Engineering Geology最新文献_第3页

Vertical land movements assessment integrating Interferometric Synthetic Aperture Radar, in-situ data, and engineering-geological model: The case study of the reclaimed farmland of the Po River Delta (Italy) 基于干涉合成孔径雷达、原位数据和工程地质模型的陆地垂直运动评价——以意大利波河三角洲复垦农田为例

IF 8.4 1区工程技术 Q1 ENGINEERING, GEOLOGICAL

Engineering Geology

Pub Date : 2026-01-10 DOI: 10.1016/j.enggeo.2026.108544

Laura Pedretti , Pietro Teatini , Tommaso Letterio , Guadalupe Bru , Carolina Guardiola-Albert , Roberto Tomás , María I. Navarro-Hernández , Alessandro Bondesan , Yuri Taddia , Claudia Meisina

Low-elevation reclaimed coastlands face significant challenges from land subsidence and sea-level rise, making long-term monitoring of ground movements crucial to ensure infrastructure safety and preserve the natural environment. This study aims to reconstruct the long-term historical ground deformation of the reclaimed farmland in the Po River Delta by: i) integrating nearly 30 years of multisource, multi-temporal, and multisensor Interferometric Synthetic Aperture Radar (InSAR) satellite data (ERS-1/2, RADARSAT-1/2, Sentinel-1); ii) combining multisource InSAR datasets generated using different algorithms covering distinct or overlapping time periods (Sentinel-1 PSI, P-SBAS, and IPTA); and iii) developing a 3D engineering-geological model focused on the under-consolidated fine-grained deposits that are more prone to subsidence. By combining multiple monitoring techniques, this multidisciplinary approach reveals that land subsidence is primarily driven by autocompaction of under-consolidated finegrained sediments, locally accelerated by building construction, as evidenced by InSAR data. The highest subsidence rates occur in the youngest reclaimed areas with thicker under-consolidated fine-grained deposits.

While integrating multisensor InSAR datasets from diverse sources to reconstruct longterm ground deformation presents challenges, it also yields valuable insights. In this work, we demonstrate that heterogeneous datasets can still be valuable when interpreted carefully and that the feasibility of combining legacy and modern InSAR data for long historical deformation reconstruction is a practical challenge in real-world data integration.

Moreover, this comprehensive approach enables updating spatial and temporal records of land movement and identifying conditioning factors for inclusion in land movement susceptibility and risk maps supporting land planning.

低海拔的填海造地面临着地面沉降和海平面上升带来的重大挑战，因此对地面运动的长期监测对于确保基础设施安全和保护自然环境至关重要。利用近30 年多源、多时段、多传感器InSAR卫星（ERS-1/2、RADARSAT-1/2、Sentinel-1）数据，重建波河三角洲垦殖农田的长期历史地表变形；ii)结合使用不同算法生成的多源InSAR数据集，涵盖不同或重叠的时间段（Sentinel-1 PSI、P-SBAS和IPTA）；iii)针对更容易下沉的欠胶结细粒矿床建立三维工程地质模型。通过结合多种监测技术，这种多学科方法揭示了地面沉降主要是由未固结的细粒沉积物的自压实驱动的，正如InSAR数据所证明的那样，建筑施工在局部加速了地面沉降。沉降速率最高的是最年轻的填海地区，下固结细粒沉积物较厚。

{"title":"Vertical land movements assessment integrating Interferometric Synthetic Aperture Radar, in-situ data, and engineering-geological model: The case study of the reclaimed farmland of the Po River Delta (Italy)","authors":"Laura Pedretti , Pietro Teatini , Tommaso Letterio , Guadalupe Bru , Carolina Guardiola-Albert , Roberto Tomás , María I. Navarro-Hernández , Alessandro Bondesan , Yuri Taddia , Claudia Meisina","doi":"10.1016/j.enggeo.2026.108544","DOIUrl":"10.1016/j.enggeo.2026.108544","url":null,"abstract":"<div><div>Low-elevation reclaimed coastlands face significant challenges from land subsidence and sea-level rise, making long-term monitoring of ground movements crucial to ensure infrastructure safety and preserve the natural environment. This study aims to reconstruct the long-term historical ground deformation of the reclaimed farmland in the Po River Delta by: i) integrating nearly 30 years of multisource, multi-temporal, and multisensor Interferometric Synthetic Aperture Radar (InSAR) satellite data (ERS-1/2, RADARSAT-1/2, Sentinel-1); ii) combining multisource InSAR datasets generated using different algorithms covering distinct or overlapping time periods (Sentinel-1 PSI, P-SBAS, and IPTA); and iii) developing a 3D engineering-geological model focused on the under-consolidated fine-grained deposits that are more prone to subsidence. By combining multiple monitoring techniques, this multidisciplinary approach reveals that land subsidence is primarily driven by autocompaction of under-consolidated finegrained sediments, locally accelerated by building construction, as evidenced by InSAR data. The highest subsidence rates occur in the youngest reclaimed areas with thicker under-consolidated fine-grained deposits.</div><div>While integrating multisensor InSAR datasets from diverse sources to reconstruct longterm ground deformation presents challenges, it also yields valuable insights. In this work, we demonstrate that heterogeneous datasets can still be valuable when interpreted carefully and that the feasibility of combining legacy and modern InSAR data for long historical deformation reconstruction is a practical challenge in real-world data integration.</div><div>Moreover, this comprehensive approach enables updating spatial and temporal records of land movement and identifying conditioning factors for inclusion in land movement susceptibility and risk maps supporting land planning.</div></div>","PeriodicalId":11567,"journal":{"name":"Engineering Geology","volume":"363 ","pages":"Article 108544"},"PeriodicalIF":8.4,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957173","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Prediction of static liquefaction landslides in loess: Integrating triaxial shear parameters into the sliding-block model 黄土静力液化滑坡预测：将三轴剪切参数纳入滑块模型

