Abstract

Summary Seismic velocity models derived via ray-based tomography and full waveform inversion (FWI) are typically non-unique, as the observed data may be explained by many possible models. Constraints may limit these possibilities, but even then many viable models can appear equally correct. Here we present a methodology for quantifying model uncertainty and use the NOAKA ocean bottom node dataset from the Norwegian North Sea to demonstrate our approach. Recognizing and quantifying uncertainties is pivotal for evaluating risks in interpreting seismic volumes which can then be used in the well planning process. We propose a model building workflow that collates statistics of the derived models and focuses on the key reflections of interest. The model statistics are used for quantifying how model uncertainty impacts the image used by interpreters in the decision making process. Our approach empowers decision-makers with a low-cost metric set to comprehend and mitigate uncertainties, offering a clearer perspective on the potential risks associated with final seismic interpretations.