Recognizing Reflected Tremor Phases: Guidance from Synthetic Seismograms

Amanda KlausESS 522 – Geophysical Data Analysis

Final ProjectJune 10, 2010

Outline

1. Background: Tremor– What it is– Where it happens

2. My tremor-related research3. Why care about tremor?4. My ESS 522 project– Synthetic seismograms– Data processing and seismogram analysis

Earthquake: July 15, 2004, 12:06, M 5.9 (Vancouver Island) - Station: YOUB

Tremor: July 25, 2004Station YOUB

BackgroundTremor: an emergent, low-amplitude seismic

signal, persistent in time and space

Where Tremor Happens: Map View

5 years of tremor location data(courtesy of Aaron Wech)

Cascadia subduction zone schematic(Source: pnsn.org, after Hyndman and

Wang 1995)

Where Tremor Happens: Depth

• Due to large uncertainties in vertical locations, tremor depth is unknown! Two possibilities:

Option 1: Tremor is distributed throughout the lower crust, and maybe the upper subducting slab. Tremor is likely caused by movement of fluids.

Option 2: Tremor is located at the interface between the continental and oceanic plates. Tremor is a manifestation of plates sliding past each other.

Tremor Depth: Why It Matters

1. If tremor is at the plate interface, tremor means plates are sliding

2. This sliding downdip of the locked zone increases stress in the locked zone

3. Locked zone ruptures in megathrust earthquakes after enough stress builds up

My research

• Look for signal of tremor bouncing off the Moho (seismic velocity discontinuity)

• Hope to see consistent delay times that indicate bouncing off of an interface ca. 7 km below

sourceMoho

ESS 522 Project

Part 1: Generation of synthetic seismograms

• Goal: investigate feasibility of seeing bounces of low-amplitude seismic energy off of the Moho

Synthetics - Method

• Did a literature search for best estimates of seismic velocities and densities

• Assembled these into a simple 1D model of flat, horizontally infinite layers, over a half-space representing the upper mantle

• Used Lu-pei Zhu’s FK code for generating synthetics– Earthquake source parameters: 7° dip, reverse

fault, magnitude 2

Input Model• Simple 1D velocity model (no dipping layers)• Source: M 2 earthquake – dip 7° east - thrust fault.

Source is at 29.99 km depth, just above interface.

Project Part 2: Data Processing

Many things to consider:• Filtering: what frequencies are we concerned

with?• Instrument response and how it affects data– Not included today

• Signal-to-noise ratio– Tremor data: very low signal-to-noise ratio

We’ll look at some of these things in the demo…

Demonstration

• Step 1: load data • Step 2: plot data: plotFKjdf( eastVert, 0 )

Demonstration

• Use reduced time to shift seismograms:plotFKjdf(eastVert, 6)

Demonstration: Adding Noise

Data_out = addrand(data_in, noise_amp)Noise_amp is % of largest data point – default is

10%

With noise added…

Filtering & Effect on Noise

Use command coralFilter coralFilter(data, cutoff_freq, type, order, phase)

Filtering Example

4th Order Butterworth Filter 16th Order Butterworth Filter

An illustration of stacking

• 10 copies of the same seismogram, with different random noise added to each

Vertical channel – station east of source, 40 km away

Noisy data, stacked

Analogous for stacking arrays

5 miles