IF 7.4 1区工程技术 Q1 ENGINEERING, GEOLOGICAL

Engineering Geology

Pub Date : 2026-01-08 DOI: 10.1016/j.enggeo.2026.108549

Fanyu Zhang, Jianbing Peng, Yixiao Zhang, Yapeng Wang, Tongwei Zhang

Static liquefaction landslides are among the most catastrophic geohazards, causing severe casualties and damage worldwide. The rapid mobility of this kind of landslide is the most spectacular in the Chinese Loess Plateau (CLP). However, it has been challenging to accurately predict initiation and failure in static liquefaction loess landslides. Here, we conduct a series of undrained triaxial compression tests on undisturbed and remolded loess samples in CLP, compiling a comprehensive database of undrained triaxial compression tests on saturated loess that combines current and published triaxial tests. Based on the database, we analyze the relationship between the normalized stress ratio and pore water pressure ratio within a stress state framework, then obtain two fitted parameters at the instability and failure points. The two ratios and the fitted parameters are integrated into the limit equilibrium equation to build a sliding-block model. The model accurately predicts the factor of safety against initiation and failure of nine static liquefaction loess landslides. The scanning electron microscope images and grain size distribution confirm that the packing structure affects shear behavior and the critical state locus in triaxial tests. Pore water pressure and boundary parameters in landslides are more sensitive to changes than those parameters extracted from the triaxial laboratory in the sliding-block model. Finally, we develop a hydromechanical coupling criterion for predicting the instability and failure of future static liquefaction landslides. These results show that the novel sliding-block model bridges the gap between triaxial shear parameters and slope field stability conditions. Our findings indicate that the model can serve as an effective method for predicting static liquefaction-induced landslides in loess and other soil types.

静态液化滑坡是最具灾难性的地质灾害之一，在世界范围内造成严重的人员伤亡和损失。这种滑坡的快速移动是中国黄土高原最为壮观的。然而，如何准确预测静态液化黄土滑坡的起裂和破坏一直是一个挑战。在这里，我们对CLP原状和重塑黄土样品进行了一系列不排水三轴压缩试验，并将现有和已发表的三轴试验结合起来，编制了饱和黄土不排水三轴压缩试验的综合数据库。在此基础上，分析了应力状态框架下归一化应力比与孔隙水压力比的关系，得到了失稳点和破坏点的两个拟合参数。将这两个比值和拟合参数积分到极限平衡方程中，建立滑块模型。该模型准确地预测了9个静态液化黄土滑坡的起爆破坏安全系数。扫描电镜图像和晶粒尺寸分布证实了填料结构对剪切性能和三轴试验临界状态轨迹的影响。滑块模型中孔隙水压力和边界参数比三轴实验室提取的参数变化更为敏感。最后，我们建立了一个预测未来静态液化滑坡失稳和破坏的水-力耦合准则。这些结果表明，新的滑块模型弥补了三轴剪切参数与边坡现场稳定条件之间的差距。研究结果表明，该模型可作为预测黄土和其他土壤类型静力液化诱发滑坡的有效方法。

{"title":"Prediction of static liquefaction landslides in loess: Integrating triaxial shear parameters into the sliding-block model","authors":"Fanyu Zhang, Jianbing Peng, Yixiao Zhang, Yapeng Wang, Tongwei Zhang","doi":"10.1016/j.enggeo.2026.108549","DOIUrl":"https://doi.org/10.1016/j.enggeo.2026.108549","url":null,"abstract":"Static liquefaction landslides are among the most catastrophic geohazards, causing severe casualties and damage worldwide. The rapid mobility of this kind of landslide is the most spectacular in the Chinese Loess Plateau (CLP). However, it has been challenging to accurately predict initiation and failure in static liquefaction loess landslides. Here, we conduct a series of undrained triaxial compression tests on undisturbed and remolded loess samples in CLP, compiling a comprehensive database of undrained triaxial compression tests on saturated loess that combines current and published triaxial tests. Based on the database, we analyze the relationship between the normalized stress ratio and pore water pressure ratio within a stress state framework, then obtain two fitted parameters at the instability and failure points. The two ratios and the fitted parameters are integrated into the limit equilibrium equation to build a sliding-block model. The model accurately predicts the factor of safety against initiation and failure of nine static liquefaction loess landslides. The scanning electron microscope images and grain size distribution confirm that the packing structure affects shear behavior and the critical state locus in triaxial tests. Pore water pressure and boundary parameters in landslides are more sensitive to changes than those parameters extracted from the triaxial laboratory in the sliding-block model. Finally, we develop a hydromechanical coupling criterion for predicting the instability and failure of future static liquefaction landslides. These results show that the novel sliding-block model bridges the gap between triaxial shear parameters and slope field stability conditions. Our findings indicate that the model can serve as an effective method for predicting static liquefaction-induced landslides in loess and other soil types.","PeriodicalId":11567,"journal":{"name":"Engineering Geology","volume":"255 1","pages":""},"PeriodicalIF":7.4,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957214","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A theoretical model for ground surface temperature under seasonal snow cover condition and numerical application in THM coupling of permafrost subgrade 季节性积雪条件下地表温度的理论模型及其在冻土路基THM耦合中的数值应用

IF 8.4 1区工程技术 Q1 ENGINEERING, GEOLOGICAL

Engineering Geology

Pub Date : 2026-01-08 DOI: 10.1016/j.enggeo.2026.108550

Gaosheng Li , Yundong Zhou , Xiangtian Xu , Yuqin Zhao , Henglin Xiao , Ruiqiang Bai , Yongtao Wang , Yuhang Liu

Snow possesses poor heat conduction properties due to its high reflectivity and low thermal conductivity, which significantly influences the energy exchange between the ground and the atmosphere. The randomness of the change in snow depth also exacerbates the complexity of the frost damage problem in highway engineering in cold regions. The conventional snow model is predicated on empirical formulas and a plethora of undetermined parameters, which renders it challenging to achieve the efficient integration of the dynamic change of snow depth and the multi-physical field coupling process of frozen soil. In this study, we proposed a simple and efficient method to equivalent the dynamic boundary conditions to periodic time-varying boundary conditions for simulating the variation of ground surface temperature under seasonal snow cover conditions. The accuracy of the method was verified using measured data. The present study developed a functional relationship between land surface temperature and ground surface temperature in seasonal snow cover areas by comparing the differences in the variation of ground surface temperature under the conditions of four kinds of snow melting time and seven kinds of maximum snow depths in the annual cycle. Finally, based on the engineering geological information of the surveyed Genhe-Mangui highway, we established a thermal-hydro-mechanical (THM) coupling model of permafrost snow-covered subgrade, and fully considered the randomness of the maximum snow depth. This work can provide a practical theoretical model for predicting the ground surface temperature of snow-covered ground, and offer a novel understanding of the frost damage caused by snow-covered subgrade.

