A Denotsing Technology Based On Mulitple Domain Combination And Its Application In Metallic Ore Exploration

With the rapid development of the society and economy, near-surface mineral resourceshave been gradually depleted, which makes deep prospection quite popular now, for oreexploration has developed from shallow surface to the second deep space(500m~2000m). Asis known to us, geological conditions in deposits areas are always very complex; ore bodiescan have different forms, uneven distribution, deep dip angle and small scale; the differenceof wave impedance between ore body and surrounding rock is too small to distinguish.What’s worse, the internal heterogeneity of ore body has extensively reduced the SNR(Signal to Noise Ratio) of seismic data with sorts of undesired perturbations in the record,making the traditional noise attenuation methods are limited to improve the SNR. In the dataprocessing of metal seismic exploration, high SNR is the base of the others. So, it’s veryimportant to find a method to improve SNR in metal seismic exploration. Although there areas many noise attenuation methods as the noise in metal deposits, usually, each of them arelimited to reduce kinds of noise. Then, to suppress all of the noise, we need to combinedifferent methods.This paper has systematically introduced sources of seismic data, and illustrated howthe random noises, organized noises and scatter noises comes, and the degree they developand affect the quality of data and so on. Moreover, I also analyse the character in differentmathematical domain and the difference between effective waves to find the biggestdiversity there. Then the suitable de-noising operator is designed to distinguish the signaland noise.Combined methods, with reasonable sequence in different domains are applied tosuppress noises in various levels. F-K transformation,2D mask filtering de-noising based onRadon transformation and median filtering in Curvelet domain are highly discussed toreduce different noises. By studying the theory of various de-noising methods, we can learnseveral special de-noising methods. For example, to suppress the moderately strong linearinterferences in Level2attenuation, F-K filtering and Radon transformation are used; wealso design Radon transform and combined transformation method in Curvelet domain toreduce the strong random noise. After computing models, the combined methods in variousdomains are proved to be practicable.This method has been tested in the metallic ore deposit in Jinchang where the seismicdata contains many different strong noises, and the result is remarkable. By comparing theresult with the conventional method of noise attenuation of Mingcai Xu, comprehensive de-noise methods can improve the SNR substantially, which shows that the method wedesign is valid.The comprehensive method in various domains applied in metal seismic exploration isable to improve the SNR highly and suppress sorts of undesired perturbations in the record.In practice, we can choose different combination modes according to the character of theseismic data to reach ideal result, thus contributing to the later data processing.