ore strength, host-rock strength, orebody thickness, shape, and dip. The results show that the WNMF model achieved an average prediction accuracy of 67.5%, considered reasonable and realistic. Further findings reveal that the WNMF model is sensitive to the imbalanced class dataset used in the experiments, thus, suggesting the need to improve the dataset's quality. These results reveal the model's effectiveness in predicting underground mining methods; therefore, with continuous improvement, the WNMF model can be effectively applied in underground MMS.