雪的反射率高，导热系数低，热传导性能差，这对地面与大气之间的能量交换影响很大。雪深变化的随机性也加剧了寒冷地区公路工程冻损问题的复杂性。传统的积雪模型基于经验公式和大量的未定参数进行预测，难以实现积雪深度动态变化与冻土多物理场耦合过程的有效整合。本文提出了一种简单有效的方法，将动态边界条件等效为周期性时变边界条件，用于模拟季节积雪条件下地表温度的变化。用实测数据验证了该方法的准确性。通过比较4种融雪时间和7种年循环最大雪深条件下地表温度变化的差异，建立了季节积雪区地表温度与地表温度的函数关系。最后，基于已勘测的根满公路工程地质信息，充分考虑最大积雪深度的随机性，建立了多年冻土积雪路基热-水-力耦合模型。该研究为积雪地表温度的预测提供了一个实用的理论模型，并对积雪路基的冻损问题有了新的认识。

{"title":"A theoretical model for ground surface temperature under seasonal snow cover condition and numerical application in THM coupling of permafrost subgrade","authors":"Gaosheng Li , Yundong Zhou , Xiangtian Xu , Yuqin Zhao , Henglin Xiao , Ruiqiang Bai , Yongtao Wang , Yuhang Liu","doi":"10.1016/j.enggeo.2026.108550","DOIUrl":"10.1016/j.enggeo.2026.108550","url":null,"abstract":"<div><div>Snow possesses poor heat conduction properties due to its high reflectivity and low thermal conductivity, which significantly influences the energy exchange between the ground and the atmosphere. The randomness of the change in snow depth also exacerbates the complexity of the frost damage problem in highway engineering in cold regions. The conventional snow model is predicated on empirical formulas and a plethora of undetermined parameters, which renders it challenging to achieve the efficient integration of the dynamic change of snow depth and the multi-physical field coupling process of frozen soil. In this study, we proposed a simple and efficient method to equivalent the dynamic boundary conditions to periodic time-varying boundary conditions for simulating the variation of ground surface temperature under seasonal snow cover conditions. The accuracy of the method was verified using measured data. The present study developed a functional relationship between land surface temperature and ground surface temperature in seasonal snow cover areas by comparing the differences in the variation of ground surface temperature under the conditions of four kinds of snow melting time and seven kinds of maximum snow depths in the annual cycle. Finally, based on the engineering geological information of the surveyed Genhe-Mangui highway, we established a thermal-hydro-mechanical (THM) coupling model of permafrost snow-covered subgrade, and fully considered the randomness of the maximum snow depth. This work can provide a practical theoretical model for predicting the ground surface temperature of snow-covered ground, and offer a novel understanding of the frost damage caused by snow-covered subgrade.</div></div>","PeriodicalId":11567,"journal":{"name":"Engineering Geology","volume":"362 ","pages":"Article 108550"},"PeriodicalIF":8.4,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145939758","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Full-waveform CNN–transformer neural network for regional coseismic landslide susceptibility modeling: A case study of the 2022 Luding earthquake, China 全波形cnn -变压器神经网络区域同震滑坡易感性模拟——以2022年泸定地震为例

IF 8.4 1区工程技术 Q1 ENGINEERING, GEOLOGICAL

Engineering Geology

Pub Date : 2026-01-07 DOI: 10.1016/j.enggeo.2025.108520

Xiaolong Zhang , Shuai Huang , Binghai Gao

Earthquake-induced landslides are among the most common and destructive geological hazards in mountainous regions, posing significant threats to infrastructure, safety, and property. Traditional landslide susceptibility models primarily rely on simplified seismic intensity metrics, such as Peak Ground Acceleration, Velocity, or Arias intensity, which fail to capture the full time-frequency structure and duration effects of seismic motion, limiting both predictive accuracy and explainability. To address these limitations, this study proposes a novel approach for regional coseismic landslide susceptibility modeling that integrates full-waveform seismic data reconstruction with a hybrid Convolutional Neural Network (CNN)–Transformer deep learning model. The method involves a waveform reconstruction process for regions with sparse seismic data, utilizing waveform standardization, spectral decomposition, spatial interpolation, and group velocity constraints to synthesize three-component ground motion time histories with a frequency bandwidth of up to 25 Hz. A CNN-Transformer hybrid model is then employed to jointly analyze the reconstructed seismic waveforms and static environmental factors, such as topographic slope and lithology, enabling high-resolution spatial predictions of coseismic landslide susceptibility. Using the 2022 Luding earthquake as a case study, experimental results show that the integrated model significantly outperforms traditional models, achieving an AUC of 0.982 and an F1-score of 0.957, compared to 0.756 and 0.805 for the traditional model. Gradient-based explainability analysis reveals that the model focuses on the mainshock period within ±10 s of peak ground displacement (PGD) in regions with consistent predictions, while in areas with divergent predictions, it relies on tail waves, multi-phase shaking, and sustained seismic motion features, which are often missed by peak-based metrics. This study advances landslide susceptibility modeling by integrating full-waveform seismic data with static environmental factors, providing a more accurate and explainable framework for predicting coseismic landslide susceptibility. The approach offers significant potential for improving engineering applications and enabling cross-regional deployment in future seismic hazard assessments.

地震引发的山体滑坡是山区最常见和最具破坏性的地质灾害之一，对基础设施、安全和财产构成重大威胁。传统的滑坡敏感性模型主要依赖于简化的地震强度指标，如峰值地面加速度、速度或阿里亚斯强度，这些指标无法捕捉地震运动的全时频结构和持续时间效应，从而限制了预测的准确性和可解释性。为了解决这些局限性，本研究提出了一种新的区域同震滑坡易感性建模方法，该方法将全波形地震数据重建与混合卷积神经网络(CNN) -Transformer深度学习模型相结合。该方法包括对地震数据稀疏区域的波形重建过程，利用波形标准化、频谱分解、空间插值和群速度约束来合成频率带宽高达25 Hz的三分量地震动时程。然后利用CNN-Transformer混合模型，对重建的地震波形与地形坡度、岩性等静态环境因子进行联合分析，实现同震滑坡易感性的高分辨率空间预测。以2022年泸定地震为例，实验结果表明，综合模型的AUC为0.982，f1得分为0.957，显著优于传统模型，传统模型的AUC为0.756，f1得分为0.805。基于梯度的可解释性分析表明，在预测结果一致的地区，该模型主要关注地表峰值位移（PGD）±10 s内的主震周期，而在预测结果不一致的地区，该模型主要依赖于尾波、多相震动和持续地震运动特征，而这些特征往往是基于峰值的指标所忽略的。本研究通过将全波形地震数据与静态环境因素相结合，推进了滑坡易感性建模，为同震滑坡易感性预测提供了一个更准确、可解释的框架。该方法为改进工程应用和在未来地震灾害评估中实现跨区域部署提供了巨大潜力。

