things happen before earthquakes

I read an interesting paper from J. Byerlee on "Model for episodic flow of high-pressure water in fault zones before earthquakes".

This is a part of a dialog between authors on the role of high-pressure water in the fault zones: where the water come from and how it effects fault zones. The flow of high-pressure water differs in their temporal behaviors: no flow, continuos flow and episodic flow. Although this paper is written in 1993, things haven't been clearer since then. Perhaps the discovery of tremor and related phenomena make things more difficult to explain.

Here is today's theme: Things happened before earthquakes. It is an interesting theory. The following story is my own interpretation of the paper and one should trace back to the original paper if interested.

Step 1: Water saturated the fault zone. Water can penetrate into crust down to depth of 20 km. The crust is largely in hydrostatic pressure. The fault zone which consists of fine crushed rocks is highly porous and permeable.

Step 2: Compaction of the fault zone at high pressure and temperature. Silica deposits during the compaction and over-saturated water flow out of the fault zone into country rocks. Seals formed after silica deposition.

Step 3: Seals formed compartments. The seal is impermeable. The 3D compartments developed from deep section of the fault zone and propagated upward. The depth distribution of compartments is from 15 km to 3 km depth.

Step 4: Trigger earthquakes. Within the country rock, in order to prevent hydrofracture, the minimum principle stress has to be greater than the pore pressure. But inside the fault zone, the formation of seal allows a smaller minimum principle stress. The pore pressure can be 85% of lithostatic and the minimum principle stress can be only 60%. Before the formation of seal, however, the minimum principle stress has to increase. This is a key point that has been stated a couple of times in the paper, without any explanations. The seal can still break when further pressed, then the water will flow into country rock from the sealed pores, the effective normal stress would increase and the fault will be further locked. When the shear stress is big enough the lithified seal between compartments will be fail, and there will be a increase in the pore pressure and the fault unclamped. Earthquake (or tremor?) will happen as a result of pore pressure change in the fault zone.

Step 5: The earthquake fractures the fault zone, and everything restart.

To me, this is quite an imagination.

craton formation

Today, I listened to Dan McKenzie's talk. It's an interesting talk in my opinion though I don't completely understand it.
I learned something:
when considering the heat flow the effect of radiogenic heating in the thickened crust is not negligible.

measuring interseismic deformation using InSAR 1

Long time no see. I just passed the qualify exam and I am on my second half journey to a PhD.

Today I am going to discuss existing technique and approaches to measure interseismic deformation using InSAR. This is an interesting topic. Currently several groups of people are working on InSAR time-series techniques. In US these groups includes Stanford, Caltech, JPL, Scripps, Berkeley, UCR, etc etc...

Let's start from Persistent Scatterers InSAR (PS). This technique is first developed by [Ferretti et al]. It works like this: align a stack of the SLC images; scale the amplitude image; compute the pixel-wise mean amplitude (m) and standard deviation of the amplitude (σ) for a whole stack; compute the D = σ/m; treat the pixel where D is smaller than a threshold as Permanent Scatters; model the phase of each PS as a mix of linear deformation and DEM error and solve a least-square problem to retrieve both the DEM error and a linear deformation rate.

Instead of using a threshold of amplitude dispersion, Lyons and Sandwell compute s = m/σ and weight the real and imaginary part of the complex image by s^2; apply a non-isotropic filter and form interferogram.

Instead of amplitude dispersion, stanford group [Hooper et al] include the phase information to identify PS. Their method is written in JGR paper. I haven't understood this method yet.

Cracks on San Andreas Fault after Baja California earthquake

Last weekend, Matt and I drove to Imperial Valley and north shore of Salton Sea to look for cracks after the M7.2 Sierra El Mayor event.
We found evidence to show the earthquake triggered slip on San Andreas Fault system in Southern California.

David told us how to use B4 data to find the fault and it works like a charm. We load the marker into Google Earth after studying the InSAR and B4 data. The accuracy has to be within meters in order to locate the cracks on the ground. It's a surprise when we found them because the cracks are even hardly visible in the real word.

It is an exciting experience to find sub-cm cracks by the means of laser altimetry and InSAR. However, I don't know how much science can come out of it, as it's really tricky to measure the cracks quantitatively.

That's my two cents.

AGU 2009

Q: What I learned from AGU on Monday?

A: There are two main continental transform fault system: one is San Andreas Fault in California and the other is Alpine fault in New Zealand.

People studied the tectonics processing, but what they're really looking at is the slip rate and strain rate in both geological and geodetic time window. It's mainly focus on the kinematics of the ground motion.

The formation of the Garlock fault is not known.

The effect of the damage zone on the fault property is not known.

The viscosity, plate effective thickness, rigidity is related to the fault deformation.

The fault geometry, topography, crust thickness and heat flow is used to model long-term slip rates.

Geological slip rates, stress direction, GPS velocity and depth of the seismicity are important properties.

Q: What I learned in AGU on Tuesday?

A: For Wenchuan earthquake, the postseismic uplift in mountain side is stronger than the postseismic subsidence in basin side. It is a consistent feature and need explanation. The decay period is 8-14 days in the logarithm decay curve from GPS measurement.

Ionosphere effect is not present if one changed SAR acquisition to another date.

There are two mechanisms for the ductile behavior of the lithosphere: viscous granulus flow and dislocation creep.

Q: What I learned from AGU on Wednesday?

A: Static displacement (near field) decay with the square of the distance. Dynamic displacement (far field) decay with the distance. The r^-1 and r^-2 curve could be useful.

The inelastic and plastic failure is important when the strain is high (> 10^-3). However, it seems such a high strain is only present within a few meters in the fault zone.

Meade and Loveless wrote an interesting paper on crustal motion modeling.

I got too tired to learn in the last two days.

That's for the end of the year.