Research on multidomain joint denoising method based on DAS data in open-pit coal mine

During the acquisition of field seismic data, complex environmental noise can easily interfere, especially in mining areas where the interference from background noise is more severe. Additionally, there is instrumental noise caused by mechanical vibrations in the optoelectronic system of Distributed Acoustic Sensing (DAS) systems. These noises can significantly reduce the signal-to-noise ratio of DAS data, increase the difficulty of subsequent data processing, and decrease the accuracy of geological structure interpretation. Therefore, noise suppression has become one of the key steps in DAS data processing. This paper proposes an adaptive denoising method for DAS data based on multidomain joint processing. Median filtering is used to remove instrumental common-mode noise in the spatial domain, while curvelet transform is used to remove random noise by exploiting the differences in coherence between signals and noise in the sparse domain. This method effectively characterizes multi-scale and multidirectional signal features through curvelet transform, combines dynamic threshold functions to achieve noise suppression and signal reconstruction, and prioritizes the suppression of instrumental noise's in-phase characteristics while retaining signal details and reconstruction accuracy. Experimental results show that this method can significantly improve the signal-to-noise ratio of DAS data, effectively preserve the geometric details of valid signals, and enhance the reliability and precision of reconstructed data. The proposed denoising method demonstrates strong robustness in complex noise environments and provides strong support for improving the application of DAS technology in seismic exploration and geological monitoring.