{"title":"Full-waveform CNN–transformer neural network for regional coseismic landslide susceptibility modeling: A case study of the 2022 Luding earthquake, China","authors":"Xiaolong Zhang , Shuai Huang , Binghai Gao","doi":"10.1016/j.enggeo.2025.108520","DOIUrl":"10.1016/j.enggeo.2025.108520","url":null,"abstract":"<div><div>Earthquake-induced landslides are among the most common and destructive geological hazards in mountainous regions, posing significant threats to infrastructure, safety, and property. Traditional landslide susceptibility models primarily rely on simplified seismic intensity metrics, such as Peak Ground Acceleration, Velocity, or Arias intensity, which fail to capture the full time-frequency structure and duration effects of seismic motion, limiting both predictive accuracy and explainability. To address these limitations, this study proposes a novel approach for regional coseismic landslide susceptibility modeling that integrates full-waveform seismic data reconstruction with a hybrid Convolutional Neural Network (CNN)–Transformer deep learning model. The method involves a waveform reconstruction process for regions with sparse seismic data, utilizing waveform standardization, spectral decomposition, spatial interpolation, and group velocity constraints to synthesize three-component ground motion time histories with a frequency bandwidth of up to 25 Hz. A CNN-Transformer hybrid model is then employed to jointly analyze the reconstructed seismic waveforms and static environmental factors, such as topographic slope and lithology, enabling high-resolution spatial predictions of coseismic landslide susceptibility. Using the 2022 Luding earthquake as a case study, experimental results show that the integrated model significantly outperforms traditional models, achieving an AUC of 0.982 and an F1-score of 0.957, compared to 0.756 and 0.805 for the traditional model. Gradient-based explainability analysis reveals that the model focuses on the mainshock period within ±10 s of peak ground displacement (PGD) in regions with consistent predictions, while in areas with divergent predictions, it relies on tail waves, multi-phase shaking, and sustained seismic motion features, which are often missed by peak-based metrics. This study advances landslide susceptibility modeling by integrating full-waveform seismic data with static environmental factors, providing a more accurate and explainable framework for predicting coseismic landslide susceptibility. The approach offers significant potential for improving engineering applications and enabling cross-regional deployment in future seismic hazard assessments.</div></div>","PeriodicalId":11567,"journal":{"name":"Engineering Geology","volume":"362 ","pages":"Article 108520"},"PeriodicalIF":8.4,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145939757","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

From hydro-meteorological thresholds towards an operational warning model for landslides at regional scale: A real-case application 从水文气象阈值到区域滑坡的业务预警模型：一个实际应用

IF 8.4 1区工程技术 Q1 ENGINEERING, GEOLOGICAL

Engineering Geology

Pub Date : 2026-01-07 DOI: 10.1016/j.enggeo.2026.108542

Sen Zhang , Gaetano Pecoraro , Da Huang , Jianbing Peng , Bei Zhang , Michele Calvello

Landslide prediction is essential for developing a landslide early warning system. Recently, hydro-meteorological thresholds combining rainfall and hydrological variables have demonstrated their effectiveness in enhancing the predictive capability of landslide occurrences. However, most territorial landslide early warning systems operational worldwide are primarily developed using only rainfall thresholds, totally neglecting the hydrological process that contributes to landslide initiation. In this study, we propose a three-step procedure aimed at developing a hydro-meteorological warning model intended for operational use employing multiple hydro-meteorological thresholds derived from a probabilistic analysis, using soil saturation and precipitation data retrieved from the ERA5-Land product. The model developed herein was tested in one of the warning zones defined by civil protection for the management of geo-hydrological risk in Campania region, Italy. Performance indicators derived adopting the “event, duration matrix, performance” (EDuMaP) method highlight that the hydro-meteorological warning model developed in this study—using real-time forecasts from the Integrated Forecasting System - High-Resolution (IFS-HRES) product—outperforms the current implemented warning model, which depends exclusively on precipitation forecasts. Specifically, the inclusion of soil saturation into the warning model leads to a significant reduction of false alarms. The results achieved herein demonstrate that hydro-meteorological thresholds can be effectively employed within landslide early warning systems for real-world applications at regional scale.

滑坡预测是建立滑坡预警系统的基础。近年来，结合降雨和水文变量的水文气象阈值在提高滑坡灾害预测能力方面的有效性得到了验证。然而，世界范围内运行的大多数区域滑坡预警系统主要是利用降雨阈值开发的，完全忽视了有助于滑坡启动的水文过程。在这项研究中，我们提出了一个三步程序，旨在利用从ERA5-Land产品中检索的土壤饱和度和降水数据，利用概率分析得出的多个水文气象阈值，开发一个用于业务使用的水文气象预警模型。本文开发的模型在意大利坎帕尼亚地区的一个预警区进行了测试，该预警区是由民防部门定义的，用于管理地质水文风险。采用“事件、持续时间矩阵、性能”（EDuMaP）方法得出的绩效指标突出表明，本研究开发的水文气象预警模型——使用高分辨率综合预报系统（IFS-HRES）产品的实时预报——优于目前实施的仅依赖降水预报的预警模型。具体来说，将土壤饱和度纳入预警模型可以显著减少误报。研究结果表明，水文气象阈值可以有效地应用于区域尺度的滑坡预警系统中。

{"title":"From hydro-meteorological thresholds towards an operational warning model for landslides at regional scale: A real-case application","authors":"Sen Zhang , Gaetano Pecoraro , Da Huang , Jianbing Peng , Bei Zhang , Michele Calvello","doi":"10.1016/j.enggeo.2026.108542","DOIUrl":"10.1016/j.enggeo.2026.108542","url":null,"abstract":"<div><div>Landslide prediction is essential for developing a landslide early warning system. Recently, hydro-meteorological thresholds combining rainfall and hydrological variables have demonstrated their effectiveness in enhancing the predictive capability of landslide occurrences. However, most territorial landslide early warning systems operational worldwide are primarily developed using only rainfall thresholds, totally neglecting the hydrological process that contributes to landslide initiation. In this study, we propose a three-step procedure aimed at developing a hydro-meteorological warning model intended for operational use employing multiple hydro-meteorological thresholds derived from a probabilistic analysis, using soil saturation and precipitation data retrieved from the ERA5-Land product. The model developed herein was tested in one of the warning zones defined by civil protection for the management of geo-hydrological risk in Campania region, Italy. Performance indicators derived adopting the “event, duration matrix, performance” (EDuMaP) method highlight that the hydro-meteorological warning model developed in this study—using real-time forecasts from the Integrated Forecasting System - High-Resolution (IFS-HRES) product—outperforms the current implemented warning model, which depends exclusively on precipitation forecasts. Specifically, the inclusion of soil saturation into the warning model leads to a significant reduction of false alarms. The results achieved herein demonstrate that hydro-meteorological thresholds can be effectively employed within landslide early warning systems for real-world applications at regional scale.</div></div>","PeriodicalId":11567,"journal":{"name":"Engineering Geology","volume":"362 ","pages":"Article 108542"},"PeriodicalIF":8.4,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957215","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Reconstruction and upscaling of local rock mass joint networks based on SinGAN 基于SinGAN的局部岩体节理网络重构与升级

