本日選文（共 5 篇）。

[1794] 針對希臘北部 Prinos CO2 儲存場址潛在誘發地震之混合地震動模型（混合 GMM）

Hybrid Ground‐Motion Model for Potential Induced Seismicity at the Prinos CO 2 Storage Site (Northern Greece)

  • 期刊：Geophysics (GeoScienceWorld) — Advance Access
  • Published：Wed, 13 May 2026 00:00:00 GMT
  • DOI：10.1785/0120250287/730782
  • 原文連結：連結

Summary (EN) The paper presents the first region-specific hybrid ground-motion model (GMM) developed for potential injection-induced seismicity at the Prinos CO2 storage site in northern Greece. The framework combines stochastic simulations calibrated against regional recordings and constrained by an a priori seismological model, with iterative parameter optimization and quantification of epistemic uncertainty via suites of acceptable parameter combinations. A synthetic dataset (2≤Mw≤6, hypocentral distances up to 80 km) and an expanded empirical dataset including machine-learning-derived events are integrated to support regional hazard assessment and to inform methodologies for induced-earthquake ground-motion characterization in Europe.

重點摘要（繁中） 本研究提出針對希臘北部 Prinos 二氧化碳儲存場潛在注入誘發地震所開發的首個區域性混合地震動模型。該框架將以先驗地震學模型為約束的隨機模擬與區域觀測資料校準結合，透過迭代優化參數並保留可接受參數組合以量化知識不確定性。研究使用合成資料集（2≤Mw≤6、震源距離最遠80 km）並整合包含機器學習衍生事件的擴充經驗資料，以提供本區域地震動及危害評估的依據並推進歐洲誘發地震地震動表徵的方法學。

[1795] 透過非遍歷建模探討義大利中部地震動之時空變異性

Exploring Spatiotemporal Variability of Ground Motion Through Nonergodic Modeling in Central Italy

  • 期刊：Geophysics (GeoScienceWorld) — Advance Access
  • Published：Wed, 13 May 2026 00:00:00 GMT
  • DOI：10.1785/0120250253/730781
  • 原文連結：連結

Summary (EN) The paper presents a new nonergodic, nonstationary (NENS) ground‑motion prediction model for central Italy built from an extensive strong‑motion dataset that includes the 2016–2017 Amatrice–Visso–Norcia sequence. The model is a nested mixed‑effects regression for Fourier amplitude spectra (0.2–30 Hz) using ∼434,000 recordings from 9,948 events (Mw 1.8–6.5), explicitly including site-, path-, and source‑specific random effects that vary across three time phases (presequence, main sequence, postsequence). Residuals show significant spatiotemporal variability and a locally variable reduction in median ground‑motion amplitudes during dissipative phases, highlighting the need to account for spatial and temporal nonergodicity and the value of high‑density instrumentation to capture transient changes.

重點摘要（繁中） 本文提出一個適用於義大利中部的新型非遍歷且非平穩（NENS）地動預測模型，使用包含2016–2017 Amatrice–Visso–Norcia震序在內的大量強震觀測資料。該模型為傅立葉振幅頻譜（0.2–30 Hz）之巢狀混合效應回歸，使用約434,000個紀錄、9,948個事件（矩震級1.8–6.5），明確納入隨時間於前序、主序及後序三階段變化的場址、波徑與震源隨機效應。殘差分析顯示顯著的時空變異性，且在耗散性階段（如流體飽和或高度散射條件）中位地動振幅局部降低，強調在義大利中部需同時考量空間與時間的非遍歷性，並顯示密集儀器網能捕捉瞬變效應的價值。

[1796] 透過自適應重要取樣快速傳播地震危害中的知識不確定性

Fast Propagation of Epistemic Uncertainty in Seismic Hazard via Adaptive Importance Sampling

  • 期刊：Geophysics (GeoScienceWorld) — Advance Access
  • Published：Wed, 13 May 2026 00:00:00 GMT
  • DOI：10.1785/0120250205/730780
  • 原文連結：連結

Summary (EN) Conventional logic-tree approaches for representing epistemic uncertainty in probabilistic seismic hazard analysis (PSHA) can introduce bias and become computationally expensive as the number of variables grows. The authors propose a Gaussian Population Monte Carlo adaptive importance sampling (G‑PMC AIS) framework that formulates PSHA with importance sampling to approximate an optimal sampling density, enabling fast estimation of mean and fractile hazards and sensitivity metrics. Numerical experiments report speedups of 13–224× over logic-tree methods and 123–3775× over Monte Carlo with continuous distributions while keeping mean-hazard coefficient of variation below 1% and fractile Kolmogorov–Smirnov distance below 5%, making the approach practical for classical parameterized PSHA models.

