Summary
In this paper we propose a strategy to estimate the impulse responses from a local target in the subsurface from surface seismic data, as an iterative sparse inversion approach in two steps. The first step is the process to estimate the up- and downgoing wavefields at a specific level nearby the target through Joint Migration Inversion. The second step is an iterative sparse inversion approach, which estimates the impulse responses from the target. The main feature of this strategy is that all multiple scattering in the data is used to enhance the illumination at target level. Currently, the first step has not yet been fully tested and the results shown are obtained only from the second step, using forward modeling and wavefield decomposition to get the up- and downgoing wavefields at the level nearby the target. The numerical tests show that the iterative sparse inversion approach does not require dense sources sampling to estimate the impulse responses from a target below a complex overburden, because of all the extra illumination via multiples.
Introduction
When imaging a target below a complex overburden, the complex propagation and multiple scattering effects complicate the interpretation and characterization of the target (Thorbecke et al., 2004). A good strategy is to first deal with the overburden effects, followed by migration or inversion just in the desired target (local-schemes) such as JMI-res (Berkhout, 2013) or target-oriented full waveform inversion (Staal et al., 2010; Gisolf and van den Berg, 2010; Haffinger et al., 2012). In exploration seismology these local-schemes are important tools for characterizing or monitoring the reservoir.
One of the main challenges for local-schemes is to get the dataset only containing information from the target area in the subsurface below a complex overburden. This process to derive the local response from the target area, where sources and receivers are projected on to a level just above this area, is usually called redatuming.
There are two main methodologies for this redatuming process. The first one is model-driven datuming, where the traveltimes from sources (or receivers) to the chosen datum level in the subsurface are computed from a prior velocity model and used as time shifts in a Kirchhoff-type integral (Berryhil, 1979, 1984, 1986; Schneider, Jr. et al., 1995; Liu and Xu, 2011). These methods have the drawbacks that they can exhibit some extrapolation artifacts from upward or downward continuation of the wavefields and they are velocity dependent, while estimating a correct velocity model in a complex geological setting is difficult. Another drawback is that they usually are not suitable to correct for complex propagation paths including multiple scattering and transmission effects.
