Multiple elastic attribute fusion approach based on affinity propagation clustering strategy for gas hydrate reservoir identification

Pre-stack seismic inversion based on sensitive elastic parameters is critical in reservoir lithology prediction and geofluid identification. The ability of a single elastic attribute to identify a reservoir depends on its sufficient sensitivity to distinguish the target reservoir from the surrounding sediments. In general, high-dimensional data space composed of multiple elastic attributes is more conducive to describing reservoir characteristics. Therefore, a multiple elastic attribute fusion method using affinity propagation clustering strategy for gas hydrate reservoir identification is proposed. Rock-physics modeling is the most effective tool to determine the influence of microscopic physical parameters on macroscopic elastic response and to quantitatively evaluate the sensitivity of elastic attributes. Consequently, a rock-physics model of hydrate sediments considering the non-negligible shear properties of hydrates is constructed to clarify reservoir-sensitive elastic parameters. Additionally, a clustering evaluation indicator is defined to determine the optimal data clustering dimension in terms of feasibility and economy of the proposed method, and to avoid bias in the results due to data redundancy. It is shown that the 3D elastic attribute space consisting of shear modulus, Young's modulus, and S-wave velocity has the best discrimination ability for hydrate reservoirs. The logging data are used to verify the feasibility and effectiveness of the proposed method. Finally, the hydrate reservoir development is accurately discriminated by using the multiple elastic attributes yield from the pre-stack seismic inversion and combined with the fusion strategy.