The invention discloses a
reflection coefficient inversion reservoir prediction method based on a weighted stacking
noise suppression strategy. The method includes six main steps: 1), through a
noise disturbance analysis, determining a source of
noise influencing the inversion effect, and distinguishing
seismic noise and inversion noise; 2) performing
seismic noise suppression preprocessing on seismic data through a post-stack
wavelet frequency division method; 3) for de-noised seismic data, constructing
reflection coefficient inversion key parameters (a
wavelet inverse transform factor F, an inversion predominant
frequency band Q, and an inversion initial model S(n)); 4) suppressing inversion noise through weighted stacking of inversion results of multiple regularization parameters to obtain
reflection coefficient inversion results; 5) performing
quality control (QC) on the inversion results through
spectrum analysis and a calibrated
sound wave curve to obtain final result data; and 6) performing sand body fine interpretation based on the reflection coefficient result data, and predicting sand body
spatial distribution characteristics. Compared with other thin reservoir
prediction methods, according to the reflection coefficient inversion reservoir prediction method, the inversion anti-noise performance is high, thin-layer interference in the original seismic data can be eliminated, the prediction precision of the
thin layer is greatly improved, and oil and gas exploration risks are reduced.