A-team Seminar - Fall, 2012

Wednesday, 3:00-4:00 PM, Green Center, Room 281.

(Seminar for Fall, 2012)

Date

Speaker

Subject

Notes

09/19/12

Bharath Shekar

Presentation about summer work.

09/26/12

Francesco Perrone

Adjoint State Method

In order to solve inverse problem in the framework of optimization theory we need the gradient of the objective function that defines our problem. In exploration seismology we want to estimate the physical parameters of an elastic medium using boundary value data (seismic waves recorded at the surface) and in order to do so we construct an inverse problem that links the wavefield in the medium with the model parameters. The adjoint-state method is a standard numerical technique that allows us to compute the derivative of an objective function that depends on a set of variables (state variables) that describe the state of the dynamical system under investigation (e.g. the elastic Earth's subsurface), and it does that by solving a limited number of forward problems. In this presentation I describe the adjoint-state method through a few examples, I use different flavors of Full Waveform Inversion to show the effects on inversion resolution, and I show how to compute the key element of the method, i.e. the sources of the adjoint problem.

10/03/12

Oscar Jarillo.

Full Waveform Inversion of microseismic data in VTI media

Monitoring of microseismic events triggered during the hydraulic stimulation of reservoirs is of great interest to obtain information about the fracturing process. In this talk I will present the outline of my project that involves full-waveform estimation of source location, source mechanism and VTI parameters from synthetic microseismic data implementing the adjoint-state method. I will also present the results from forward modeling using a moment-tensor source and some basic aspects of moment-tensor inversion from borehole data.

10/10/12

No seminar

10/17/12

Round-table discussion

10/24/12

Bharath Shekar

Point-source radiation in attenuative anisotropic media

Perturbation ray-theory allows us to compute asymptotic Green's function in attenuative, anisotropic media. As a "by-product" of this method, we obtain approximations for the complex-valued slowness vector. The complex-valued slowness vector (bivector) describes the orientation of the planes of constant phase and constant amplitude. I will show some numerical examples which gives us a qualitative description of the behaviour of the complex-valued slowness vector. I will then describe some analytical methods to evaluate the slowness vector.

10/31/12

No seminar.

11/7/12

SEG

11/14/12

Round-table discussion about anisotropy-related talks at the SEG.

11/21/12

Thanksgiving break

11/28/12

Pengfei Cai

Joint migration velocity analysis of PP- and PS-waves for Volve OBC data

I will first briefly review the methodology of joint MVA of PP- and PS-waves, which consists of applying PP+PS=SS method, flattening PP and SS CIGs and codepthing. Then some synthetic models will be shown to test joint MVA. At last, I will show some results from a 2D line of the Volve 3D OBC dataset.

12/05/12

Bharath Shekar

Point-source radiation in attenuative media.

We will discuss/prove the following: 1. In visco-acoustic media, waves from a point-source propagate with a zero inhomogeneity angle (I have seen this stated as a "fact" in many papers, but I have not seen a proof.) 2. Saddle-point condition for evaluating the wavefield from point sources in viscoelastic media. Although this has been investigated in a few papers, the saddle-point condition was derived from a geometric argument. Therein, the interpretation of the saddle-point condition was erroneous. We will provide an alternative derivation and carefully interpret the saddle-point condition. 3. The behaviour of the inhomogeneity angle vis-à-vis velocity and attenuation anisotropy.