Acta Geophysica最新文献

{"title":"Geophysically constrained deep learning for lithology prediction using seismic data","authors":"Yan Huang,&nbsp;Wenliang Nie,&nbsp;Wei Huang,&nbsp;Gang Fang,&nbsp;Xiangfei Nie,&nbsp;Xiaotao Wen","doi":"10.1007/s11600-026-02050-9","DOIUrl":"10.1007/s11600-026-02050-9","url":null,"abstract":"<div><p>Deep learning methods have demonstrated promising prospects in seismic lithology prediction owing to their powerful nonlinear modelling capabilities. However, existing approaches still face two key challenges. First, a limited number of methods directly exploit the morphological features and temporal continuity of seismic waveforms for lithology prediction. Second, data-driven deep learning methods lack explicit constraints derived from geophysical principles, resulting in limited generalisation capability and physical consistency. To address these issues, this paper proposes a geophysically constrained deep learning method for lithology prediction. The proposed method directly takes seismic waveform time series as input features and employs a dual-branch temporal convolutional network architecture to capture short-range and long-range temporal dependencies, respectively. Combined with the multi-head self-attention mechanism of the transformer encoder, the method establishes global contextual relationships to effectively extract information related to seismic waveform morphological variations and temporal continuity. Meanwhile, a geophysical knowledge-guided loss function is constructed, which maps lithology prediction results to acoustic impedance through a rock physics model and generates synthetic seismic records via forward modelling. The network training process is then constrained by the degree of fit between synthetic and observed seismic data. Finally, the proposed method is applied to lithology prediction in the Pearl River Mouth Basin. Experimental results demonstrate that the proposed method significantly outperforms the single-branch temporal convolutional network method and the unconstrained dual-branch temporal convolutional network-transformer method. Furthermore, the proposed method can effectively identify thin-bed sandstones, and the predicted results exhibit favourable spatial continuity and physical plausibility.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132102","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Geophysically constrained deep learning for lithology prediction using seismic data","authors":"Yan Huang,&nbsp;Wenliang Nie,&nbsp;Wei Huang,&nbsp;Gang Fang,&nbsp;Xiangfei Nie,&nbsp;Xiaotao Wen","doi":"10.1007/s11600-026-02050-9","DOIUrl":"10.1007/s11600-026-02050-9","url":null,"abstract":"<div><p>Deep learning methods have demonstrated promising prospects in seismic lithology prediction owing to their powerful nonlinear modelling capabilities. However, existing approaches still face two key challenges. First, a limited number of methods directly exploit the morphological features and temporal continuity of seismic waveforms for lithology prediction. Second, data-driven deep learning methods lack explicit constraints derived from geophysical principles, resulting in limited generalisation capability and physical consistency. To address these issues, this paper proposes a geophysically constrained deep learning method for lithology prediction. The proposed method directly takes seismic waveform time series as input features and employs a dual-branch temporal convolutional network architecture to capture short-range and long-range temporal dependencies, respectively. Combined with the multi-head self-attention mechanism of the transformer encoder, the method establishes global contextual relationships to effectively extract information related to seismic waveform morphological variations and temporal continuity. Meanwhile, a geophysical knowledge-guided loss function is constructed, which maps lithology prediction results to acoustic impedance through a rock physics model and generates synthetic seismic records via forward modelling. The network training process is then constrained by the degree of fit between synthetic and observed seismic data. Finally, the proposed method is applied to lithology prediction in the Pearl River Mouth Basin. Experimental results demonstrate that the proposed method significantly outperforms the single-branch temporal convolutional network method and the unconstrained dual-branch temporal convolutional network-transformer method. Furthermore, the proposed method can effectively identify thin-bed sandstones, and the predicted results exhibit favourable spatial continuity and physical plausibility.