Abstract:  In the eastern Ordos basin,  thin reservoirs in the Third Submember of the Second Member of Shanxi Formation (Shan23 Submember) of Daning-Jixian Block B are frequently closely associated with  carbonate rocks atop  Taiyuan Formation. These reservoirs are tight, with  complex lithology, small impedance difference between sandstone and mudstones strong heterogeneity, and great difficulty in reservoir prediction. In order to effectively improve the accuracy of quantitative interpretation of seismic data and implement the distribution characteristics of thin reservoirs in Block B, a novel prediction method for thin tight sandstone reservoirs amalgamating seismic phase, frequency extension and inversion phase has been proposed. This method firstly makes clear the characteristics of core facies, electrical logging facies and seismic facies of the reservoir based on  geological and logging data. Subsequently, spectral recovery and frequency extension technologies are used to improve the resolution of seismic data. Lastly, the lithology curves are regenerated, and waveform indicative inversion is performed using the high-resolution seismic data after frequency extension and the planar distribution of thin reservoirs is depicted to realize the quantitative prediction of tight and thin reservoirs in the Shan23 Submember of the study area. The results show that there are four types of seismic response characteristics for the target layer: ①Subaqueous distributary channel facies, characterized by a combination of strong, medium-weak and strong amplitude; ②Lateral area of  subaqueous distributary channel facies, characterized by a combination of strong, weak and strong amplitude; ③Sheet sand and distal sand bar facies, characterized by a combination of medium-weak, medium-strong and strong amplitude; ④Distributary bay facies, characterized by a combination of weak and strong amplitude. The dominant frequency of the seismic data is  increased from 30 Hz to 40 Hz, and the frequency bandwidth is  expanded from 10-60 Hz to 8-80 Hz, making more clear images for thin reservoirs in seismic data after frequency extension. The reconstructed curves show a good linear relationship between the sandstone thickness obtained from the post-frequency-extension seismic data inversion and the sandstone thickness from well, with a determination coefficient of 0.911 and a conformity of 84%. Based on the inversion data, horizontal well drilling is  guided, achieving a reservoir encounter rate of 95%.

Key words:  , reservoir prediction, complex lithology, thin reservoir, curve reconstruction, waveform indicative inversion