Effect of noise on microseismic event detection and imaging using ICOVA statistical noise modelling method

Despite the evidence that noise does not conform to the White Gaussian Noise (WGN) assumption, the robustness
of new processing and imaging algorithms are still tested with WGN. This paper presents an alternative noise modelling method, based on multivariate statistics, to generate realistic noise for incorporation in synthetic datasets. The realistic noise model captures the complex nature of noise arising from multiple sources and the varying signal-to-noise (SNR) observed at the different stations across the array. This complex noise structure results in microseismic events being detected at lower SNR than would be implied using a WGN model. It also successfully re-creates smearing of energy during imaging of microseismic events at low SNRs. This modelling method provides an opportunity to test the robustness of new algorithms under realistic noise conditions prior to recording data in the field.