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132161","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Resolving deep tectonic frameworks in extensional basins using advanced filtering of gravity data: case studies from Egypt and Nigeria","authors":"Ahmed M. Eldosouky,&nbsp;Stephen E. Ekwok,&nbsp;Anthony E. Akpan,&nbsp;Luan Thanh Pham,&nbsp;Ahmed Henaish,&nbsp;Mohamed A. Abd El‑Wahed","doi":"10.1007/s11600-026-02033-w","DOIUrl":"10.1007/s11600-026-02033-w","url":null,"abstract":"<div><p>This study provides a comprehensive analysis of the deep tectonic trends and structural framework of two classic extensional sedimentary basins: the Northern Western Desert Basins (NWDB) of Egypt and the Southern Benue Trough (SBT) of Nigeria. By utilizing Bouguer gravity data from the global WGM2012 model and applying the Enhanced Horizontal Gradient Amplitude (EHGA) and upward continuation filters, the research achieves high-precision delineation of subsurface geological structures. Unlike conventional gradient filters, the EHGA approach overcomes limitations by effectively resolving both shallow and deep-seated boundaries, resulting in clearer and more reliable edge detection. The findings reveal that the NWDB is mainly characterized by ENE–WSW and WNW–ESE trending lineaments, while the SBT is dominated by NE–SW and secondary NW–SE structural fabrics, highlighting their distinct tectonic histories and deformation events. The integration of edge enhancement results from different upward continuation data allows for multi-scale mapping of rift-related features, consistent with published seismic- and borehole-constrained structural interpretations. This methodology not only provides new insights into the geodynamic evolution, fault architecture, and petroleum potential of these basins but also establishes a robust framework for future structural analysis in similar sedimentary settings globally. The study further underscores the importance of gravity data and filtering techniques in regions where surface exposures are limited or seismic data are scarce, demonstrating their value in enhancing structural mapping, understanding tectonic complexities, and supporting natural resource exploration.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132095","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Resolving deep tectonic frameworks in extensional basins using advanced filtering of gravity data: case studies from Egypt and Nigeria","authors":"Ahmed M. Eldosouky,&nbsp;Stephen E. Ekwok,&nbsp;Anthony E. Akpan,&nbsp;Luan Thanh Pham,&nbsp;Ahmed Henaish,&nbsp;Mohamed A. Abd El‑Wahed","doi":"10.1007/s11600-026-02033-w","DOIUrl":"10.1007/s11600-026-02033-w","url":null,"abstract":"<div><p>This study provides a comprehensive analysis of the deep tectonic trends and structural framework of two classic extensional sedimentary basins: the Northern Western Desert Basins (NWDB) of Egypt and the Southern Benue Trough (SBT) of Nigeria. By utilizing Bouguer gravity data from the global WGM2012 model and applying the Enhanced Horizontal Gradient Amplitude (EHGA) and upward continuation filters, the research achieves high-precision delineation of subsurface geological structures. Unlike conventional gradient filters, the EHGA approach overcomes limitations by effectively resolving both shallow and deep-seated boundaries, resulting in clearer and more reliable edge detection. The findings reveal that the NWDB is mainly characterized by ENE–WSW and WNW–ESE trending lineaments, while the SBT is dominated by NE–SW and secondary NW–SE structural fabrics, highlighting their distinct tectonic histories and deformation events. The integration of edge enhancement results from different upward continuation data allows for multi-scale mapping of rift-related features, consistent with published seismic- and borehole-constrained structural interpretations. This methodology not only provides new insights into the geodynamic evolution, fault architecture, and petroleum potential of these basins but also establishes a robust framework for future structural analysis in similar sedimentary settings globally. The study further underscores the importance of gravity data and filtering techniques in regions where surface exposures are limited or seismic data are scarce, demonstrating their value in enhancing structural mapping, understanding tectonic complexities, and supporting natural resource exploration.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132094","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine learning–driven flood susceptibility mapping using multi-source geospatial data in the Lower Gangetic sub-basin, India","authors":"Ambrish Kumar,&nbsp;Vipin Kumar Mishra,&nbsp;Rama Pal,&nbsp;Rajesh Kaushal,&nbsp;M. Madhu","doi":"10.1007/s11600-026-02044-7","DOIUrl":"10.1007/s11600-026-02044-7","url":null,"abstract":"<div><p>Among natural hazards, floods cause some of the heaviest global losses, especially in alluvial river systems where topography, hydrology, land-use change, and shifting climate patterns interact in complex ways. This work presents a GIS-based flood susceptibility assessment for the Burhi Gandak Basin (BGB) in Bihar, India, built on eight