(1. 中南大學 有色金屬成礦預測教育部重點實驗室，長沙 410083；
2. 中南大學 地球科學與信息物理學院，長沙 410083；3. 有色金屬華東地質勘查局，南京 210007)
摘 要: 對野外實測大地電磁數(shù)據(jù)進行反演解釋時，由于人文噪聲的干擾和采集的數(shù)據(jù)不完全，增加了地球物理反演解釋的難度，所以研究不同噪聲和缺失數(shù)據(jù)信息對大地電磁正則化反演的影響非常必要。利用正則化理論研究不同噪聲、數(shù)據(jù)“信息不完全”下對大地電磁反演的影響，分析解釋最平坦模型、最平滑模型和最小模型的特點。結果表明：合適的正則化反演初始模型的選擇，對反演的結果至關重要。最平坦模型、最平滑模型較好，最小模型效果一般；隨著噪聲的增加，反演效果逐漸變差；數(shù)據(jù)信息不完全的時候，如果其它先驗信息豐富，仍然能夠得到較好的反演結果。
關鍵字: 高斯噪聲；缺失數(shù)據(jù)；正則化反演；大地電磁測深
regular inversion of magnetotelluric response
（1. Key Laboratory of Metallogenic Prediction of Nonferrous Metals, Ministry of Education,
Central South University, Changsha 410083, China;
2. School of Geosciences and Info-Physics, Central South University, Changsha 410083, China;
3. East China Mineral Exploration and Development Bureau for Nonferrous Metal, Nanjing 210007, China）
Abstract:Because human noise and incomplete data information were existed when the data was collected in the field, the difficulty of magnetotelluric (MT) inversion for interpretation increases naturally. Therefore, the study on magnetotelluric regular inversion on human noise and incomplete data information becomes necessary. Using the results of different amplitudes of Gaussian noise and the data of incomplete information studied with regularization theory, the characteristics of the flattest model, the smoothest model and the smallest model were analyzed and interpreted. The results show that: a suitable initial model is the essential choice to inversion, the flattest model and the smoothing model are better for magnetotelluric inversion than the smallest model; as the noise increases, the inversion becomes worse. If the additional prior information is rich, a good inversion results can still be got from the data incomplete information.
Key words: Gaussian noise; missing data; regularized inversion; magnetotelluric sounding
