Applied Computing and Geosciences最新文献

{"title":"Simulating tectonic stress fields in mineralization dynamics using a deep neural networks-based surrogate model","authors":"Fan Xiao ,&nbsp;Ao Tang ,&nbsp;Dunhui Xiao ,&nbsp;Shu Jiang ,&nbsp;Qiuming Cheng","doi":"10.1016/j.acags.2026.100318","DOIUrl":"10.1016/j.acags.2026.100318","url":null,"abstract":"<div><div>Tectonically induced fault systems create favorable environments for hydrothermal fluid migration and mineral precipitation, ultimately leading to the formation of certain deposits. Therefore, the numerical simulation of tectonic stress fields in mineralizing dynamic systems is crucial for understanding mineralization processes and for predictive modeling of ore prospectivity related to structure-controlled deposits. However, uncertainties in rock-physical parameters can undermine the reliability of simulation results. Although trial-and-error methods may seem to offer a solution to this issue, they typically necessitate extensive numerical simulations, leading to high computational costs. In this study, we developed a deep learning-based surrogate modeling method, using the Fankou lead-zinc deposit as a case study. This method enables rapid and accurate calculations of complex stress field distributions within mineralizing dynamic systems. We employed the Sobol method to analyze the sensitivity of the first, second and third principal stresses, to critial rock-physical parameters, including density, Poisson's ratio, and elastic modulus. This analysis allowed us to identify the critical parameters for the mineralization dynamics model. Subsequently, we utilized Latin hypercube sampling (LHS) to generate parameter sets within the key parameter space for numerical simulations, resulting in high-fidelity datasets of three distinct stress field distributions. Finally, we constructed an end-to-end surrogate model for each of the three stress fields based on deep neural networks, using the rock-physical parameters generated by LHS as input variables and the high-fidelity datasets obtained from numerical simulations as prediction variables. Evaluation metrics in the test dataset, including the correlation coefficient, root mean square error, mean relative error, and mean absolute error, indicate that the surrogate models perform well, effectively capturing the spatial distribution of the tectonic stress field within the mineralizing dynamic system. Our results also demonstrate that the average computational efficiency of the data-driven surrogate models is approximately 200 times greater than that of numerical simulation methods, providing an effective and rapid computational framework for parameter inversion and optimization related to the tectonic stress field in complex mineralizing dynamic systems.</div></div>","PeriodicalId":33804,"journal":{"name":"Applied Computing and Geosciences","volume":"29 ","pages":"Article 100318"},"PeriodicalIF":3.2,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977263","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Comparison of deep learning models for 1D magnetotelluric inversion","authors":"Hakim Saibi ,&nbsp;Abdelhadi Hireche ,&nbsp;Takeshi Tsuji ,&nbsp;Mohammed Y. Ali ,&nbsp;Ahmad B. Ahmad","doi":"10.1016/j.acags.2026.100320","DOIUrl":"10.1016/j.acags.2026.100320","url":null,"abstract":"<div><div>This paper presents a comparative study of three deep learning (DL) architectures for one-dimensional magnetotelluric (MT) data inversion: long short-term memory (LSTM), gated recurrent unit (GRU), and Informer models. We developed a comprehensive framework for generating realistic synthetic MT data, training the models, and evaluating their performance through multiple quantitative metrics. Our synthetic dataset comprised 100,000 samples with 25 periods spanning 10⁻³ to 10<span><math><msup><mrow></mrow><mrow><mn>3</mn></mrow></msup></math></span> seconds, created using statistical parameters derived from real MT data. Each model was trained on apparent resistivity and phase responses to recover subsurface resistivity profiles. The results show that the recurrent neural network architectures (LSTM and GRU) slightly outperform the attention-based Informer model, with the LSTM achieving the best performance (MSE of 0.06455 <span><math><mrow><mi>Ω</mi><mspace></mspace><msup><mrow><mtext>m</mtext></mrow><mrow><mn>2</mn></mrow></msup></mrow></math></span>, R<span><math><msup><mrow></mrow><mrow><mn>2</mn></mrow></msup></math></span> of 0.45234). Despite their differing architectures, all the models successfully captured the major subsurface resistivity contrasts. When applied to real MT data from the UAE, the tested models showed promising results in terms of reconstructing subsurface structures. Overall, this study demonstrates the viability of DL approaches for MT inversion, with potential applications in efficient field-based subsurface imaging.