Inversions - RokDoc-Chronoseis
Having fully understood the information from forward modelling and achieving a good well tie, the next stage of a predictive workflow is to undertake an inversion to represent the rock’s elastic properties. Ikon Services can perform a number of different inversions as described below and will often evaluate the strengths and weaknesses of the different types of inversion as part of the feasibility study.
Coloured inversion – relative impedance inversion – All round inversion but does not remove the wavelet. This can follow on into seismic net-pay analysis (for thin reservoirs - Connolly) and generation net pay probability maps, but it does require good well control. Inversion products can be relative or absolute.
Extended elastic impedance – generate optimised fluid, lithology/property stacks based on EEI analysis of offset well data. Inversion products can be relative or absolute.
Relative Fluid and lithology indicators (Fluid Factor) - This uses a combination of weighted stacks and coloured inversion.
Spectral inversion – Generates relative impedance which is broader bandwidth than the input seismic, and can be run without well data. It does this through spectral decomposition of the seismic data – this is a proprietary inversion technique and is as yet unavailable outside of Ikon. Inversion products can be relative or absolute.
Deterministic simultaneous inversion (can also be done as a relative inversion) – Use multiple stacks or angle gathers to invert to Zp, Zs and Rhob (the latter dependant on far angle range). These can be combined to generate poisson ratio, lambda-rho, mu-rho etc.
AI, EI, EEI, Rp, Rs etc – Deterministic inversion (pre and post stack) using sparse spike / model based inversion techniques to provide absolute impedance cubes based on optimal discriminators. These cubes can be combined to create poisson ratio, lambda-rho etc.
Neural network inversion – transform reflectivities directly into impedances.
Neural network rock property inversion – transform impedance or reflectivity attributes into meaningful rock properties. This can be taken as far as providing porosity, saturation and lithology cubes for use in reservoir models/simulators, but it requires good quality data.
Stochastic inversion – probabilistic inversion of seismic to high frequency impedance properties. These may then be transformed to reservoir properties. We provide a host of different ways of analysing the output of n realisations of said properties. Performed at the reservoir scale - the benefits here are that it provides a high resolution estimate of uncertainty from the seismic - watch video.
Joint stochastic inversion – a new and proprietary method of inverting multiple seismic inputs at the same time, to provide multiple high bandwidth impedance attributes. The term ‘joint’ refers to the fact that the initial impedance properties are ‘linked’ through the medium of a rock physics model, meaning that the output of the inversion is more meaningful.
ImageGenetics – Unique proprietary pattern recognition tool for use in mining 3D datasets for AVO attributes. Combine up to four seismic datasets and develop pattern templates for each based on various seismic signatures at key well or prospect locations. The workflow generates geobodies and surfaces which can then be used to aid the prospect identification, risking and ranking process.
Similarity – generation of textural attributes such as similarity (coherency) tuned to the dataset/interval in question. Enhances imaging of subtle stratigraphic features and fault and fracture geometries.
Understanding driven by Rock Physics models
Once the data is converted from seismic reflectivities to impedances Ikon is able to use the learning from the rock physics and forward modeling, to predict lithology and fluids via the following practices:
- Direct Bayesian to Rock Properties
- Impedance to reservoir properties from rock physics model
- Joint probability of values > or < cut-offs
- Joint probability of values in ranges
- NTG at probability level
- Probability of net thickness > value
- Probability of values inside a polygon
- Connected volume CDF (by any criterion)
1D to 3D reservoir evaluation & characterisation