Methods for improving the vertical resolution of natural gamma ray logs.

As major oilfields in China enter the mid-to-high water-cut development stage, conventional reservoirs are gradually depleting, and thin reservoirs have become the focus of hydrocarbon exploration and development. However, the evaluation of thin reservoirs faces severe challenges due to the surrounding rock effect, which significantly reduces the vertical resolution of conventional logging curves and causes deviations between measured responses and true formation parameters. This study aims to enhance the vertical resolution of natural gamma-ray logs, thereby improving the identification of thin-layer information and providing technical support for accurate evaluation and efficient development of thin reservoirs. Based on logging principles and time-frequency analysis theory, two resolution enhancement schemes are proposed. First, wavelet analysis is applied to the original logging signals for multi-scale decomposition, effectively extracting and enhancing high-frequency details that characterize thin layers while suppressing background noise from surrounding rocks. Second, the deconvolution method is employed to eliminate the smoothing effect of the logging tool response function on the original formation signals, achieving a "sharpening" effect on the curves. Rapid forward modeling is introduced to validate the processing results of both methods, and their respective advantages and disadvantages are compared and analyzed. Both wavelet transform and five-point deconvolution significantly enhance thin-layer signals, bringing the processed formation parameters closer to the true values. The wavelet transform achieves the best performance when the decomposition level is set to 3, while the five-point deconvolution reconstructs the curves using a derived formula based on the instrument response function. Both methods increase the vertical resolution of natural gamma-ray logs to 0.2 m. Furthermore, the resolution improvement effect of processed field-measured natural gamma-ray logs is verified using multiple logging curves including micro-resistivity (RXO) and compensated density (DEN) logs, and the verification results show favorable consistency. These techniques demonstrate strong feasibility and practical value for thin-layer identification. The wavelet transform is flexible and integrates denoising with resolution enhancement, but its performance heavily depends on parameter selection. The five-point deconvolution method has clearer physical meaning and offers better stability, yet its resolution improvement is relatively limited. Unlike conventional standalone deconvolution or filtering-based resolution correction approaches, which suffer severe noise interference and indistinct bed boundaries when applied to heterogeneous interbedded formations, the dual collaborative processing framework proposed in this study combines multi-scale frequency decomposition and tool response inversion, delivering an innovative integrated solution customized for geologically complex thin interbed sequences.