Reconstructed low-frequency data frequency domain full-waveform inversion method based on convolution idea and accurate earthquake source

Abstract

The present invention relates to a reconstructed low-frequency data frequency domain full-waveform inversion method based on a convolution idea and an accurate earthquake source. The method comprisesthe steps that: a reliable earthquake source function is inverted by employing direct wave components of observation data; low-frequency data of self-excited self-closing points is reconstructed basedon the convolution idea; and finally, the reconstructed data is applied to perform a low-frequency starting part of multi-scale full-waveform inversion to allow the inversion to have a good beginningand finally obtain a good result. The method provided by the invention effectively avoids occurrence of cycle skips, so that the inversion result is greatly improved, and minimum convergence of various errors is effectively recovered. An idea of a supergun of dynamic random coding is employed at the final inversion part, so that the efficiency of the full-waveform inversion is effectively improved. The reconstructed low-frequency information of the self-excited self-closing points are low-frequency information on the aspect of mathematics to provide large-scale disturbance for the full-waveform inversion, are real low-frequency information on the aspect of objective physics, and can be used for other data processing directions.

Description

technical field [0001] The invention relates to underground imaging for seismic exploration, especially based on the idea of ​​convolution and accurate source reconstruction of low-frequency information missing from seismic data, combined with super shot and multi-scale ideas to perform frequency-domain full waveform inversion. Background technique: [0002] With the continuous deepening of oil and gas development, the velocity model obtained by conventional imaging technology has been unable to meet the demand for high-precision underground velocity models in the process of searching for oil and gas deposits and in the process of exploitation. In this context, combined with the rapid update of computer technology, full waveform inversion (FWI) has received widespread attention and rapid development, and has become a major research hotspot in geophysics today. In the 1980s, Tarantola (1984) proposed the full waveform inversion in the time domain. He first proposed to use the...

AI Technical Summary

Problems solved by technology

Hu (2014) proposed using the principle of wave interference to obtain a relatively low-frequency analog data for full waveform inversion by using two relatively high-frequency data interference, but it is difficult to reflect the physical meaning of interference in specific numerical calculations

[0004] The existing frequency-domain full waveform inversion is based on multi-scale methods, that is, the inversion starts from low frequencies first, and the seismic information of frequencies below 5 Hz is generally missing in the actual data, and the information of 5-7 Hz is not necessarily reliable in some cases , so in practical applications, frequency-domain full waveform inversion cannot have a good start, and it is difficult to obtain good results

Images