IF 8.4 1区工程技术 Q1 ENGINEERING, GEOLOGICAL

Engineering Geology

Pub Date : 2026-01-07 DOI: 10.1016/j.enggeo.2026.108546

Jun Xiang , Qinjie Zhang , Xiling Liu , Xing Zhao , Tubing Yin , Zhiguo Li

Accurate identification and modeling of rock mass joint networks are crucial for assessing the quality and stability of the rock mass. However, traditional methods are often limited by sparse data and predefined statistical assumptions, making it difficult to capture the multi-scale self-similar characteristics of complex joint systems. To address this challenge, we propose a self-similarity upscaling approach for rock mass joints based on the SinGAN model. Leveraging its pyramid-like multi-scale generator, the method learns self-similar statistical features from a single joint image and enables accurate transmission of multi-scale structural information. Three field-acquired rock joint outcrop images were processed into binary images and used for model training. Model performance was evaluated based on joint intensity, orientation, and length distribution. The results show that SinGAN-generated images exhibit strong consistency with the originals, effectively preserving the variability of joint intensity, dominant orientation clusters, and the log-normal distribution of joint length. Integrating the upscaled images with a simplified GSI-based rock mass classification revealed a systematic decline in grading scores with increasing scale, consistent with the mechanical response of natural rock masses. Compared with traditional methods, the proposed approach leverages a data-driven framework to achieve unsupervised learning of the self-similarity statistical features of rock mass joint networks, significantly enhancing both the efficiency and accuracy of joint modeling in complex geological settings, and bridging the gap between laboratory-scale observations and field-scale predictions. This study highlights the potential of generative adversarial networks for quantitative multi-scale geological modeling and provides reliable data support for engineering design and geohazard risk assessment.

岩体节理网络的准确识别和建模对于评价岩体的质量和稳定性至关重要。然而，传统方法往往受到稀疏数据和预定义统计假设的限制，难以捕捉复杂关节系统的多尺度自相似特征。为了解决这一挑战，我们提出了一种基于SinGAN模型的岩体节理自相似升级方法。该方法利用其金字塔状的多尺度生成器，从单个关节图像中学习自相似的统计特征，从而实现多尺度结构信息的准确传输。将三幅野外采集的岩石节理露头图像处理成二值图像，用于模型训练。根据关节强度、方向和长度分布来评估模型的性能。结果表明，singan生成的图像与原始图像具有较强的一致性，有效地保留了关节强度、优势方向簇和关节长度的对数正态分布的可变性。将升级后的图像与基于简化gsi的岩体分类相结合，发现分级分数随着尺度的增加而系统性下降，这与天然岩体的力学响应一致。与传统方法相比，该方法利用数据驱动框架实现了岩体节理网络自相似统计特征的无监督学习，显著提高了复杂地质环境下节理建模的效率和准确性，弥合了实验室尺度观测与现场尺度预测之间的差距。该研究强调了生成对抗网络在定量多尺度地质建模中的潜力，并为工程设计和地质灾害风险评估提供了可靠的数据支持。

{"title":"Reconstruction and upscaling of local rock mass joint networks based on SinGAN","authors":"Jun Xiang , Qinjie Zhang , Xiling Liu , Xing Zhao , Tubing Yin , Zhiguo Li","doi":"10.1016/j.enggeo.2026.108546","DOIUrl":"10.1016/j.enggeo.2026.108546","url":null,"abstract":"<div><div>Accurate identification and modeling of rock mass joint networks are crucial for assessing the quality and stability of the rock mass. However, traditional methods are often limited by sparse data and predefined statistical assumptions, making it difficult to capture the multi-scale self-similar characteristics of complex joint systems. To address this challenge, we propose a self-similarity upscaling approach for rock mass joints based on the SinGAN model. Leveraging its pyramid-like multi-scale generator, the method learns self-similar statistical features from a single joint image and enables accurate transmission of multi-scale structural information. Three field-acquired rock joint outcrop images were processed into binary images and used for model training. Model performance was evaluated based on joint intensity, orientation, and length distribution. The results show that SinGAN-generated images exhibit strong consistency with the originals, effectively preserving the variability of joint intensity, dominant orientation clusters, and the log-normal distribution of joint length. Integrating the upscaled images with a simplified GSI-based rock mass classification revealed a systematic decline in grading scores with increasing scale, consistent with the mechanical response of natural rock masses. Compared with traditional methods, the proposed approach leverages a data-driven framework to achieve unsupervised learning of the self-similarity statistical features of rock mass joint networks, significantly enhancing both the efficiency and accuracy of joint modeling in complex geological settings, and bridging the gap between laboratory-scale observations and field-scale predictions. This study highlights the potential of generative adversarial networks for quantitative multi-scale geological modeling and provides reliable data support for engineering design and geohazard risk assessment.</div></div>","PeriodicalId":11567,"journal":{"name":"Engineering Geology","volume":"363 ","pages":"Article 108546"},"PeriodicalIF":8.4,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145950194","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Impacts of plant roots on debris-flow bed erosion in laboratory experiments 植物根系对室内泥石流床侵蚀的影响

IF 8.4 1区工程技术 Q1 ENGINEERING, GEOLOGICAL

Engineering Geology

Pub Date : 2026-01-06 DOI: 10.1016/j.enggeo.2025.108513

Anna J. van den Broek, Dagmar T. Mennes, Maarten G. Kleinhans, Lonneke Roelofs, Jana Eichel, Daniel Draebing, Tjalling de Haas

