Study On BEMD Method And PM Model And Their Application In Geochemical Anomalies Identification

How to separate the geochemical noises and regional geologic background information from the original nonlinear and non-stationary geochemical data, and extract small scale local anomalies caused by geological factors related to ore-prospecting such as ore-body, altered minerals and concealed rock body is one of the most important tasks of exploration geochemists. However, because of the complex geological processes and multiple impositions of mineralizing process, the concentration and distribution of geochemical elements display a random, non-linear and non-stationary characteristics, which makes the extraction and decomposition of geochemical information very difficult. The traditional geochemical information extraction and decomposition methods, such as trend-surface methods and eigenvalue decomposition method, mostly do not take the spatial heterogeneity of geochemical data into consideration, therefore the decomposition results cannot reflect the real situation of the geochemical elements distribution. Hence it is difficult to explain. Especially in view of the present problems in the process of prospecting the coverage area, the layer shielding effect makes the geochemical information collection very weak; consequently, traditional information extraction method is limited in application.In recent years, many scholars introduced the Hilbert-Huang Transform (HHT) and its expanded theories into the application areas of earth science, and through Empirical Mode Decomposition (EMD) to get some intrinsic mode functions (IMFs) to extract meaningful geological information. Unlike the time-frequency analysis method based on Fourier transform, (such as short time Fourier transform (STFT), Wigner-Ville distribution (WVD) and Wavelet transform (WT)), HHT method by introducing the concept of intrinsic mode function, does not require pre-determined decomposition basis functions; through EMD the nonlinear and non-stationary signal is decomposed adaptively into a finite number of the IMFs in the time series, and then the meaning instantaneous frequency of each IMF is obtained with the Hilbert transform (HT), then the signal characteristics can be analyzed in energy-frequency-time domain. Anisotropic diffusion Perona Malik model (PM) model can suppress the noises while retaining the edge character of the image. This advantage makes it widely used in the field of image processing. However, when processing high intensity noise data, the PM model produces strong "block effect", and in worst case it will lead to distortion of data. Catte model is an important improvement model of PM model for dealing with the strong intensity "noise"data "failure":the high intensity noise data was smoothed to reduce the noise gradient values firstly, then PM diffusion model was used for edge character identification of the data. Bi-dimensional empirical mode decomposition (BEMD), PM model and Catte model was applied in the coverage area of geochemical information decomposition to extract the weak anomaly information in this study.Our study area is a forest vegetation covered area located in Southwestern Fujian province, which is at the south of continental volcanic Cu polymetallic metallogenic belt, eastern part of our country. This area has intensive Mesozoic volcanic activities, and has undergone multi-cycle movements and magmatic activity and mineralization many times, which has potential for porphyry-epithermal Cu polymetallic deposits. At present, Cu polymetallic deposits have been discovered including Luoboling, Zijinshan, Bitian, Xiping, and so on. The stream sediment dataset representing 1:20,0000 scale was chosen to investigate the mineralization features. Previous studies have demonstrated that the Cu polymetallic deposits are strongly correlated to element associations of Ag-As-Pb-Mo-Sb-Zn-Sn-Au-Cd-Cu. BEMD and its improved algorithms were used to decompose the complex and superposition geochemical information, PM model and its improved algorithm were used to evaluate spatial patterns and recognize anomalies of Copper mineralization of Ag-As-Pb-Mo-Sb-Zn-Sn-Au-Cd-Cu, which describes the different frequency domain scales of spatial distribution of mineral resources, and for quantitative simulation and identification of geochemical anomalies provides new methods, The main conclusions as follows:(1)The geochemical combination anomlies information were decomposed by BIMFs-PM modelThe signals fuse the Bi-intrinsic mode functions (BIMFs) which represent different frequency scales with the same specific meaning can indicate a particular meaning of the decomposed signal. The fused BIMFs were embedded PM model for secondary noise suppression and edge feature extraction, which can highlight the essential characteristics of the decomposed signal. In view of the characteristics of the geochemical data, the BEMD method based on Kriging interpolation considered the spatial structure between geochemical data. The decomposed BIMFs satisfy local orthogonality without the problem of mode mixing, and thus provide better quality of the decomposed data. The anomaly group of Copper mineralization related to elements Ag-As-Pb-Mo-Sb-Zn-Sn-Au-Cd-Cu were decomposed by BEMD based on Kriging interpolation. The BIMF2, BIMF3 and BIMF4 which indicating the different frequency scales geochemical anomalies were combined for BIMF234. The PM model was applied to smooth the precision errors during data processing, and enhance the anomaly information of the final anomaly map. The results showed that BIMFs-PM can not only extract the mineralized anomalies, but also can extract information of formations or granite (e.g., BIMF3).(2) The geochemical combination anomlies information were decomposed by BEEMD methodBecause BEMD decomposition relies on much the basis of experience, many key technologies remain to be discussed (such as the choice of suitable interpolation technology when dealing with different types of data, the mode mixing problem improvement, and so on). In this study, to get the good quality of decomposed BIMFs, for the data computing resources request and algorithm speed consideration, we proposed bi-dimensional ensemble empirical mode decomposition (BEEMD) based on Cubic spline interpolation to relieve the 2D mode mixing problem. Firstly, BEEMD, and BEMD method were applied to Lena image decomposing of recognition, and the results show that:BEMD decomposition results have obvious 2D mode mixing. The BEEMD method turn the Lena image as a two-dimensional whole data to decomposition, which can eliminate the 2D mode mixing problem, the decomposed BIMFs have a better looking, and some parts of the image, such as hair, nose, arm, and the edge of the hat are made clearer. Then, BEMD and BEEMD are compared to identify geochemical anomalies using a case study from Cu polymetallic mineralization district in southwestern Fujian province, (China). The results showed that:because of the 2D mode mixing of BEMD, the decomposed BIMFs have false anomalies and false information, and BEEMD can eliminate the 2D mode mixing for the decomposing process, give more robust and reliable results than BEMD, making the data easier to interpret; the BIMFs and the residue obtained by BEEMD can be combined into new components, which can represent geochemical anomalies, unwanted noise and the background of the study area.(3) Catte model and its application in geochemical anomalies identificationFirstly, PM model and Catte model were applied to Lena image decomposing for noise processing and the feature character recognition of Lena image with different intensity noise. The results showed that:with the noise intensity of Lena image increasing, the PM model fails gradually; but Catte model get a good work of denoising and retaining image edge character information. Then Catte model was used to extract geochemical anomalies, and the results were compared with the results which are obtained by the PM model. Catte model and PM model are compared to identify geochemical anomalies of Cu polymetallic mineralization district in southwestern Fujian province, (China). Results show that:the geochemical anomalies extracted by PM model show obvious "block", not consistent with actual distribution trend of the elements. While the obtained anomalies spatial distribution trend by Catte model was consistent with BIMFs-PM model and S-A fractal model and the geochemical anomalies obtained by Catte are closer to local geochemical anomalies, which show that Catte model used for extracting geochemical anomalies of combined elements is effective.