Understanding seismic resolution is critical when trying to build accurate reservoir models by integration of data coming from different disciplines. Lack of low frequencies in seismic data makes it band limited that limits its uses only for structural model but its availability and areal coverage suggests that it has valuable information to constrain the reservoir model. Increasing resolution to double through deterministic inversion can be achieved whereas additional constraints are needed to further increase this resolution. Thus, it is paramount to improve the resolution of seismic and detection of reservoir and be able to identify reservoir thickness and extension as accurately as possible in order to build reliable reservoir model. In this study, the uncertainty in seismic resolution versus detection and quantification of rock properties are investigated in terms of seismic inversion.
The framework of improving seismic resolution presented in this study is based on integrating seismic and well data through seismic inversion. It considers reprocessing of seismic data for frequency enhancement, pre-conditioning, rock physics modelling, high quality well-to-seismic tie and wavelet extraction, low frequency modelling, stratigraphic grid definition, property proportion distribution, geostatistical variograms and lithology trends. A detailed method description is provided that has proven to have high stability. By using a series of simple examples it is demonstrated that understanding seismic resolution is challenging and significant efforts are needed to obtain a realistic reservoir image.
Under careful control of the seismic wavelet characteristics including phase, frequency and amplitude; and provided the noise level of the seismic data is sufficiently low; useful information can be extracted from high resolution seismic inversion derivatives. It can serve the purpose of detecting beds much thinner than the conventionally accepted detection limit. The results reveal that high resolution seismic is not necessarily the data with high detectability.
The novelty of this investigation is in the ability to improve seismic resolution and detection through robust deterministic and stochastic inversion to solve the most important problem on the ‘level of detail that one can see in seismic’. The observations made on real data leading to the idea of explaining seismic resolution and detection myth versus reality.