Seismic inversion has been a method for geologists and geophysicists in reservoir characterization for the last few decades. The methodology takes all the information available about a reservoir such as well information, field analogues, trends and rock physics to piece together a reservoir model to predict and better understand what the reservoir may look like in unpenetrated locations.
However, seismic inversion, as with any other attempt to characterize the subsurface, involves a significant degree of uncertainty, making it a difficult process to master. To reduce this uncertainty as much as possible, the following five factors should be considered when performing any seismic inversion.
1. Define What You Are Inverting For
Seismic inversion can be performed with one of several goals in mind. The first thing you should consider when doing a seismic inversion is why you are doing this particular inversion. Are you looking for fluids? Do you aim to understand facies distribution? Is your goal to understand the reservoir architecture?
Clearly defining the goal of the inversion is essential because the input needed to perform the inversion may not be the same in each case. Take One Dimensional Stochastic Inversion (ODiSI), a novel approach to seismic inversion, as an example. ODiSI is a facies-based inversion, which means that all facies have to have quantifiable elastic properties and enough acoustic separation to identify anything meaningful in the data.
2. QC the Data at Every Step of the Inversion Process
Seismic inversion is driven by a convolution between real word; direct and indirect, measurements and a model of the earth. Therefore, at every step of any seismic inversion one must consider the possible and probable outcomes. Do the results look real? Could they be real? What has the result shown me?
A simple form of seismic inversion is that of colored or relative inversion, whereby seismic reflectivity data is converted to that of impedance; moving to a rock property (impedance) from an interface property (reflectivity). Colored Inversion is achieved through convolution of an earth model which is created from well information and measured seismic data. The best method for judging the success of such a process is through visual QC and requires a degree of scientific opinion.
3. Be Pragmatic
Seismic inversion is a theory-laden field and a rules-based approach for understanding the subsurface. To perform seismic inversion, you need a firm grasp on these theories and rules. However, practical seismic inversion often requires you to deviate from established theories as certain errors cannot be quantified. Blindly applying theories and rules without controlling the quality of each step in the process will often lead to unsatisfactory results.
When performing a seismic inversion, it is often valuable to think outside of the box and try different approaches that may require a departure from established theories. Know the theory, but do not be afraid to test something different – as long as it makes sense.
4. Remember That the Output Reflects the Input
Seismic inversion relies on prior information as input. Prior information is based on what you already know, but when performing seismic inversion, you also need to be conscious of what you do not know. After all, your well data is biased – nobody sets out to drill a well with no reservoir. Your input will influence the output.
When parameterizing your model, avoid over-constraining it. When working with ODiSI, for instance, you should provide ODiSI with any possible scenarios, and ODiSI will pick the best answer. The danger of over-constraining, by not allowing the creation of different scenarios, is to obtain meaningless results.
Instead, consider what is possible, not only what is probable, and let this influence your input and how you constrain your model.
5. Consider the Resolution of Your Seismic Data
The result of a seismic inversion will reflect the input you have used to perform the inversion. As the main input for seismic inversion is seismic data, your results will have the same resolution. The resolution of your seismic constrains all seismic inversion. Your models may be at a fine scale, but the seismic loop is ultimately what provides all the information and is what you are inverting.
When using ODiSI, for instance, the resolution of the pseudo-wells it produces will be very fine. However, the main input is seismic data, so the output will naturally have the same resolution as your seismic.