GNGTS 2024 - Atti del 42° Convegno Nazionale

Session 3.3 GNGTS 2024 SD, allowing the inference of diferent frequencies through the applicaton of a dedicated Scaling Factor (SF). The results illustrate the efectveness of the proposed approach in both low and high-frequency inferences using exemplary seismic data from the Viking Graben Line 12. The low-frequency model successfully shifs the central frequency from 14Hz to 8Hz while preserving amplitude and accurately predictng signal interference. The high-frequency model demonstrates reliable inference when compared to reference data the applicaton of a band-pass flter between 10-40Hz, revealing a beter detectability of diferent features, such as dipping refectors and difracton paterns, that were challenging to interpret in the lower frequency input data, while are confrmed by the reference data. Overall, the proposed LSTM Neural Network-based approach proves to be a promising soluton for addressing frequency gaps in seismic signal processing, ofering high adaptability, scalability, and enhanced predictve capabilites for both low- and high-frequency components predictons. References Camacho-Ramírez, E., González-Flores, E., & Campos-Enriquez, J. (2016). Discrete wavelet transform-based multresoluton analysis and spectral enhancement to characterize heavy oil reservoirs in the southern gulf of mexico region. Interpretaton, 4(4), T497-T505 . htps://doi.org/ 10.1190/int-2015-0184.1 Chopra, S., Alexeev, V., & Sudhakar, V. (2003). High-frequency restoraton of surface seismic data. The Leading Edge, 22(8), 730-738 . htps://doi.org/10.1190/1.1605071 Du, W., Xin, P., Wang, P., & Sun, Y. (2016). Multple-track ant body atribute extracton method improved.. htps://doi.org/10.2991/iceeg-16.2016.72 Guo, J. and Wang, Y. (2004). Recovery of a target refecton underneath coal seams. Journal of Geophysics and Engineering, 1(1), 46-50 . htps://doi.org/10.1088/1742-2132/1/1/005 Karsli, H. (2006). Further improvement of temporal resoluton of seismic data by autoregressive (ar) spectral extrapolaton. Journal of Applied Geophysics, 59(4), 324-336 . htps:// doi.org/10.1016/j.jappgeo.2005.11.001 Karsli, H. (2010). An applicaton of the autoregressive extrapolaton technique to enhance deconvoluton results: a 2d marine data example. Geophysical Prospectng, 59(1), 56-65 . htps:// doi.org/10.1111/j.1365-2478.2010.00895.x Karsli, H., Dondurur, D., & Çifçi, G. (2006). Applicaton of complex-trace analysis to seismic data for random-noise suppression and temporal resoluton improvement. Geophysics, 71(3), V79- V86 . htps://doi.org/10.1190/1.2196875 Li, Y. and Demanet, L. (2016). Full-waveform inversion with extrapolated low-frequency data. Geophysics, 81(6), R339-R348 . htps://doi.org/10.1190/geo2016-0038.1 Nakayama, S. and Blacquière, G. (2021). Machine-learning-based data recovery and its contributon to seismic acquisiton: simultaneous applicaton of deblending, trace reconstructon, and low-frequency extrapolaton. Geophysics, 86(2), P13-P24. htps://doi.org/10.1190/ geo2020-0303.1