An evaluation method for sweet spot prediction and classification in deep low-permeability tight gas reservoirs: a case study of Pinghu Formation in gasfield X of southern central anticline belt, Xihu Sag, East China Sea Basin

The study aims to address the challenges of developing gas reservoirs in the Eocene Pinghu Formation of the southern part of the central anticline belt in the Xihu Sag, East China Sea Basin, which are characterized by deep burial (>3 900 m), strong heterogeneity, and low-permeability tightness limiting gas production. A comprehensive evaluation method for sweet spots identification and classification in deep, low-permeability, tight gas reservoirs, integrating lithology, physical properties, and gas-bearing characteristics, was proposed. By conducting reservoir classification modeling and amplitude variation with offset (AVO) analysis via fluid replacement modeling, it was clarified that the velocity ratio and Poisson impedance were sensitive to lithology, the Poisson damping factor was sensitive to physical properties, and the fluid factor and bulk modulus were sensitive to gas-bearing properties. Based on AVO response characteristics, a geophysical response pattern for four classes of sweet spots was established.Relying on facies-controlled pre-stack elastic inversion, key parameters such as velocity ratio, Poisson impedance, Poisson damping factor, fluid factor, and bulk modulus were estimated to achieve a comprehensive prediction of reservoir lithology, physical properties, and gas-bearing characteristics. A multi-attribute sweet-spot evaluation framework based on sparse principal component analysis (SPCA) and adaptive weighting (AW) was further proposed. The L1-regularization was used to enable sparsity in attribute loadings to automatically screen the main seismic attributes, and adaptive weights were constructed based on sparsity and loading strength to improve attribute fusion, ultimately selecting high-quality sweet spots with good physical properties and high gas saturation. Application of this method to target layers in the gas field showed that the comprehensive matching rate between the prediction results and the well log interpretation and gas testing results exceeded 80%. This approach effectively identified favorable zones for sweet spot development in gas reservoirs, providing a reliable basis for rolling evaluation, well location deployment, and efficient development of deep, low-permeability, tight gas reservoirs.