Land seismic surveys are often carried out in regions with a complex structure of the upper part of the section. The main complicating factors are significant differences in relief heights, inhomogeneity of the low-velocity zone and the presence of high-velocity layers of permafrost. An important stage of processing, in this case, is the construction of a velocity model of the near-surface area based on the times of the first arrivals of seismic waves and its further accounting when constructing seismic images. The presence of boundaries in the resulting model may be necessary when choosing a floating or final datum or embedding in a depth model.
The article discusses approaches to automating the procedure for constructing a near-surface model within the framework of the reflected waves method. First, a method for automatic picking of the first break times was implemented based on convolutional neural networks. Testing on real data has shown that the use of neural networks provides a more robust acquisition of first break times compared to standard approaches implemented in processing packages. Secondly, a method is proposed for constructing a layered near-surface velocity model based on the first break times. We use ray seismic tomography to build a smooth velocity model, then we convert it to a layered model. Testing on synthetic data simulating the geological conditions of Western Siberia has shown the possibility of building the layered near-surface model with good accuracy. Finally, an example of the implementation of methods in the form of modules in a processing package for their further use for processing real data is shown.
References
1. Dolgikh Yu.N., Mnogourovnevaya seysmorazvedka i kinematicheskaya inversiya dannykh MOV – OGT v usloviyakh neodnorodnoy VChR (Multilevel seismic prospecting and kinematic inversion of the reflection and common depth point methods data in the conditions of the inhomogeneous upper part of the section), Moscow: EAGE Geomodel', 2014, 212 p.
2. Sysoev A.P., Prikladnye zadachi kompensatsii neodnorodnosti verkhney chasti razreza pri obrabotke i interpretatsii seysmicheskikh dannykh (Applied problems of compensation of heterogeneity of the upper part of the section during processing and interpretation of seismic data), Novosibirsk: Publ. of IPGG SB RAS, 2011, 88 p.
3. Davletkhanov R., Consideration of near-surface heterogeneities by static corrections or their inclusion in the reservoir model of the medium - what to choose (In Russ.), Tekhnologii seysmorazvedki, 2015, no. 1, pp. 76–91.
4. Akram J., Eaton D.W., A review and appraisal of arrival-time picking methods for downhole microseismic data arrival-time picking methods, Geophysics, 2016, V. 81, pp. KS71-KS91.
5. Kong Q., Trugman D.T., Ross Z.E. et al., Machine learning in seismology: Turning data into insights, Seismological Research Letters, 2018, V. 90, no. 1, pp. 3–14.
6. Boganik G., Gurvich I., Seysmorazvedka (Seismic survey), Tver: AIS Publ., 2006, 744 p.
7. Yilmaz Ö., Seismic data analysis, Tulsa: Society of exploration geophysicists, 2001, V. 1, 1809 p.
8. Seismic tomography: with applications in global seismology and exploration geophysics: edited by Nolet G.,Luxembourg:, Springer Science & Business Media, 2012, V. 5, 385 p.
9. Zelt C.A., Traveltime tomography using controlled-source seismic data, Encyclopedia of Solid Earth Geophysics, 2011, pp. 1453–1473.
10. Nikitin A.A., Serdyukov A.S., Duchkov A.A., Cache-efficient parallel eikonal solver for multicore CPUs, Computational Geosciences, 2018, V. 22, no. 3, pp. 775–787.
11. Shalashnikov A.V., Finikov D.B., Khokhlov N.I., Ivanov A.M., New approaches in optimization of calculation of wave fields directly related to the selected target area of seismic response (In Russ.), Geofizicheskie tekhnologii, 2019, no. 1, pp. 4–32.
12. Zelt C.A., Sain K., Naumenko J.V., Sawyer D.S., Assessment of crustal velocity models using seismic refraction and reflection tomography, Geophysical Journal International, 2003, V. 153, no. 3, pp. 609–626.
