Quantitative seismic pre-stack analysis of potential gas-hydrate resources in the Makran Accretionary Prism, offshore Iran

Abstract

Gas hydrates are classified as a major unconventional resource and assumed to be a future potential energy source. High pressures, low temperature stability conditions of the hydrates restrict their occurrence to the permafrost and deep sea regions (from the continental slop to the abyssal). In these regions where the well information is most often absent or sparse, the amplitude-variation-with-angle (AVA) is a common method of hydrate/free gas assessment. In this method, various linear reflection coefficient equations are fitted to a series of data points to compute some pre-stack attributes. These attributes have the advantage of being an interface property rather than a layer property which makes them an easy to use characterization tool in lack of well data. In recent years, rock physics templates (nomograms) have been presented to estimate the hydrate/free gas saturations solely based on these interface properties. AVA inversion approach was applied to the seismic data in the Makran Accretionary Prism, offshore Iran, to estimate the bottom simulating reflector (BSR)-vicinity hydrate and gas saturations. This inversion process included four pre-stack attributes of the intercept, gradient, normal incidence (NI) and Poisson reflectivity (PR) and was limited to the BSR. Estimated BSR-vicinity saturations were ranged from 7%–8.5% for hydrate and 5%–6.5% for free gas at three representative locations. Accuracy of each attribute was analyzed by a synthetic modeling. In this modeling, the effective medium theory (EMT) was considered to calculate the physical properties of hydrate and gas-bearing sediments. Lithological properties and the acquisition geometry were set to the corresponding values of the field data. Results of the synthetic AVA inversion showed that if the hydrate saturation does not exceed 30%, the error of the estimation for pre-stack attributes would be in acceptable range. The expected estimation errors for actual attributes were derived from this synthetic modeling.