重點摘要（繁中） 傳統用於表示機率地震危險分析（PSHA）認識論不確定性的邏輯樹方法，因為粗糙的離散化在變數增多時會導致估計偏差且計算量大幅增加。作者提出一種基於重要性抽樣的高斯族群蒙地卡羅自適應重要性抽樣（G‑PMC AIS）框架，透過近似最佳抽樣密度及其邊際化來快速估算平均與分位危險以及進行敏感度分析。數值實驗顯示，相較於邏輯樹方法加速13–224倍、相較於連續分布蒙地卡羅加速123–3775倍，同時將平均危險的變異係數控制在1%以下、分位估計的Kolmogorov–Smirnov距離低於5%，可用於傳統參數化的PSHA專案以大幅降低計算成本。

[1797] 使用深度神經網路的單站三分量地震定位

Single‐Station Three‐Component Seismic Event Localization Using Deep Neural Networks

  • 期刊：Seismological Research Letters (SRL) — Advance Access
  • Published：Wed, 13 May 2026 00:00:00 GMT
  • DOI：10.1785/0220250285/730779
  • 原文連結：連結

Summary (EN) The paper addresses limitations of conventional single‐station seismic localization—dependence on regional velocity models for epicentral distance and unreliable polarization‐based back‐azimuth estimates under low SNR—by proposing a deep neural network approach. Two networks take three‐component waveforms to independently predict epicentral distance and back azimuth in a fully data‑driven framework requiring no prior information or manual intervention. The model was trained on a dataset of 367,000 high‑quality events and was evaluated systematically, including comparisons with mainstream localization methods and robustness tests under low‑SNR conditions.

重點摘要（繁中） 本文針對傳統單站三分量地震定位的兩大限制——震中距離依賴區域速度模型且誤差會傳遞、以及基於波形極化的方位估計在低訊噪比下不可靠，提出一種深度神經網路方法。該方法以兩個網路分別輸入三分量波形來獨立預測震中距離與回方位，完全由資料驅動，無需先驗資訊或人工介入。作者以36.7萬筆高品質事件進行訓練，並進行包括與主流定位模型比較及低訊噪比下健壯性測試等系統性評估。

[1817] 數百萬年尺度上張性斷層系統位移速率的時空變化及其對地震風險的影響

Spatial and Temporal Variations in Slip Rate Over Millions of Years on an Extensional Fault System and Implications for Seismic Hazard

  • 期刊：Tectonics (AGU/Wiley) — eTOC
  • Published：Wed, 13 May 2026 03:41:31 -0700
  • DOI：10.1029/2025TC009239
  • 原文連結：連結

Summary (EN) The authors analyze slip rates from a seismically imaged, presently inactive extensional fault network offshore NW Australia that formed during extension from the Upper Triassic (209.5 Ma) to the Early Cretaceous (137.3 Ma). They find low maximum slip rates (<0.1 mm/yr) but substantial spatial and temporal variation on million‑year timescales (10^6–10^7 years), including along‑strike migration of the location of peak slip by several kilometers. Although system‑wide earthquake recurrence remains similar, recurrence rates on individual faults change with the slip‑rate variations, introducing uncertainty that should be incorporated into probabilistic seismic hazard assessments.

重點摘要（繁中） 作者分析澳洲西北外海一個地震剖面成像的、目前不活動的伸展性斷層網絡，該區在上三叠紀（209.5 Ma）至早白堊紀（137.3 Ma）期間因大陸裂解而伸展形成。研究發現最大滑動速率低（<0.1 mm/yr），但在百萬到千萬年（10^6–10^7 年）尺度上存在顯著的空間與時間變化，包含最大滑動位置沿走向移動數公里。雖然整個斷層系統的地震復發率整體上相似，但單一斷層的復發率會隨滑動速率變化而改變，這類不確定性應納入機率性地震危險度評估。