</div></div>","PeriodicalId":33804,"journal":{"name":"Applied Computing and Geosciences","volume":"29 ","pages":"Article 100320"},"PeriodicalIF":3.2,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977262","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Corrigendum to “Enhancing neuro-symbolic AI for mineral prediction via LLM-guided knowledge integration” [Appl. Comput. Geosci. 29 (2026) 100310]","authors":"Weilin Chen,&nbsp;Jiyin Zhang,&nbsp;Chenhao Li,&nbsp;Xiaogang Ma","doi":"10.1016/j.acags.2026.100321","DOIUrl":"10.1016/j.acags.2026.100321","url":null,"abstract":"","PeriodicalId":33804,"journal":{"name":"Applied Computing and Geosciences","volume":"29 ","pages":"Article 100321"},"PeriodicalIF":3.2,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146037979","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Automated identification of stylolites in geological whole-core images using hybrid deep learning networks","authors":"Saeid Sadeghnejad ,&nbsp;Ali Hasanloo ,&nbsp;Abolfazl Moslemipour ,&nbsp;Thorsten Schäfer","doi":"10.1016/j.acags.2026.100335","DOIUrl":"10.1016/j.acags.2026.100335","url":null,"abstract":"<div><div>Stylolites are complex geological features that usually appear as irregular surfaces. Fromed by pressure-driven dissolution, they are commonly filled with insoluble materials compared to surrounding host rocks. In underground reservoir flow (e.g., hydrogen, carbon dioxide, or natural gas storage), stylolites can influence reservoir fluid flow. They can act either as flow barriers or as flow pathways in reservoirs. Therefore, identifying stylolites in reservoir rocks is an essential task for the correct prediction of underground fluid flow. In this study, we develop a hybrid deep-learning workflow to automatically detect and classify stylolites in slabbed whole-core images. First, the “You Only Look Once” object-detection architecture is used to locate core-rock boundaries in raw whole core images containing multiple cores. These core images are then subdivided into 1,827 smaller image patches. We employ two convolutional neural network architectures of ReNet-50 and ResNeXt-50 for the final classification task. As the dataset, 150 m cores from three wells of a carbonate reservoir are used. The core image patches are first classified manually into five different image classes of stylolites, fractures, vertical plugs, horizontal plugs, and intact rock. As the dataset is imbalanced with respect to the number of images in each class, data-augmentation techniques such as flipping, cropping, rotation, and adjustments to brightness and contrast are implemented. This expands the dataset images to almost 16,000 images in total. Both networks are pre-trained on the ImageNet dataset and fine-tuned on the augmented dataset. To control overfitting, additional regularization techniques like dropout and adaptive learning-rate scheduling are used. Results show that ResNeXt-50 achieves the best classification performance of 92% on previously unseen whole-core images.</div></div>","PeriodicalId":33804,"journal":{"name":"Applied Computing and Geosciences","volume":"29 ","pages":"Article 100335"},"PeriodicalIF":3.2,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147396676","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessing future flood hazards in the major rivers of Bangladesh using CMIP6 projections and integrated hydrologic-hydraulic modeling","authors":"Aishia Fyruz Aishi ,&nbsp;Md Kamruzzaman Tusar ,&nbsp;Md Ariful Islam","doi":"10.1016/j.acags.2026.100327","DOIUrl":"10.1016/j.acags.2026.100327","url":null,"abstract":"<div><div>Climate change is expected to significantly alter flood dynamics in Bangladesh, one of the world's most flood-prone countries. Yet quantitative assessments of future flood hazards in this region remain sparse. This study pioneers an integrated approach to project future flood characteristics in Bangladesh's major rivers - Ganges, Jamuna, and Padma - for the period 2051-2070 under the SSP2-4.5 and SSP5-8.5 scenarios. We uniquely combine bias-corrected climate projections from the Coupled Model Intercomparison Project Phase 6 (CMIP6) with hydrological modeling using the Hydrologic Engineering Center's Hydrologic Modeling System and River Analysis System, providing a comprehensive assessment of flood hazard evolution. Our analysis, based on data from the Bangladesh Water Development Board (BWDB) and multiple Global Climate Models (GCMs), includes rainfall-runoff simulations, model calibration, flood frequency analysis, and one-dimensional steady-flow simulations. Results show projected increases in peak discharges of 6%, 10%, and 5% for the Ganges, Jamuna, and Padma rivers, respectively, compared to the baseline from 1996 to 2022. The Ganges shows the highest potential increase, up to 26.51% for the 100-year return period discharge, with the water level rising up to 0.71 m. Despite the highest increased discharge projections, the Jamuna exhibits unexpected decreases in projected water levels for design flood discharges of greater return periods compared to the baseline. The Padma shows moderate increases in extreme flood risk. These river-specific responses underscore the complexity of climate-driven hydrologic change and highlight the importance of integrated modeling approaches in guiding adaptive flood risk management and climate resilience planning in deltaic environments.