Research On Improved Downhole Microseismic Noise Suppression Algorithm Based On Shearlet Transform

As the growing demand for oil and gas in the world,the exploitation of unconventional reservoirs has become a hot spot.At present,microseismic monitoring plays an important role in reservoir monitoring as well as resource characterization studies.The microseismic acquisition process has two types: surface and downhole monitoring.Downhole events are character for its weak energy,low signal to noise ratio(SNR),and high frequency.Therefore,improving the SNR is a necessary step in microseismic data processing.At the same time,as the conventional filtering method is limited when dealing with such signals,it is very important to design the noise suppression algorithm for downhole microseismic signal.Shearlet transform is a new multiscale geometric analysis technology developed in recent years.It is also a near optimal sparse representation for multidimensional functions.In addition,the Shearlet transform can be associated with the multiresolution analysis and establish the unity of the continuous and discrete forms.Based on its excellent properties,it is used to suppress the noise of downhole microseismic data in this paper.However,due to the high frequency and weak energy of data,it is difficult to separate the coefficients of signal and noise in the Shearlet domain through a simple threshold function.Therefore,with the help of the properties of downhole microseismic signal and Shearlet transform,this paper proposes two algorithms: block matching Shearlet transform(BMST)and scale classification Shearlet transform(SCST).In view of the two characteristics of the spatio-temporal similarity and spatio-temporal orientation of the microseismic signals,the BMST is proposed to suppress the noise of the microseismic signals in the well.Note that the Shearlets are directional filters,so the values of coefficients are relevant to the matching degree in directions between signal and shearlets.If the orientation of signal is roughly in line with that of shearlet,the values will be large.Otherwise,they will be small.By means of block matching,the numerical differences of coefficients between valid signals and noise become greater,so they can be separated easily in shearlet domain.Shearlet transform is a multiscale transform.Therefore,the distribution of signals at various scales should be also different.It is unreasonable to use the same threshold for all scales.In this paper,we propose a novel microseismic noise attenuation algorithm based on scale classification Shearlet transform(SCST).By analyzing the spectrum and energy distribution of the Shearlet coefficients of microseismic data,we divide the scales into two types: low-frequency scales which contain less useful signal and high-frequency scales which contain more useful signal.After classification,we perform two different methods to deal with the coefficients on different scales.For the Low-frequency scales,the noise is attenuated by thresholding method.As for the high-frequency scales,we propose to use a generalized gauss distribution(GGD)model based non-local means(NLM)filter,which takes advantage of the temporal and spatial similarity of microseismic data.Based on the study of downhole microseismic signal and Shearlet transform,this paper constructs two improved noise suppression schemes.Meanwhile,BMST has a great increase in SNR,which is suitable for higher requirements of SNR of data;SCST is less complex,and is suitable for real-time noise suppression.