Machine Learning (ML) algorithms. Nine flood-conditioning variables were employed in the analysis: rainfall; terrain-derived parameters comprising the Digital Elevation Model (DEM), slope, Topographic Wetness Index (TWI), and Stream Power Index (SPI); the hydrological parameter distance from the river; and land-surface descriptors including Land Use/Land Cover (LULC), Normalized Difference Vegetation Index (NDVI), and Normalized Difference Water Index (NDWI). These variables were chosen for their established hydrological relevance. Predictor redundancy was checked through Variance Inflation Factor (VIF) and tolerance diagnostics. Rainfall (VIF = 17.12) and NDWI (VIF = 11.58) showed comparatively high collinearity, yet both were kept in the model because of their hydrological importance and because ensemble learners are known to tolerate correlated inputs reasonably well. A flood inventory of 160 flood and 371 non-flood locations was used to build and test XGBoost, Bagging, Random Forest, LightGBM, Naïve Bayes, M5P, Artificial Neural Network (ANN), and Logistic Regression. Model quality was judged through the Matthews Correlation Coefficient (MCC), recall, precision, Cohen’s Kappa, F1-score, specificity, accuracy, and Area Under the ROC Curve (AUC-ROC), supplemented by 10-fold and spatial cross-validation. XGBoost delivered the strongest results, reaching 98.1% accuracy, 95.7% precision, 98.1% recall, 98.1% specificity, an F1-score of 0.969, a Kappa of 0.956, an MCC of 0.956, and an AUC-ROC of 0.975. SHAP (Shapley Additive exPlanations) analysis pointed to distance from river, DEM, rainfall, LULC, NDWI, slope, SPI, and TWI as the conditioning factors exerting the greatest influence, which strengthens confidence in the model's interpretability. Taken together, the proposed GIS-ML pipeline offers a dependable, interpretable, and transferable route to flood susceptibility mapping, with practical value for flood mitigation planning, sustainable land management, and climate-adaptive governance of vulnerable basins of the rivers.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132030","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine learning–driven flood susceptibility mapping using multi-source geospatial data in the Lower Gangetic sub-basin, India","authors":"Ambrish Kumar,&nbsp;Vipin Kumar Mishra,&nbsp;Rama Pal,&nbsp;Rajesh Kaushal,&nbsp;M. Madhu","doi":"10.1007/s11600-026-02044-7","DOIUrl":"10.1007/s11600-026-02044-7","url":null,"abstract":"<div><p>Among natural hazards, floods cause some of the heaviest global losses, especially in alluvial river systems where topography, hydrology, land-use change, and shifting climate patterns interact in complex ways. This work presents a GIS-based flood susceptibility assessment for the Burhi Gandak Basin (BGB) in Bihar, India, built on eight Machine Learning (ML) algorithms. Nine flood-conditioning variables were employed in the analysis: rainfall; terrain-derived parameters comprising the Digital Elevation Model (DEM), slope, Topographic Wetness Index (TWI), and Stream Power Index (SPI); the hydrological parameter distance from the river; and land-surface descriptors including Land Use/Land Cover (LULC), Normalized Difference Vegetation Index (NDVI), and Normalized Difference Water Index (NDWI). These variables were chosen for their established hydrological relevance. Predictor redundancy was checked through Variance Inflation Factor (VIF) and tolerance diagnostics. Rainfall (VIF = 17.12) and NDWI (VIF = 11.58) showed comparatively high collinearity, yet both were kept in the model because of their hydrological importance and because ensemble learners are known to tolerate correlated inputs reasonably well. A flood inventory of 160 flood and 371 non-flood locations was used to build and test XGBoost, Bagging, Random Forest, LightGBM, Naïve Bayes, M5P, Artificial Neural Network (ANN), and Logistic Regression. Model quality was judged through the Matthews Correlation Coefficient (MCC), recall, precision, Cohen’s Kappa, F1-score, specificity, accuracy, and Area Under the ROC Curve (AUC-ROC), supplemented by 10-fold and spatial cross-validation. XGBoost delivered the strongest results, reaching 98.1% accuracy, 95.7% precision, 98.1% recall, 98.1% specificity, an F1-score of 0.969, a Kappa of 0.956, an MCC of 0.956, and an AUC-ROC of 0.975. SHAP (Shapley Additive exPlanations) analysis pointed to distance from river, DEM, rainfall, LULC, NDWI, slope, SPI, and TWI as the conditioning factors exerting the greatest influence, which strengthens confidence in the model's interpretability. Taken together, the proposed GIS-ML pipeline offers a dependable, interpretable, and transferable route to flood susceptibility mapping, with practical value for flood mitigation planning, sustainable land management, and climate-adaptive governance of vulnerable basins of the rivers.