GNGTS 2014 - Atti del 33° Convegno Nazionale

GNGTS 2014 S essione 3.1 59 The convolution with missing surface data is replaced by a simple time-shift by the water layer primary reflection traveltime τ sn . The adaptive subtraction step is still required to separate the interfering wavefields, however no explicit source wavelet deconvolution is needed once again. Imaging of multiple energy. Until few years ago, multiple energy has been considered as noise, thus multiple reflection wavefields have been discarded after the separation process. However, multiple reflections contain a wealth of information of the subsurface, that can be used in seismic processing to improve the resolution of reservoir images (Berkhout, 2006). Indeed, recently the geophysicists’ mind-set about multiples has changed: an increasing interest actually exists in finding new methods for the exploitation of such an information. Wavefield extrapolation imaging techniques can be adapted to correctly handle both primary and multiple seismic reflections, and recent results proved how non-linear imaging can take advantage of internal multiples, too (when a detailed velocity model that matches the sharp discontinuities at multiple generators is available). In O’Brien (2013) the imaging of “half order” multiple arrivals (i.e., source-injected energy, after being reflected in the subsurface, is then downward reflected by sea surface, and reaches the receiver array in the borehole as a transmitted wave) proves how the use of “multiple noise” can improve subsurface illumination for WVSP survey, solving at the same time the lack of lateral and shallow illumination. Jiang et al. , (2007) discuss different approaches to image multiples, with different complexity and sensitivity to velocity model knowledge: among them, a clever “receiver-mirror imaging” technique (i.e., a new acquisition geometry is built by reverting receiver depths z r ’=-z r , and imaging is performed with a double-vertical-size velocity model mirrored in respect of the free-surface, as shown in Fig. 2a) allows to migrate such data without the need to modify standard imaging engines, but only applying simple modification to acquisition geometry. “Half-order” (i.e. mirrored receiver) multiple arrivals and primary reflections do not interfere, because of the difference in ray-path lengths. However, extended recording time must be taken into account when acquiring data. When more than one reflection is considered in the subsurface, such techniques may fail because of the inability to separate the interaction between primary and multiples from different interfaces or from different multiple orders (i.e., number of reflections in the subsurface) (Fortini et al. , 2013). In fact, such a cross-talk can be hard to be attenuated after the imaging step, especially for WVSP acquisitions because of their low fold of coverage. Thus, modified imaging algorithms must be implemented to (at least partially) overcome these problems (Zhang et al. , 2014). Fig. 1 – a) double mirror migration flowchart. Top-row: a two-step iteration of SRME separate first order only water- layer multiple reflections; bottom-left: data after multiples removal are imaged by receiver mirror migration; bottom- right: first order water-layer multiples are imaged after simple source geometry transform. b) model-based water-layer SRME approach: dotted blue lines in the water-layer represent the synthetic ray-paths s-n while green ray-paths n-r are extracted from WVSP data.