Nonlinear seismic inversion of total organic carbon for hydrocarbon source rocks

ABSTRACT The seismic prediction of hydrocarbon source rocks critically relies on the total organic carbon (TOC) estimation from seismic data, which represents a critical yet challenging task for potential resource assessment. Common TOC inversion methods rely on statistical correlations between TOC and seismic parameters, which lack a direct physical connection to seismic data. To address this gap, a seismic reflection coefficient equation incorporating TOC is derived by integrating a published rock-physics model of hydrocarbon source rocks with the Gray approximation of the Zoeppritz equation. The rock-physics model quantifies the relations between elastic properties and the TOC of hydrocarbon source rocks at varying maturity stages. The Gray approximation links the P-wave reflection coefficient to the reflectivity of rock elastic properties, assuming small property differences at the interface. To ensure the accuracy of the seismic reflection coefficient equation incorporating TOC and to guarantee the reliability of TOC inversion results, the derivation circumvents approximations to the rock-physics model. Theoretical model tests and comparisons confirm that the seismic reflection coefficient equation, incorporating TOC, effectively portrays the quantitative relationship between seismic data and TOC. Building upon the seismic reflection coefficient equation incorporating TOC, a nonlinear seismic inversion of TOC is developed. Within a Bayesian framework, assuming the prior probability density function (PDF) conforms to the mixed Gaussian model and the likelihood function conforms to the Gaussian model, the posterior PDF of the TOC is deduced. Given the nonlinearity of the posterior PDF of the TOC and the independence of each posterior probability density component, a differential evolution Markov chain Monte Carlo algorithm is used to simulate the different posterior probability density components of the TOC. Synthetic data tests and field data applications validate the efficacy and stability of our nonlinear seismic inversion method in TOC prediction. The resultant TOC inversion profiles finely delineate the spatial distribution of hydrocarbon source rocks.