Online Journal for E&P Geoscientists

Fei, Tong ^*1; Luo, Yi ¹; Qin, Fuhao ¹; Zhang, Sanzong ²
(1) EXPEC ARC, Saudi Aramco Oil Company, Dhahran, Saudi Arabia. (2) Geoscience, KAUST, Thuwal, Saudi Arabia.

Full waveform inversion (FWI) is an iterative forward modeling and inversion procedure that minimizes the difference of the observed and calculated seismic data. This method, based on acoustic or elastic wave equation, has recently gained widespread interest for use in petroleum exploration applications. FWI is well known for its ability to produce high resolution and accurate subsurface models in areas of complex geology, such as the Red Sea; hence, it may improve the seismic imaging quality, especially in the subsalt zones. Here, we implemented acoustic FWI in the time and space domains and carried out various tests on both a simple syncline model, and a more complex model containing salt bodies and steeply dipping events. Test results for both models show that the convergence of the FWI is relatively insensitive to the choice of initial models; however, a good starting model helps FWI converge more quickly. Moreover, the test results show that the success of FWI relies on the existence of the low-frequency component within the data. Starting FWI using the low-frequency (< 5Hz) component of the data, and gradually including the higher frequency data, we can obtain a good high resolution velocity model for seismic imaging. Without the low frequency data, the current FWI implementation cannot produce a reliable velocity model. Therefore, new strategies must be developed to make FWI technology less dependent on the low-frequency content of the data.