Debris flows often increase in size due to bed erosion and entrainment, enhancing their hazardous potential. However, the effects of plant rooting on debris-flow erosion on ubiquitous vegetated slopes remain unknown, which hinders debris-flow hazard assessment. Here, we investigated the effects of roots on debris-flow bed erosion using scaled experiments in a 5 m long, 0.3 m wide laboratory flume with an erodible bed. Roots of fast-growing Sorghum bicolor (Sudan grass) seedlings were used as proxies for tree roots to quantify the effect of varying rooting characteristics on erosion. Our results indicate that erosion decreases non-linearly with increasing Root Length Density (

R L D

) and Root Area Ratio (

R A R

). Increases in either parameter enhance root–soil contact, thereby improving soil stability and reducing erosion. Among the two,

R L D

, and thus the combined effect of root length and root density, appears most influential, as

R A R

does not capture the three-dimensional structure of the root system. Our experimental results suggest that increasing root-soil contact at the debris-flow bed reduces erosion, decreasing or even preventing debris-flow volume growth. These findings imply that alterations in vegetation characteristics, such as those resulting from forest fires or reforestation, affect debris-flow erosion and open up possibilities for biogeomorphic scale experiments for slope processes.

由于河床侵蚀和夹带，泥石流的规模往往会增大，从而增加了其潜在的危险性。然而，在普遍存在的植被斜坡上，植物根系对泥石流侵蚀的影响尚不清楚，这阻碍了泥石流危害评估。在这里，我们研究了根对泥石流河床侵蚀的影响，在一个长5米、宽0.3米的实验室水槽中进行了规模实验，并带有可侵蚀的河床。以速生高粱（苏丹草）幼苗根系为代表，量化不同根系特征对侵蚀的影响。结果表明：土壤侵蚀随根长密度（RLD）和根面积比（RAR）的增加呈非线性减少；任何一个参数的增加都能增强根与土壤的接触，从而提高土壤稳定性，减少侵蚀。在这两者中，RLD以及根长度和根密度的综合效应似乎影响最大，因为RAR不能捕捉根系的三维结构。我们的实验结果表明，增加泥石流床根部与土壤的接触可以减少侵蚀，减少甚至阻止泥石流体积的增长。这些发现表明，植被特征的改变，如森林火灾或重新造林所造成的变化，会影响泥石流侵蚀，并为斜坡过程的生物地貌尺度实验开辟了可能性。

{"title":"Impacts of plant roots on debris-flow bed erosion in laboratory experiments","authors":"Anna J. van den Broek, Dagmar T. Mennes, Maarten G. Kleinhans, Lonneke Roelofs, Jana Eichel, Daniel Draebing, Tjalling de Haas","doi":"10.1016/j.enggeo.2025.108513","DOIUrl":"10.1016/j.enggeo.2025.108513","url":null,"abstract":"<div><div>Debris flows often increase in size due to bed erosion and entrainment, enhancing their hazardous potential. However, the effects of plant rooting on debris-flow erosion on ubiquitous vegetated slopes remain unknown, which hinders debris-flow hazard assessment. Here, we investigated the effects of roots on debris-flow bed erosion using scaled experiments in a 5 m long, 0.3 m wide laboratory flume with an erodible bed. Roots of fast-growing <em>Sorghum bicolor</em> (Sudan grass) seedlings were used as proxies for tree roots to quantify the effect of varying rooting characteristics on erosion. Our results indicate that erosion decreases non-linearly with increasing Root Length Density (<span><math><mrow><mi>R</mi><mi>L</mi><mi>D</mi></mrow></math></span>) and Root Area Ratio (<span><math><mrow><mi>R</mi><mi>A</mi><mi>R</mi></mrow></math></span>). Increases in either parameter enhance root–soil contact, thereby improving soil stability and reducing erosion. Among the two, <span><math><mrow><mi>R</mi><mi>L</mi><mi>D</mi></mrow></math></span>, and thus the combined effect of root length and root density, appears most influential, as <span><math><mrow><mi>R</mi><mi>A</mi><mi>R</mi></mrow></math></span> does not capture the three-dimensional structure of the root system. Our experimental results suggest that increasing root-soil contact at the debris-flow bed reduces erosion, decreasing or even preventing debris-flow volume growth. These findings imply that alterations in vegetation characteristics, such as those resulting from forest fires or reforestation, affect debris-flow erosion and open up possibilities for biogeomorphic scale experiments for slope processes.</div></div>","PeriodicalId":11567,"journal":{"name":"Engineering Geology","volume":"362 ","pages":"Article 108513"},"PeriodicalIF":8.4,"publicationDate":"2026-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145902892","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Regional-scale inventory and initial analysis of liquefaction triggered by the 2025 Mw 7.7 Mandalay earthquake, Myanmar 2025年缅甸曼德勒7.7兆瓦地震引发的区域尺度液化的库存和初步分析

IF 7.4 1区工程技术 Q1 ENGINEERING, GEOLOGICAL

Engineering Geology

Pub Date : 2026-01-05 DOI: 10.1016/j.enggeo.2026.108543

Sotiris Valkaniotis, George Papathanassiou, Janusz Wasowski, Maria Taftsoglou, Ranjan Kumar Dahal

引用次数: 0

Bio-geotechnical reinforcement of purple soil slopes: The synergistic effects of xanthan gum biopolymer and planting density 紫色土坡的生物土工加固：黄原胶生物聚合物与种植密度的协同效应

IF 8.4 1区工程技术 Q1 ENGINEERING, GEOLOGICAL

Engineering Geology

Pub Date : 2026-01-03 DOI: 10.1016/j.enggeo.2025.108540

Zhongkai Liu , Qian Peng , Qi Yang , Huafeng Deng , Yao Xiao , Daxiang Liu , Yueshu Yang