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132474","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"GeoPI-STGMNet for joint inversion of petrophysical properties and fluid types in interbedded sandstone–mudstone reservoirs","authors":"Junjie Huang,&nbsp;Yufeng Gu,&nbsp;Jian Zhou,&nbsp;Yanjiao Jiang,&nbsp;Zhidong Bao","doi":"10.1007/s11600-026-02047-4","DOIUrl":"10.1007/s11600-026-02047-4","url":null,"abstract":"<div><p>Accurate and geologically consistent joint interpretation of petrophysical properties and fluid types remains challenging in thinly interbedded sandstone–mudstone reservoirs because of strong vertical heterogeneity, overlapping logging responses, and the coupling among shale content, pore structure, flow capacity, and fluid distribution. This study develops GeoPI-STGMNet for the simultaneous inversion of shale volume (<span>(V_{{{text{sh}}}})</span>), porosity (<span>(varphi)</span>), logarithmic permeability (<span>(log K)</span>), and water saturation (<span>(S_{{text{w}}})</span>), together with fluid-type identification from conventional well logs. The main methodological contribution is the integration of stratigraphic sequence modeling, inter-well topological information, petrophysical constraints, and task-gated multi-task learning within a unified framework for simultaneous petrophysical inversion and fluid-type identification. In this framework, depth-dependent stratigraphic information and inter-well relationships are jointly represented, while petrophysical consistency is used to regularize the multi-task predictions. Experiments on the SGH well series of the YRTOL Block, Youshashan Oilfield, show that GeoPI-STGMNet achieves an average <span>(R^{2})</span> of 0.942 for the joint prediction of <span>(V_{{{text{sh}}}})</span>, <span>(varphi)</span>, <span>(log K)</span>, and <span>(S_{{text{w}}})</span>, together with an Accuracy of 0.938 for seven-class fluid-type identification, outperforming the evaluated conventional interpretation and deep learning baselines. Continuous well-section results further show improved preservation of thin-bed boundaries, oil–water transition zones, and water-flooded intervals. The proposed framework provides a unified and physically interpretable approach for multi-parameter formation evaluation and fluid characterization in heterogeneous interbedded reservoirs.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132473","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Variability of horizontal magnetic tensor estimation and its impact on reference methods in magnetotellurics","authors":"Krzysztof Nowożyński,&nbsp;Waldemar Jóźwiak,&nbsp;Tomasz Ernst","doi":"10.1007/s11600-026-02006-z","DOIUrl":"10.1007/s11600-026-02006-z","url":null,"abstract":"<div><p>This study analyzes the instability of interstation magnetic transfer function estimation, known as the horizontal magnetic tensor (HMT), with a particular focus on how this variability affects the effectiveness of reference methods. HMT describes the relationship between horizontal magnetic field variations recorded at two observation sites and is increasingly used to investigate regional conductivity structures and to support reference processing techniques in magnetotellurics. Despite its widespread application, relatively little attention has been paid to its temporal stability and the implications of possible source-related variability. Previous research has shown that deviations from the ideal plane wave, manifested as the presence of an external vertical magnetic field component, exhibit seasonal characteristics influenced by day length and regional variations, especially for periods exceeding 1400 s. This phenomenon is attributed to variations in the source’s characteristics or the presence of an additional, unstable wave source. Notably, this additional field may also include a horizontal component, which could influence HMT estimation and consequently affect the effectiveness of reference methods. This study analyzed measurement series from observatories in Belsk (Poland) and Niemegk (Germany) over 22 years to determine whether the degree of deviation from the spatially uniform plane wave could have a local character. The results showed that, unlike tippers, HMT exhibits seasonal variability across the entire analyzed frequency range, indicating the variable nature of the source. Winter estimates differ significantly from summer ones, possibly due to shorter daylight duration. The impact of HMT variability on the effectiveness of reference methods was then analyzed. The results indicate that seasonality does not affect the performance of the classic single-reference method. However, in the case of the multi-reference method proposed by Egbert, HMT variability influences the results, potentially rendering the method ineffective.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s11600-026-02006-z.