Adaptive seismic noise reduction using Wiener filter

Abstract

Seismic noise disrupts the earthquake observation system due to the frequency and amplitude of seismic noise similar to the earthquake signal. The filter process is one of the methods that can be used to reduce seismic noise. In this study, the Wiener filter algorithm was designed with the Decision-Directed method for Apriori SNR estimation. This filter was chosen because it is adaptive, so it can adjust to environmental conditions without requiring manual parameter settings. The data used are earthquake signals that occur in the Palu area, Central Sulawesi, which are recorded on PKA29 temporary seismic station from February 3 to April 28, 2015. After each signal data has been filtered, then it is evaluated by calculating SNR differences before and after filtering, the signal's dominant frequency, and the cross-correlation of the signal before and after filtering. As a result, the Wiener filter is able to reduce the noise content in earthquake signals according to noisy frequencies before earthquake signals. The impact is that SNR has increased with an average of 8.056 dB. In addition, this filter is also able to maintain the shape of earthquake signals. This is indicated by the normalization value of the cross-correlation between signals before and after the filter which ranges from 0.703 to 1.00.