Bio-geotechnical engineering in the Three Gorges Reservoir Area (TGRA) faces prominent stability challenges during the vegetation establishment period. Xanthan gum (XG) demonstrates potential for enhancing early-stage stability of root-soil composites, but the interdependent effects between XG content and planting density remain unclear. This study systematically investigates the synergistic effects between XG content (0 %, 0.6 % and 1.2 % by dry soil mass) and Cynodon dactylon planting density (18 and 36 g/m²) on purple soil stability through multi-mode tests. Results demonstrate that increased XG content significantly enhances soil cohesion and aggregate stability, while the 1.2 % XG content markedly inhibits plant germination and growth. Notably, mechanistic analysis reveals that under the combined effects of evaporation and soil layer thickness, high XG content induces surface cementation through upward capillary migration of XG molecules and soil cations, followed by crystallization and cross-linking upon dehydration. This process promotes the formation of a white cementation layer, which subsequently leads to preferential cracking, and seeds are consequently forced to germinate from within these cracks. Furthermore, in thicker soil layers, high XG content contributes to prolonged moisture retention and induces localized anaerobic conditions. This anaerobic environment enhances the activity of anaerobic microorganisms, leading to the formation of black metal sulfide deposits. The higher planting density (36 g/m²) can mitigate these effects by improving soil aeration and drainage through root development. Finally, Entropy-weighted TOPSIS evaluation identifies 0.6 % XG with 36 g/m² planting density as the recommended combination, effectively balancing immediate soil reinforcement with sustainable vegetation establishment. Compared to untreated purple soil, this optimized treatment achieves a 95.67 % increase in disintegration resistance index, a 196.64 % increase in cohesion, and a 31.09 % reduction in surface crack ratio. The findings could provide theoretical guidance for bio-geotechnical engineering design in TGRA and similar regions, offering references for biopolymer-vegetation interaction studies.

三峡库区生物岩土工程在植被建立期面临着突出的稳定性挑战。黄原胶（XG）具有提高根土复合材料早期稳定性的潜力，但XG含量与种植密度之间的相互作用尚不清楚。本研究通过多模式试验系统研究了干土XG含量（0、0.6%和1.2%）和长尾草种植密度（18和36 g/m2）对紫色土稳定性的协同效应。结果表明，增加XG含量可显著提高土壤黏聚力和团聚体稳定性，而1.2%的XG含量可显著抑制植物的萌发和生长。值得注意的是，机理分析表明，在蒸发和土层厚度的共同作用下，高XG含量通过XG分子和土壤阳离子向上的毛细迁移诱导表面胶结，脱水后发生结晶和交联。这个过程促进了白色胶结层的形成，随后导致优先开裂，种子因此被迫从这些裂缝中发芽。此外，在较厚的土层中，高XG含量有助于延长水分保持时间并诱导局部厌氧条件。这种厌氧环境增强了厌氧微生物的活性，导致黑色金属硫化物沉积物的形成。较高的种植密度（36 g/m2）可以通过根系发育改善土壤的通气和排水，从而缓解这些影响。最后，熵权TOPSIS评价确定了0.6% XG和36 g/m2种植密度作为推荐组合，有效地平衡了立即土壤加固和可持续植被的建立。与未处理紫土相比，优化处理后的紫土抗崩解指数提高了95.67%，黏聚力提高了196.64%，表面裂缝率降低了31.09%。研究结果可为TGRA及类似地区的生物岩土工程设计提供理论指导，为生物聚合物与植被相互作用的研究提供参考。

{"title":"Bio-geotechnical reinforcement of purple soil slopes: The synergistic effects of xanthan gum biopolymer and planting density","authors":"Zhongkai Liu , Qian Peng , Qi Yang , Huafeng Deng , Yao Xiao , Daxiang Liu , Yueshu Yang","doi":"10.1016/j.enggeo.2025.108540","DOIUrl":"10.1016/j.enggeo.2025.108540","url":null,"abstract":"<div><div>Bio-geotechnical engineering in the Three Gorges Reservoir Area (TGRA) faces prominent stability challenges during the vegetation establishment period. Xanthan gum (XG) demonstrates potential for enhancing early-stage stability of root-soil composites, but the interdependent effects between XG content and planting density remain unclear. This study systematically investigates the synergistic effects between XG content (0 %, 0.6 % and 1.2 % by dry soil mass) and <em>Cynodon dactylon</em> planting density (18 and 36 g/m<sup>2</sup>) on purple soil stability through multi-mode tests. Results demonstrate that increased XG content significantly enhances soil cohesion and aggregate stability, while the 1.2 % XG content markedly inhibits plant germination and growth. Notably, mechanistic analysis reveals that under the combined effects of evaporation and soil layer thickness, high XG content induces surface cementation through upward capillary migration of XG molecules and soil cations, followed by crystallization and cross-linking upon dehydration. This process promotes the formation of a white cementation layer, which subsequently leads to preferential cracking, and seeds are consequently forced to germinate from within these cracks. Furthermore, in thicker soil layers, high XG content contributes to prolonged moisture retention and induces localized anaerobic conditions. This anaerobic environment enhances the activity of anaerobic microorganisms, leading to the formation of black metal sulfide deposits. The higher planting density (36 g/m<sup>2</sup>) can mitigate these effects by improving soil aeration and drainage through root development. Finally, Entropy-weighted TOPSIS evaluation identifies 0.6 % XG with 36 g/m<sup>2</sup> planting density as the recommended combination, effectively balancing immediate soil reinforcement with sustainable vegetation establishment. Compared to untreated purple soil, this optimized treatment achieves a 95.67 % increase in disintegration resistance index, a 196.64 % increase in cohesion, and a 31.09 % reduction in surface crack ratio. The findings could provide theoretical guidance for bio-geotechnical engineering design in TGRA and similar regions, offering references for biopolymer-vegetation interaction studies.</div></div>","PeriodicalId":11567,"journal":{"name":"Engineering Geology","volume":"362 ","pages":"Article 108540"},"PeriodicalIF":8.4,"publicationDate":"2026-01-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145893883","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Seismic site characterization using satellite-derived terrain morphometry and geological data: A machine learning approach for predominant frequency prediction 利用卫星衍生的地形形态测量学和地质数据进行地震现场表征：一种用于主要频率预测的机器学习方法