</div></div>","PeriodicalId":33804,"journal":{"name":"Applied Computing and Geosciences","volume":"29 ","pages":"Article 100327"},"PeriodicalIF":3.2,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146188290","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"PetroMind: A multimodal petrographic model for rock image classification and lithological description generation","authors":"Zhongliang Chen ,&nbsp;Chaojie Zheng ,&nbsp;Mingming Zhang ,&nbsp;Zhaoqi Hu ,&nbsp;Jianchao Duan","doi":"10.1016/j.acags.2025.100314","DOIUrl":"10.1016/j.acags.2025.100314","url":null,"abstract":"<div><div>Recent studies have revealed the remarkable capabilities of multimodal large models (MLLMs) in general vision-and-language tasks, generating increasing interest in their application to geoscientific domains. Rock image recognition and lithological description constitute fundamental skills for geologists and represent one of the earliest application scenarios in geosciences where artificial intelligence technologies have been actively explored. As rock image recognition and description are inherently multimodal tasks, involving both visual and textual data, models are required to understand the interactions between modalities and how these interactions affect information transfer. However, the simultaneous comprehension of rock images and their corresponding lithological descriptions remains a significant challenge for current machine learning frameworks. The impact of multimodal co-training on downstream geoscientific performance is still unconfirmed. This study presents PetroMind, a domain-adapted multimodal petrographic foundation model built upon Qwen2.5-VL, together with SA-Rock, a novel LLM-based metric designed to assess the semantic accuracy of rock image generation descriptions to evaluate the capabilities of MLLMs in petrographic multimodal tasks. In the rock image classification task, PetroMind achieves performance comparable to ViT-Base-Patch16-224, with accuracy and Macro-F1 scores exceeding or approaching 97 %. This demonstrates PetroMind's strong capability in long-tailed image learning and highlights its effectiveness in substantially improving classification accuracy for few-shot rock image categories. In the lithological description generation task, PetroMind attains BLEU-4 and SA-Rock scores of 0.644 and 7.864, respectively, indicating good few-shot learning performance. The SA-Rock metric shows that the model produces highly accurate descriptions of rock structure, texture, and colour, while leaving considerable scope for improvement in the description of mineral composition. Ablation experiments further indicate that task-specific LoRA adapters are more effective for high-resource tasks such as image classification, whereas a single shared LoRA adapter demonstrates superior multi-task interaction in low-resource captioning scenarios. This study demonstrates the potential of MLLM-based architectures in jointly understanding rock images and their associated geological descriptions.</div></div>","PeriodicalId":33804,"journal":{"name":"Applied Computing and Geosciences","volume":"29 ","pages":"Article 100314"},"PeriodicalIF":3.2,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977530","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Accelerating multi-well placement optimization with machine learning-based proxy models for CO2 enhanced oil recovery and storage","authors":"Jinjie Mao,&nbsp;Ashkan Jahanbani Ghahfarokhi","doi":"10.1016/j.acags.2026.100317","DOIUrl":"10.1016/j.acags.2026.100317","url":null,"abstract":"<div><div>The urgent need to mitigate greenhouse gas emissions, particularly CO₂, has spurred extensive research into Geological CO₂ Storage. Utilizing depleted hydrocarbon reservoirs for both CO₂-enhanced oil recovery and storage presents a multifaceted approach to addressing environmental, economic, and technological challenges. This paper focuses on the strategic placement of wells to optimize both hydrocarbon recovery and CO₂ storage in partially depleted oil reservoirs.