台灣上部地殼構造之地震衰減與多尺度、菲涅爾體積 波線追跡法的斷層掃描成像研究;Seismic Attenuation and Multi-Scale Fresnel Volume Tomographic Studies of Upper Crustal Structure in Taiwan

NCU Institutional Repository > 地球科學學院 > 地球物理研究所 > 博碩士論文 > Item 987654321/97708

請使用永久網址來引用或連結此文件: https://ir.lib.ncu.edu.tw/handle/987654321/97708

題名:	台灣上部地殼構造之地震衰減與多尺度、菲涅爾體積波線追跡法的斷層掃描成像研究;Seismic Attenuation and Multi-Scale Fresnel Volume Tomographic Studies of Upper Crustal Structure in Taiwan
作者:	蘇揚達;Suryantara, M. Syahdan Akbar
貢獻者:	地球科學學系
關鍵詞:	地震處理;軟體開發;地震衰減;Vp-Vs 關係;網格系統;菲涅爾體積;Seismic Processing;Software Development;Seismic Attenuation;Vp Vs relationship;Mesh System;Fresnel Volume
日期:	2025-08-28
上傳時間:	2025-10-17 11:48:49 (UTC+8)
出版者:	國立中央大學
摘要:	我的碩士論文研究最初源自於對不同人員和方法提出的各種層析成像模型進行驗證階段時提出的問題和擔憂。透過驗證（主要透過走時和波形到達模擬），我發現不同提出和發布的模型各有優缺點。我的論文重點在於我想要研究的主題。我主要涵蓋的主題包括：（1）保幅地震處理，以便準確拾取振幅和走時資訊；（2）設計基於混合DNN、STFT、EMD和 STA/LTA方法的半自動拾取策略；自主開發的用於拾取振幅和初至p波和s波走時的工具；（3）研究、估計和分離對散射和固有衰減的影響；（4）根據近地面測井記錄收集的資訊以及主動震源勘探推斷的速度模型，建立經驗Vp-Vs關係。目標是消除基於走時被動源層析成像研究反演的速度分佈估計過高而導致的走時誤差。主動和被動層析成像研究提出的速度模型之間的不匹配會在嘗試合併模型時產生人為的不連續性。這項工作更新了參考速度模型。因此，這項嘗試可以與以下工作連結。（5）建立必要的初始一維和三維 Vp-Vs 參考模型，作為菲涅耳體積射線走時層析成像初始模型設定過程的一部分。（6）為模型建構建構逼真的非結構化或結構化模型參數化。最後，我們提出了一種替代方法，即應用（7）菲涅爾體積射線追蹤和線性化走時反演方案來估計台灣的速度模型。在所有可用的用於層析成像研究的射線追蹤方案中，菲涅爾體積射線追蹤是首次嘗試。多尺度方法實現並研究頻率相關效應對旅行時間和振幅的影響，同時對透過線性化旅行時間層析反演得出的速度模型進行空間取樣。馬可夫鏈蒙特卡羅 (MCMC) 方法用於研究合理收斂標準下的可能模型。通常，建立具有所需屬性的馬可夫鏈並不難。更困難的問題是確定需要多少步驟才能在可接受的誤差內收斂到平穩模型分佈。透過檢查任意源接收器位置可以達到平穩分佈。評估收斂性的標準經驗方法是運行幾個獨立的模擬馬可夫鏈，並檢查所有採樣參數的鏈間與鏈內方差之比。;My master thesis study was initially motivated by questions and concerns raised during the verification stages on various tomographic models being proposed by different people and methods. Through the verifications, mainly by travel-time and waveform arrival simulations, there are many pros and cons on different proposed and published models. My thesis emphasis on the subjects that I want to pursuit. The subjects that I tend to cover are: (1) Amplitude preserving seismic processing in order to have accurate picks on amplitude and travel-time information, (2) Design semi-automatic picking strategy by hybrid DNN, STFT, EMD and STA/LTA methods. A self-developed tools for the convenience to pick amplitude and first arrived p-wave and s-wave travel-time. (3) Study, estimate and separate the effects on scattering and intrinsic attenuation. (4) Establish the empirical Vp-Vs relationship from the information collected by near-surface well-log and inferred velocity models from active source surveys. The goal is to remove the travel-time errors caused by the over-estimated velocity distribution inverted from travel-time based passive source tomography studies. The miss-matches between the velocity models proposed by the active and passive tomography studies can produce artificial discontinuities while attempting to merge models. This effort update the reference velocity models. Therefore, the attempt can be linked to the following efforts. (5) Establish a necessary initial 1D and 3D Vp-Vs reference models as a part of initial model setup procedure for Fresnel volume ray travel-time tomography. (6) Construct realistic unstructured or structured model parameterization for model building. Lastly, we propose an alternative approach by applying (7) Fresnel volume ray tracing and linearized travel time inversion schemes to estimate the velocity model in Taiwan. Fresnel volume ray tracing is the first attempt among all the available ray tracing schemes being used and applied for tomographic studies. The multi-scaled approach which implement and investigate frequency-dependent effects on travel-time and amplitude while spatially sampled the velocity model being derived through linearized travel-time tomographic inversion. Markov chain Monte Carlo (MCMC) methods are used to study possible models under reasonable convergence criteria. Usually it is not hard to construct a Markov chain with the desired properties. The more difficult problem is to determine how many steps are needed to converge to the stationary model distribution within an acceptable error. The stationary distribution can be reached by checking arbitrary source-receiver positions. A standard empirical method to assess convergence is to run several independent simulated Markov chains and check that the ratio of inter-chain to intra-chain variances for all the sampled parameters sampled.
顯示於類別:	[地球物理研究所] 博碩士論文

文件中的檔案:

檔案	描述	大小	格式	瀏覽次數
index.html		0Kb	HTML	30	檢視/開啟

在NCUIR中所有的資料項目都受到原著作權保護.

社群 sharing

資料載入中.....