Array Coherence

Abstract:

To observe weak signals it is important to first do the preprocessing so that weak signals are preserved. This is best described by the coherence between stations.We demonstrate that a normalized coherence calculated from phase angles only is more robust to outliers. This processing has been popular for observing microseisms and for extracting noise cross correlation results.

We illustrate coherence-based processing for tracking storms in the Southern Ocean from the southern Californian array.The preprocessing is popular where stable traveltimes between stations are required. However, for extracting directions of propagating noise or attenuation the discarding of amplitudes can be problematic.

For large arrays (e.g. Long Beach array or USArray) there are often weak coherent arrivals within a fraction of the array. These cannot be observed with the whole array or using individual sensors. Forming first a coherence matrix, we identify the elements in the noise crosscorrelation matrix with sufficiently (?) large entries. Based on these entries further processing can identify the source location.