Magnetotelluric impedance estimation Based on clustering analysis

De-noising or removing noise the magnetotelluric (MT) data using conventional methods is dependent on ratio in time series collected. To treat this problem, we propose a clustering analysis method based Impedance Euclidean Distance (IED) and Improved K-Means Cluster Method (IKCM). Firstly, divided into same length segment estimated impedance of interest frequency. Secondly, IED evaluated by real imaginary part impedance, which conduct sample IKCM. Thirdly IKCM performanced optimal class determined according to coherence, compactness smoothness criterion. The resistivity phase are result best class. Finally, evaluate proposed synthetic set field collected at Tianba Village Yunnan Province, China. Experimental results demonstrate that presented approach can be used select out free less from set, evaluation also indicate superior robust terms continuity curves.