IF 8.4 1区工程技术 Q1 ENGINEERING, GEOLOGICAL

Engineering Geology

Pub Date : 2026-01-02 DOI: 10.1016/j.enggeo.2025.108541

Harish Thakur, P. Anbazhagan

<div><div>Predominant frequency (<em>fo</em>) characterization across large seismically active regions remains challenging due to limited field measurements and cost constraints. Existing <em>fo</em> mapping approaches rely exclusively on spatial interpolation methods (kriging, inverse distance weighting, natural neighbor) that redistribute measured values without incorporating terrain morphometry, geological context, or subsurface parameters as predictors. This study develops a DEM-based machine learning methodology for regional-scale <em>fo</em> prediction in the Himalayan region and Indo-Gangetic Plains, addressing critical data scarcity in earthquake-prone developing countries. We compiled 4400 <em>fo</em> measurements from 26 published HVSR studies using systematic georeferencing procedures to ensure spatial consistency. The methodology employs a two-stage regression kriging framework: (1) stacked ensemble machine learning models trained on 20 predictor variables using GLO-30 DEM morphometric parameters (elevation, slope, curvature indices), geological classifications, and bedrock depth information to capture nonlinear terrain-frequency relationships; and (2) ordinary kriging of model residuals to account for spatial correlation patterns. Cross-validation partitioning ensures unbiased residuals, while Bayesian optimization determines optimal hyperparameters for base model selection. Feature importance analysis reveals that valley bottom identification (MRVBF), geological formation characteristics, and bedrock depth provide primary predictive capability (Shapley values ∼0.15–0.18), demonstrating that terrain morphometry and subsurface parameters effectively control <em>fo</em> variation at regional scales. The stacked ensemble achieves R<sup>2</sup> = 0.516 and RMSE = 0.634 log units, with variogram analysis revealing spatial correlation extending 7.3 km and structured variance accounting for 52 % of model residuals. High-resolution <em>fo</em> maps (50 m grid) generated for Delhi, Kathmandu, and Dhaka differentiate site response zones: low frequencies (<1.0 Hz) in deep sedimentary basins versus high frequencies (>3.0 Hz) in bedrock-controlled areas.</div><div>This work represents the first regional-scale application of DEM-derived terrain morphometry for direct <em>fo</em> prediction, utilizing a much larger compiled dataset for this purpose than previous basin-scale studies. Unlike previous studies that employed purely interpolation techniques without predictive parameters, this hybrid framework integrates physical predictors (terrain morphometry, geology, bedrock depth) with spatial modelling to produce more robust <em>fo</em> maps. Results demonstrate that incorporating satellite-derived morphometric and geological parameters—readily available globally—significantly enhances prediction reliability beyond interpolation-only approaches. This cost-effective methodology enables preliminary seismic hazard assessment in data-sparse mounta

由于有限的现场测量和成本限制，大型地震活跃区域的主要频率（fo）表征仍然具有挑战性。现有的测绘方法完全依赖于空间插值方法（克里格法、逆距离加权法、自然邻域法），这些方法重新分配测量值，而没有将地形形态、地质背景或地下参数作为预测因素。本研究开发了一种基于dem的机器学习方法，用于喜马拉雅地区和印度恒河平原的区域尺度预测，解决了地震多发发展中国家的关键数据短缺问题。我们使用系统的地理参考程序，从26项已发表的HVSR研究中收集了4400个测量值，以确保空间一致性。该方法采用两阶段回归克里格框架：(1)利用gloo -30 DEM形态参数（高程、坡度、曲率指数）、地质分类和基岩深度信息，训练20个预测变量的堆叠集成机器学习模型，捕捉非线性地形-频率关系；(2)对模型残差进行普通克里格，以解释空间相关模式。交叉验证分区确保残差无偏，贝叶斯优化确定最优超参数，用于基础模型选择。特征重要性分析表明，谷底识别（MRVBF）、地质构造特征和基岩深度提供了主要的预测能力（Shapley值~0.15 ~ 0.18），表明地形形态和地下参数在区域尺度上有效控制了变化。叠加集合的R2 = 0.516,RMSE = 0.634 log units，方差分析显示空间相关延伸7.3 km，结构方差占模型残差的52. %。为德里、加德满都和达卡制作的高分辨率地图（50 m网格）区分了场地响应区域：深沉积盆地的低频（<1.0 Hz）与基岩控制区的高频（>3.0 Hz）。

{"title":"Seismic site characterization using satellite-derived terrain morphometry and geological data: A machine learning approach for predominant frequency prediction","authors":"Harish Thakur, P. Anbazhagan","doi":"10.1016/j.enggeo.2025.108541","DOIUrl":"10.1016/j.enggeo.2025.108541","url":null,"abstract":"<div><div>Predominant frequency (<em>fo</em>) characterization across large seismically active regions remains challenging due to limited field measurements and cost constraints. Existing <em>fo</em> mapping approaches rely exclusively on spatial interpolation methods (kriging, inverse distance weighting, natural neighbor) that redistribute measured values without incorporating terrain morphometry, geological context, or subsurface parameters as predictors. This study develops a DEM-based machine learning methodology for regional-scale <em>fo</em> prediction in the Himalayan region and Indo-Gangetic Plains, addressing critical data scarcity in earthquake-prone developing countries. We compiled 4400 <em>fo</em> measurements from 26 published HVSR studies using systematic georeferencing procedures to ensure spatial consistency. The methodology employs a two-stage regression kriging framework: (1) stacked ensemble machine learning models trained on 20 predictor variables using GLO-30 DEM morphometric parameters (elevation, slope, curvature indices), geological classifications, and bedrock depth information to capture nonlinear terrain-frequency relationships; and (2) ordinary kriging of model residuals to account for spatial correlation patterns. Cross-validation partitioning ensures unbiased residuals, while Bayesian optimization determines optimal hyperparameters for base model selection. Feature importance analysis reveals that valley bottom identification (MRVBF), geological formation characteristics, and bedrock depth provide primary predictive capability (Shapley values ∼0.15–0.18), demonstrating that terrain morphometry and subsurface parameters effectively control <em>fo</em> variation at regional scales. The stacked ensemble achieves R<sup>2</sup> = 0.516 and RMSE = 0.634 log units, with variogram analysis revealing spatial correlation extending 7.3 km and structured variance accounting for 52 % of model residuals. High-resolution <em>fo</em> maps (50 m grid) generated for Delhi, Kathmandu, and Dhaka differentiate site response zones: low frequencies (<1.0 Hz) in deep sedimentary basins versus high frequencies (>3.0 Hz) in bedrock-controlled areas.</div><div>This work represents the first regional-scale application of DEM-derived terrain morphometry for direct <em>fo</em> prediction, utilizing a much larger compiled dataset for this purpose than previous basin-scale studies. Unlike previous studies that employed purely interpolation techniques without predictive parameters, this hybrid framework integrates physical predictors (terrain morphometry, geology, bedrock depth) with spatial modelling to produce more robust <em>fo</em> maps. Results demonstrate that incorporating satellite-derived morphometric and geological parameters—readily available globally—significantly enhances prediction reliability beyond interpolation-only approaches. This cost-effective methodology enables preliminary seismic hazard assessment in data-sparse mounta","PeriodicalId":11567,"journal":{"name":"Engineering Geology","volume":"362 ","pages":"Article 108541"},"PeriodicalIF":8.4,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145919765","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0