Making Marchenko imaging work with field data and the bumpy road to 3D Alex Jia*, Antoine Guitton, and Roel Snieder Center for Wave Phenomena, Colorado School of Mines
RTM Marchenko Imaging pros whole model easy implementation target orientated remove artifacts by internal multiples cons artifacts by internal multiples require wavelet learning curve (e.g. acquisition geometry)
outline Complications: Marchenko imaging: how and why? missing near offsets data calibration Why is 3D Marchenko necessary and challenging?
surface response to virtual source Marchenko equations background velocity model surface data surface response to virtual source (reverse VSP data) first arrival
up-going and down-going Green’s function Marchenko equations background velocity model surface data up-going and down-going Green’s function first arrival
up-going and down-going Green’s function Marchenko equations background velocity model surface data up-going and down-going Green’s function Marchenko image first arrival
1 G+ 2 3
a b G-
missing near offsets: 7 traces, 214 m
retrieved Green’s function full offset missing near offsets difference
missing near offset stronger for shallow events and near offset events maximum missing near offset D: depth of reflector T: dominant wave period assumption: hyperbolic assumption for flat multi-layered earth (details in CWP report)
full offset missing near offsets RTM
data calibration input 1: surface reflection response input 2 : first arrival amplitude should match
retrieved Green’s function scale=0.5 scale=1 scale=1.5
data calibration scale=0.5 scale=1 scale=1.5
* synthetic first arrival field 0.001 0.001 The question you may ask is how we calibrate the field data? Because we have a background velocity model and we use this model to construct our first arrival. We can also use this model to simulate surface data with amplitudes that can match the first arrival. Thus, we now have the simulated surface data that has the required amplitude, and we have the field data we need to calibrated. Hence, we can obtain a scaling factor by comparing the amplitude the field data and modelled data, and use this scaling factor to calibrate the field data.
0.001 synthetic first arrival 0.001 * field (scaled)
2D can be deceptive
Green’s function in 3D structure x=0.5, y=1, z=0.5 x=0.5, y=1, z=0 Goal: retrieve the Green’s function from a VS at (x=0.5, y=1, z=0.5) to a surface receiver at (x=0.5, y=1,z=0)
retrieved x=0.5, y=1, z=0.5 x=0.5, y=1, z=0
retrieved x=0.5, y=1, z=0.5 x=0.5, y=1, z=0
retrieved modelled x=0.5, y=1, z=0.5 x=0.5, y=1, z=0
(dense in-line, sparse x-line) 3D Marchenko ocean bottom data streamer data (dense in-line, sparse x-line) (http://www.prism-exploration.com/seismic-ops) (Wang et al. 2009)