The M sandstone member within the Cretaceous Napo Formation is a critical hydrocarbon reservoir in the Oriente Basin's foredeep belt. However, its characterization is severely challenged by its thin-bed nature (2–8 m) and rapid lateral facies variations. The available 3D seismic data has a dominant frequency of about 45 Hz, yielding an apparent resolution of approximately 26 m. Conventional seismic techniques, such as deconvolution, spectral whitening, and inverse Q filtering, rely on stationary wavelet assumptions and are sensitive to noise and parameter inaccuracies. This limitation leads to the misinterpretation of the “isolated reservoir” in the M01 well, which has long hindered progress in exploration and development in the area. To address these challenges, we develop a novel high-resolution seismic processing workflow centered on precise reflectivity inversion. Our approach systematically integrates three key innovations: First, targeted denoising is applied to enhance the signal-to-noise ratio, ensuring the stability of subsequent inversion. Second, we overcome the stationary wavelet assumption of conventional methods by employing the generalized S-transform to construct a discrete, time-varying wavelet library, which accurately captures the seismic wavelet's frequency attenuation with depth. Third, we integrate this time-varying wavelet model with a reflectivity inversion algorithm based on odd-even reflection theory and a least-squares optimization, enabling high-fidelity recovery of thin-layer reflection coefficients. The application of this workflow in the M Oilfield yields a transformative improvement in seismic resolution. The processed data achieves a significant broadening of the effective frequency bandwidth, elevating the dominant frequency from 45 Hz to 85 Hz. This enhancement consequently improves the theoretically detectable thickness from approximately 14 m to approximately 5 m. Most importantly, the processed seismic profiles clearly delineate ultra-thin sand bodies, ranging from 3 to 5 m, with drilling results exhibiting high consistency with seismic interpretation (e.g., successful wells M01 and M05-2 versus dry wells M05 and M05-1). The resulting seismic attributes also reveal clearer geological features, such as delta outlines and underwater distributary channels. This study successfully resolves the exploration dilemma that persisted in the area for over two decades, overturning the “isolated reservoir” model and revealing the continuous distribution characteristics of the M sandstone. The demonstrated workflow provides a robust and systematic solution for characterizing thin-bed reservoirs, directly leading to increased exploration success. The established technical framework offers a valuable and practical reference for the detailed exploration and development of similar thin sandstone reservoirs in foredeep and slope settings worldwide.

As seismic signals propagate underground, the subsurface media absorb high–frequency components, decreasing seismic resolution and limiting the identification and characterization of thin reservoirs. To address the limitation, this paper proposes a multichannel spectral fitting (MSF) method. The MSF method aims to enhance seismic resolution by considering the spectral characteristics and the correlations between adjacent seismic traces. The key to the MSF method involves utilizing the amplitude spectrum of the attenuated Ricker wavelet to construct an objective function for spectral fitting, leading to improved seismic resolution. Furthermore, the MSF method establishes the correlation between adjacent seismic traces as a constraint to stably solve the target parameters based on the entire seismic spectrum, which helps obtain horizontally consistent and more realistic seismic signals. Synthetic and field seismic examples demonstrate that the proposed method not only provides higher-resolution seismic signals but also reveals more fine details of thin reservoirs compared to the time-variant spectral whitening method. It is concluded that the MSF method is a promising tool for seismic signal processing.