Seismic finite-difference (FD) modelling is a term that describes the propagation, recording and processing of virtual seismic waves through a computational model. It is an important tool for addressing a wide variety of questions related to seismic acquisition, processing and interpretation for the simple reason that seismic images are best assessed when the subsurface is well defined. FD modelling is particularly useful when data is to be acquired over targets that present complex imaging challenges, such as those underneath salt bodies or a highly heterogeneous shallow section (e.g. Regone, 2007; Kabir et al., 2009).

This paper discusses the application of finite-difference modelling to the central part of the Azeri-Chirag-Gunashli (ACG) field offshore Azerbaijan (Figure 1) and builds on the earlier work of Crosby et al. (2014). ACG is a steeply-dipping NW-SE oriented anticline approximately 10 km wide and 50 km long. The targets reside within a series of Late Miocene stacked braided fluvio-deltaic sequences with intervening shales (e.g. Reynolds et al. 1998). Imaging is complicated by the presence of steep dips on the flanks and a highly heterogeneous shallow section including mud volcanoes, complex faults and shallow gas bodies at the crest. It is clear that any model must capture these details if it is to be useful.

Historically, the principal limitations of seismic modelling have not just been computational, but have also resulted from the level of detail in the model itself. Models have either been constructed manually using a set of interpreted framework surfaces or derive their properties from existing seismic images (e.g. Regone, 2007; Kabir et al,. 2009). In the first case, reflectivity is often too simplistic; whilst in the second case artefacts and bandwidth limitations are introduced as truth into the model. A better approach is to simulate geological features directly when building the model. Two recent papers (Etgen and L'Heureux, 2013; L'Heureux and Etgen, 2013) present an automatic, probabilistic method for generating stochastic stratigraphy using a set of facies templates, and then morphing the resultant layering to structure. This approach can generate geologically realistic models to arbitrary levels of detail. Here we show the result of applying this method to Central Azeri, together with some preliminary synthetic seismic images.

You can access this article if you purchase or spend a download.