aHomestake Array and Wiener Filtering Array Coherence Wiener - PowerPoint PPT Presentation

aHomestake Array and Wiener Filtering M. Coughlin Introduction aHomestake Array and Wiener Filtering Array Coherence Wiener Filtering Velocity Measurements Michael Coughlin, Jan Harms Conclusion August 11, 2016 1 / 16

Introduction aHomestake With the original Homestake array (http://arxiv.org/abs/1403.7756): Array and Wiener Filtering M. Coughlin Introduction We demonstrated that we can achieve more than an order of 1 Array magnitude seismic-noise cancellation between about 0.05-0.5 Hz Coherence using Wiener filters with only a few seismometers separated by a Wiener Filtering distance of order 500 m. Velocity Measurements At least a factor 50 NN reduction should in principle be feasible 2 Conclusion at the Homestake site around 0.1 Hz (subject to assumptions about scattering). We have showed that this subtraction performance can be 3 achieved without regularly updating the filter, indicating that the average properties of seismic fields at Homestake do not change significantly over timescales of weeks in this frequency band. 2 / 16

Introduction (continued) aHomestake Caveats to the analysis: Array and Wiener Filtering M. Coughlin Assumed that seismic scattering at the Homestake site is 1 Introduction representative for seismic scattering of the entire region that Array needs to be included for NN estimates. Coherence Array not large enough to explore optimal array design and the 2 Wiener Filtering many technical issues associated with the calculation of Wiener Velocity Measurements filters based on a large number of reference channels Conclusion Residual spectra contained a microseismic peak ... why? (body 3 waves and surface waves? scattering?) 3 / 16

Wiener Filter (iHomestake) aHomestake Array and Wiener Filtering M. Coughlin Introduction Array Coherence Wiener Filtering Velocity Measurements Conclusion (a) Wiener Filter 4 / 16

Introduction (continued) aHomestake Benefits of aHomestake array: Array and Wiener Filtering M. Coughlin The array has significantly larger horizontal spacing than used in 1 Introduction the iHomestake analysis. Array Significantly more channels! 2 Coherence Wiener Filtering We can try to use the larger aHomestake array to test these: Velocity Measurements Distinguishing between body and surface waves Conclusion 1 Whether a larger array with greater variation in station distances 2 would yield even better subtraction over a broader range of frequencies 5 / 16

Seismic Spectra aHomestake Array and Wiener Filtering 10 -6 M. Coughlin 300 800 Introduction D4850 WTP Array Hz] 10 -7 Coherence Wiener Filtering Seismic Spectrum [m/s / Velocity Measurements 10 -8 Conclusion 10 -9 10 -10 10 -2 10 -1 10 0 10 1 Frequency [Hz] (b) Seismic Spectra 6 / 16

Coherence vs. Relative Location (0.2 Hz) aHomestake Array and Wiener Filtering M. Coughlin 8000 0 Introduction 6000 -0.5 Array Coherence log10(1 - Coherence) 4000 -1 Wiener Filtering Velocity 2000 Y [m] Measurements -1.5 Conclusion 0 -2 -2000 -2.5 -4000 -6000 -3 -10000 -5000 0 5000 X [m] (c) Coherence vs. Relative Location 7 / 16

Coherence as a function of distance (0.2 Hz) aHomestake Array and Wiener Filtering M. Coughlin 1 1500 Introduction Array 1000 0.8 Coherence Coherence at 0.2 Hz Relative Elevation [m] Wiener Filtering 500 Velocity Measurements 0.6 0 Conclusion -500 0.4 -1000 -1500 0.2 0 2000 4000 6000 8000 10000 Horizontal Distance [m] (d) Coherence as a function of distance (0.2 Hz) 8 / 16

Coherence as a function of distance (1.5 Hz) aHomestake Array and Wiener Filtering M. Coughlin 1 1500 Introduction Array 1000 0.5 Coherence Coherence at 1.5 Hz Relative Elevation [m] Wiener Filtering 500 Velocity Measurements 0 0 Conclusion -500 -0.5 -1000 -1500 -1 0 2000 4000 6000 8000 10000 Horizontal Distance [m] (e) Coherence as a function of distance (1.5 Hz) 9 / 16

Parameters aHomestake Array and Wiener Target: 800:HHZ Filtering 1 M. Coughlin Used all sub-surface seismometers (8) 2 Introduction Vertical channels only 3 Array 1 Hour filter, 23 Hour subtraction 4 Coherence Wiener Filtering Velocity Measurements Conclusion 10 / 16

Low Frequency aHomestake Array and Wiener Filtering 10 -5 M. Coughlin Hz] Original Introduction Residual 10 -6 Array FF Coherence LNM Seismic Spectrum [(m/s)/ HNM Wiener Filtering 10 -7 Velocity Measurements Conclusion 10 -8 10 -9 10 -10 10 -2 10 -1 10 0 Frequency [Hz] (f) PSD 11 / 16

High frequency aHomestake Array and Wiener Filtering 10 -5 M. Coughlin Hz] Original Introduction Residual 10 -6 Array FF Coherence LNM Seismic Spectrum [(m/s)/ HNM Wiener Filtering 10 -7 Velocity Measurements Conclusion 10 -8 10 -9 10 -10 10 -2 10 -1 10 0 10 1 Frequency [Hz] (g) PSD 12 / 16

Parameters aHomestake Array and Wiener 3D KF-map decomposition Filtering 1 ( � k = ( k x , k y , k z ) → � k = ( k r , k θ , k φ ) → v = 2 πf k r .) M. Coughlin Used all seismometers Introduction 2 Array Vertical channels only 3 Coherence 1 week of data 4 Wiener Filtering Velocity Measurements Conclusion 13 / 16

Velocity vs. frequency aHomestake Array and Wiener Filtering M. Coughlin 10000 Introduction 9000 Array 8000 Coherence 7000 Wiener Filtering Velocity [m/s] 6000 Velocity Measurements 5000 Conclusion 4000 3000 2000 1000 0 10 -2 10 -1 10 0 10 1 Frequency [Hz] (h) Velocity vs. frequency 14 / 16

Angle ( tan − 1 ( y/x ) ) vs. frequency vs. time aHomestake Array and Wiener Filtering M. Coughlin 1.5 Introduction 6 Array 1 Coherence 5 Wiener Filtering 0.5 Velocity Time [Days] 4 Angle [rad] Measurements 0 Conclusion 3 -0.5 2 -1 1 -1.5 0 10 -2 10 -1 10 0 10 1 Frequency [Hz] (i) Angle ( tan − 1 ( y/x ) ) vs. frequency vs. time 15 / 16

Conclusions aHomestake Array and Wiener The aHomestake array covers a wider aperture than that of the Filtering 1 original Homestake array M. Coughlin We can use the aHomestake array to explore the effects of the Introduction 2 assumptions of the original analysis Array Coherence Numerical issues seem to be limiting the efficacy of the Wiener 3 Wiener Filtering filters Velocity Measurements Conclusion 16 / 16