</div><div>Traditional well placement optimization is computationally intensive due to geological complexity, especially under multi-well configurations where each injector and producer is treated as a free decision variable rather than being fixed to predefined patterns, spacing, or areas. To address this challenge, this study introduces a novel data-driven well placement optimization framework that combines physics-guided intelligent data-driven models with evolutionary optimization algorithms. Quality Map-constrained random sampling is employed to generate training datasets, with input features including well geometries, petrophysical parameters, fluid properties, and time of flight that serves as a physics-based feature to improve model generalization. The outputs encompass cumulative fluids production over a fixed period. Various machine learning (ML) methods, including Multiple Perceptron (MLP), Extreme Gradient Boosting (XGBoost), and Linear Regression as a baseline model, are explored, yielding fast and accurate predictions. We investigate the strengths and limitations of ML algorithms across different dataset sizes, where MLP-based and XGBoost-based proxies performed desirably in terms of accuracy and computational efficiency, especially with larger datasets.</div><div>Integrating these proxies trained upon 1000 datapoints into a genetic algorithm (GA) effectively optimizes the injector and producer locations in a heterogeneous, three-dimensional reservoir, with net cash flow (NCF) as the objective function. MLP-GA and XGBoost-GA both outperform traditional simulator-based GA optimization in terms of computational efficiency while achieving accurate optimal well locations and NCF values.</div></div>","PeriodicalId":33804,"journal":{"name":"Applied Computing and Geosciences","volume":"29 ","pages":"Article 100317"},"PeriodicalIF":3.2,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147396667","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A systematic analysis of machine learning methods for rock wettability prediction: Algorithm performance and computational considerations","authors":"Emil von Rosen Vange,&nbsp;Abduljamiu Amao,&nbsp;Theis Solling","doi":"10.1016/j.acags.2026.100323","DOIUrl":"10.1016/j.acags.2026.100323","url":null,"abstract":"<div><div>Rock wettability fundamentally controls fluid distribution in porous media and directly impacts enhanced oil recovery applications. Traditional laboratory measurements, while accurate, are time-consuming and provide limited spatial resolution, which constrains comprehensive reservoir characterization. Here we present a machine learning approach for rapid wettability prediction from X-ray fluorescence (XRF) elemental data, addressing the need for efficient screening methods in reservoir analysis. We systematically evaluated seven machine learning algorithms combined with four preprocessing strategies on 243 data points from 61 rock samples across diverse lithologies. The most effective approach utilized Extreme Gradient Boosting (XGBoost) with Synthetic Minority Oversampling Technique (SMOTE), achieving 96.7% accuracy on held-out test data. The model demonstrated robust performance across all wettability classes, correctly identifying 100% of strongly water-wet samples, 75% of intermediate-wet samples, and 90% of strongly oil-wet samples. Learning curve and power analyses confirmed dataset adequacy, with statistical power of 0.87 for detecting medium effect sizes. Feature importance analysis identified rubidium, bromine, and arsenic as key elemental indicators of wettability classes. SHAP (SHapley Additive exPlanations) analysis revealed that higher rubidium concentrations are associated with more water-wet behavior, consistent with its large ionic radius modifying surface electrostatics at mineral-fluid interfaces. These findings align with transition metals influencing surface chemistry through oxidation state changes, while also revealing previously unrecognized roles for trace elements in wettability control.</div></div>","PeriodicalId":33804,"journal":{"name":"Applied Computing and Geosciences","volume":"29 ","pages":"Article 100323"},"PeriodicalIF":3.2,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146188291","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A shale-oil reservoir multi-mineral inversion method based on well-logging data using an improved random-forest approach: methodology and application","authors":"Jianhong Guo ,&nbsp;Changsheng Wang ,&nbsp;Yanmei Wang ,&nbsp;Wenjing Zhang ,&nbsp;Qing Zhao ,&nbsp;Hongyuan Wei ,&nbsp;Xin Nie","doi":"10.1016/j.acags.2026.100334","DOIUrl":"10.1016/j.acags.2026.100334","url":null,"abstract":"<div><div>High-precision quantitative characterisation of reservoir mineral composition is central to hydrocarbon exploration and development. However, integrated mineral–organic evaluation in organic-rich shale reservoirs remains technically challenging, owing to the high cost and limited scalability of elemental logging, and the susceptibility of conventional logs to organic-matter effects. The core innovation of this study is the construction of a comprehensive training dataset that explicitly decouples organic and mineral signals—achieved