Least-squares Reverse Time Migration Based On GPU/CPU Acceleration

With the exploration of oil and gas,the main task of oil gas exploration is complicated structural oil gas reservoir exploration and deep carbonate reservoir exploration.Reverse time migration is the use of the more deviation method,the effect is also more ideal.The method as a result of the two-way wave equation extrapolation method,so it can imaging of steep dip angle,turning wave can also correct imaging.But conventional reverse time migration method in imaging will be introducing imaging noise at low frequencies,decreases the resolution of small scale structure imaging effect difference,the effect of complex structure imaging is fuzzy,especially there is no accurate imaging of porous,fractural model carbonate.Moreover,the method can not be accurate amplitude imaging for the amplitude of the method.The amplitude of the shallow deep in the deep,the imaging energy is lower,and the imaging amplitude is lower.The above problems will restrict the application of the least square inverse time migration in deep reservoir exploration.In order to solve the above problem,the inverse time migration is put into the frame of least square inversion,and the least square inverse is put forward.Inverse least squares migration is the least squares inversion point of view,in the Born approximation based on,in a known background velocity field(low wavenumber component)under the premise,on the reflection coefficient(high wave number components)were repeatedly iterative matching.This method can effectively solve the problem of unequal amplitude preserving and Asanaka's deep amplitude distribution in the conventional reverse time migration.The least square inverse time migration will be the conventional inverse time migration as the first imaging result,and constantly iterative and update the final imaging results.Because of the multiple matching iteration of the amplitude,the least square inverse time migration imaging results have higher amplitude preserving,fidelity and resolution.By marmousi2 model and carbonate rock model trial proved the method of small scale structure imaging effect is good,can accurate on complex structure imaging,especially in the western region,porous,fractural model carbonate fine imaging,can accurate distinguish sewing hole top and bottom as well as the flanking boundary,diffraction point of convergence.Moreover,the method has good amplitude preserving,and has realized the true amplitude imaging,and can realize the fine imaging of complex oil and gas reservoir.At the same time to offset the massive calculation of inverse least squares problems introduced hybrid GPU Programming technology,greatly improve the computational efficiency and computation of model of computing time statistics proved based on GPU accelerated the least squares inverse time migration has higher computational efficiency.These characteristics is very important for our country to unconventional oil and gas exploration and deep carbonate reservoir exploration.