Elastic FWI with rock physics constraints
Qi Hu, Kristopher A. Innanen
Current efforts to use elastic full waveform inversion (EFWI) go beyond imaging of complex structures and aim at determination of reservoir-scale rock physics properties. However, the nonlinearity of EFWI and parameter crosstalk can prevent its convergence toward the actual model. Parameters such as density and fluid saturation are more difficult to retrieve because of their limited contributions to seismic data. We develop a method for EFWI that uses rock physics constraints to mitigate such limited sensitivity. These constraints are in the form of explicit velocity-density relations for different lithologic facies as a function of position, and are imposed through a model regularization term in the objective function. We implement two different workflows of constraining EFWI for elastic and rock physics properties. One is a sequential approach that consists of first inverting for velocity and density through EFWI and then transforming the elastic attributes to rock physics properties. The other is a joint approach where we parameterize EFWI with rock physics properties, allowing elastic and rock physics properties to be simultaneously updated. Constraining each workflow helps improve density and saturation recoveries. We also illustrate that the joint approach is superior to the sequential inversion in terms of computational cost and the ability to ensure consistency between elastic and rock physics properties.