This work provides an example of a technique to estimate reservoir properties from inverted seismicimpedances. High-quality seismic data has made inversion of that data for elastic properties reliable.Petrophysical and elastic properties from well-log data, along with a rock-physics model, explain therelationships among the elastic properties and the reservoir properties of interest. However, ambiguity andnon-uniqueness are present in those relationships. The inversion of seismic data for P-impedance (IP) andS-impedance (IS) requires pre-stack seismic data, and a particular algorithm to combine low-frequencyinformation. The inversion provides IP and IS for every time sample at each CDP. A calibratedrock-physics model translates seismic-scale impedances to reservoir properties. The calibration of themodel typically is done using well-log curve information around the interval of interest. This paperdemonstrates the seismic inversion routine and the subsequent mapping of the inverted impedances torock properties using data from the Marco Polo field. The rock-physics model chosen was the soft-sandmodel because of the geological trends identified from well data and the interpretation of the depositionalenvironment. In addition, the well-log data indicated the presence of five facies, including a gas-sand,oil-sand, two brine-sands, and shale facies. A Bayesian classification technique mapped the seismicimpedances to the most likely facies. A statistical technique was necessary to account for the non-uniquerelationships among the elastic and reservoir properties. The results are realizations of the most likelyfacies. Probabilistic estimates of porosity and saturation for the hydrocarbon-bearing facies came fromjoint conditional distributions of IP and the ratio of P- to S-velocity (VP/VS). Maps of the probabilitiescontain the associated uncertainty in each results. Limitations to this technique are three-fold. First is thatthe relationship is non-unique between impedances and the rock properties, whereby one value ofimpedance relates to different combinations of rock properties. Second, the resolution of the seismic datais the resolution of the rock properties. Third, the rock-property estimates have errors in them due to errorsin the rock physics model, errors in the inverted data, and errors in the match between the data and themodel.
展开▼