by integrating elemental logging, conventional logs, and core measurements. This dataset serves as high-fidelity labels, enabling a machine learning model to accurately predict mineral and kerogen contents using only conventional logs. Taking the Chang 7 Member shale-oil reservoir in the Ordos Basin as a case study, the workflow comprises three steps. First, inorganic mineral fractions are accurately inverted from elemental-logging data and used as baseline constraints. Second, a kerogen-content inversion model is calibrated by integrating conventional logs, elemental-logging results, and core measurements, enabling a robust separation of organic matter from the mineral matrix and yielding a complete mineral–organic reservoir-parameter dataset. Third, an improved Random Forest (RF) model optimized using the Sparrow–Bald Eagle Optimisation Algorithm (SBOA) is established, with conventional logs and derived total organic carbon (TOC) curves as input features to simultaneously predict the contents of five mineral groups and kerogen. Application results demonstrate that the SBOA–RF model achieves high predictive accuracy, with a mean relative error (MRE) of 6.71% for clay minerals and an MRE of 0.92, outperforming back-propagation neural networks (BPNN), gradient boosting decision trees (GBDT), and conventional approaches; moreover, SBOA is computationally more efficient than random search for hyperparameter optimisation. Porosity computed with a variable dry-rock skeleton model yields an average MRE of 9.06%, corroborating the reliability of the predicted mineral and organic contents. The model further exhibits strong generalisation in blind wells not used for training, with inversion results in good agreement with elemental-logging outputs and core X-ray diffraction (XRD) data. By reducing reliance on elemental logging, the proposed method provides a robust data foundation for reservoir-parameter evaluation and lithofacies classification in organic-rich shale intervals where elemental logs are unavailable, with substantial engineering relevance.</div></div>","PeriodicalId":33804,"journal":{"name":"Applied Computing and Geosciences","volume":"29 ","pages":"Article 100334"},"PeriodicalIF":3.2,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147396663","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Tractable approximate Gaussian inference for interpretable mineral prospectivity modelling","authors":"Lahiru M.A. Nagasingha ,&nbsp;Charles L. Bérubé ,&nbsp;Reza Ghanati","doi":"10.1016/j.acags.2026.100332","DOIUrl":"10.1016/j.acags.2026.100332","url":null,"abstract":"<div><div>Bayesian deep learning methods are powerful tools for mineral prospectivity modelling (MPM) thanks to their ability to quantify uncertainty. However, their reliance on stochastic gradient-based optimization or iterative variational updates presents challenges, and tools for interpreting the sources of MPM uncertainty remain limited. We address these challenges by applying tractable approximate Gaussian inference (TAGI), an analytical Bayesian method, to model the prospectivity of magmatic Ni (<span><math><mo>±</mo></math></span>Cu <span><math><mo>±</mo></math></span>Co <span><math><mo>±</mo></math></span>Platinum group elements) sulphide mineral systems across Canada. First, we use the adaptive synthetic sampling technique to handle class imbalance, and second, we benchmark the TAGI approach against a Bayesian neural network (BNN) with variational inference and a Monte Carlo (MC) dropout model. Comprehensive evaluations of performance, computational cost, and robustness show that TAGI outperforms the other Bayesian approaches, achieving an area under the receiver operating characteristic curve value of 0.91 and recall of 0.70 compared to 0.88 and 0.23 for BNN and 0.87 and 0.22 for MC dropout. Additionally, TAGI also trains 1.4 times faster per epoch and demonstrates 39% less performance degradation under added noise compared to the second-best performing method. Furthermore, the TAGI MPM results capture 85% of known test deposits within just 15% of the most prospective land area. Finally, Shapley additive explanations analysis confirms that the TAGI-based model learns geologically sensible relationships and reveals that lithology and geological period data are the dominant drivers of both its prospectivity predictions and its associated uncertainty. This result highlights that, when integrated into an MPM framework, TAGI provides not only a powerful predictive tool but also an effective decision-support system, as it quantifies both prospectivity and model reliability across geological domains.</div></div>","PeriodicalId":33804,"journal":{"name":"Applied Computing and Geosciences","volume":"29 ","pages":"Article 100332"},"PeriodicalIF":3.2,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147396666","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}