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132475","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"GeoPI-STGMNet for joint inversion of petrophysical properties and fluid types in interbedded sandstone–mudstone reservoirs","authors":"Junjie Huang,&nbsp;Yufeng Gu,&nbsp;Jian Zhou,&nbsp;Yanjiao Jiang,&nbsp;Zhidong Bao","doi":"10.1007/s11600-026-02047-4","DOIUrl":"10.1007/s11600-026-02047-4","url":null,"abstract":"<div><p>Accurate and geologically consistent joint interpretation of petrophysical properties and fluid types remains challenging in thinly interbedded sandstone–mudstone reservoirs because of strong vertical heterogeneity, overlapping logging responses, and the coupling among shale content, pore structure, flow capacity, and fluid distribution. This study develops GeoPI-STGMNet for the simultaneous inversion of shale volume (<span>(V_{{{text{sh}}}})</span>), porosity (<span>(varphi)</span>), logarithmic permeability (<span>(log K)</span>), and water saturation (<span>(S_{{text{w}}})</span>), together with fluid-type identification from conventional well logs. The main methodological contribution is the integration of stratigraphic sequence modeling, inter-well topological information, petrophysical constraints, and task-gated multi-task learning within a unified framework for simultaneous petrophysical inversion and fluid-type identification. In this framework, depth-dependent stratigraphic information and inter-well relationships are jointly represented, while petrophysical consistency is used to regularize the multi-task predictions. Experiments on the SGH well series of the YRTOL Block, Youshashan Oilfield, show that GeoPI-STGMNet achieves an average <span>(R^{2})</span> of 0.942 for the joint prediction of <span>(V_{{{text{sh}}}})</span>, <span>(varphi)</span>, <span>(log K)</span>, and <span>(S_{{text{w}}})</span>, together with an Accuracy of 0.938 for seven-class fluid-type identification, outperforming the evaluated conventional interpretation and deep learning baselines. Continuous well-section results further show improved preservation of thin-bed boundaries, oil–water transition zones, and water-flooded intervals. The proposed framework provides a unified and physically interpretable approach for multi-parameter formation evaluation and fluid characterization in heterogeneous interbedded reservoirs.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132029","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Variability of horizontal magnetic tensor estimation and its impact on reference methods in magnetotellurics","authors":"Krzysztof Nowożyński,&nbsp;Waldemar Jóźwiak,&nbsp;Tomasz Ernst","doi":"10.1007/s11600-026-02006-z","DOIUrl":"10.1007/s11600-026-02006-z","url":null,"abstract":"<div><p>This study analyzes the instability of interstation magnetic transfer function estimation, known as the horizontal magnetic tensor (HMT), with a particular focus on how this variability affects the effectiveness of reference methods. HMT describes the relationship between horizontal magnetic field variations recorded at two observation sites and is increasingly used to investigate regional conductivity structures and to support reference processing techniques in magnetotellurics. Despite its widespread application, relatively little attention has been paid to its temporal stability and the implications of possible source-related variability. Previous research has shown that deviations from the ideal plane wave, manifested as the presence of an external vertical magnetic field component, exhibit seasonal characteristics influenced by day length and regional variations, especially for periods exceeding 1400 s. This phenomenon is attributed to variations in the source’s characteristics or the presence of an additional, unstable wave source. Notably, this additional field may also include a horizontal component, which could influence HMT estimation and consequently affect the effectiveness of reference methods. This study analyzed measurement series from observatories in Belsk (Poland) and Niemegk (Germany) over 22 years to determine whether the degree of deviation from the spatially uniform plane wave could have a local character. The results showed that, unlike tippers, HMT exhibits seasonal variability across the entire analyzed frequency range, indicating the variable nature of the source. Winter estimates differ significantly from summer ones, possibly due to shorter daylight duration. The impact of HMT variability on the effectiveness of reference methods was then analyzed. The results indicate that seasonality does not affect the performance of the classic single-reference method. However, in the case of the multi-reference method proposed by Egbert, HMT variability influences the results, potentially rendering the method ineffective.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s11600-026-02006-z.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132472","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}