Stable and efficient seismic impedance inversion using quantum annealing with L1 norm regularization

Abstract

Seismic impedance inversion makes a significant contribution to locating hydrocarbons and interpreting seismic data. However, it suffers from non-unique solutions, and a direct linear inversion produces large errors. Global optimization methods, like simulated annealing, have been applied in seismic impedance inversion and achieved promising inversion results. Over the past decades, there has been an increasing interest in quantum computing. Due to its natural parallelism, quantum computing has a powerful computational capability and certain advantages in solving complex inverse problems. Within this article, we present a stable and efficient impedance inversion using quantum annealing with L1 norm regularization, which significantly improves the inversion accuracy compared to the traditional simulated annealing method. Tests on a one-dimensional ten-layer model with noisy data demonstrate that the new method exhibits significantly improved accuracy and stability. Additionally, we perform seismic impedance inversion for synthetic data from the Overthrust model and field data using two methods. These results demonstrate that the quantum annealing impedance inversion with L1 norm regularization dramatically enhances the accuracy and anti-noise ability.