Abstract:Seismic velocity inversion plays a vital role in various applied seismology processes. A series of deep learning methods have been developed that rely purely on manually provided labels for supervision; however, their performances depend heavily on using large training data sets with corresponding velocity models. Because no physical laws are used in the training phase, it is usually challenging to generalize trained neural networks to a new data domain. To mitigate these issues, we have embedded a seismic forward modeling step at the end of a network to remap the inversion result back to seismic data and thus train the neural network through self-supervised loss, i.e., the misfit between the network input and output. As a result, we eliminate the need for many labeled velocity models, and physical laws are introduced when back-propagating gradients through the seismic forward modeling step. We verify the effectiveness of our approach through comprehensive experiments on synthetic data sets, where self-supervised learning outperforms the fully supervised approach, which accesses much more labeled data. The superior performance is even more significant when compared with a new data domain that has velocity models with faults and more geologic layers. Finally, in case of unknown and more complex data types, we develop a network-constrained full-waveform inversion (FWI) method. This method refines the initial prediction of the network by iteratively optimizing network parameters other than the velocity model, as found with the conventional FWI method, and demonstrates clear advantages in terms of interface and velocity accuracy. With these measures (self-supervised learning and network-constrained FWI), our physics-driven self-supervised learning system successfully mitigates issues such as the dependence on large labeled data sets, the absence of physical laws, and the difficulty in adapting to new data domains.

Extremely Weak Supervision Inversion of Multi-physical Properties

Multi-Wave Stochastic Inversion Physical Parameters Prediction Method Driven by Linearized Rock Physics

Physics-Driven Deep Learning Inversion for Direct Current Resistivity Survey Data

Coupled Hydrogeophysical Inversion to Identify Non-Gaussian Hydraulic Conductivity Field by Jointly Assimilating Geochemical and Time-Lapse Geophysical Data

ContrasInver: Ultra-Sparse Label Semi-Supervised Regression for Multidimensional Seismic Inversion

An Intriguing Property of Geophysics Inversion

Multiscale 3-D Stochastic Inversion of Frequency-Domain Airborne Electromagnetic Data

Gravity and Magnetic Data Inversion With Random Projection

Physics-driven self-supervised learning system for seismic velocity inversion

A Deep Learning Gravity Inversion Method Based on a Self-Constrained Network and Its Application

Solving multiphysics-based inverse problems with learned surrogates and constraints

Application of a model-driven simultaneous prestack inversion of rock physical properties in ultra-deep Ordovician carbonate reservoirs in the Shunbei area

Full-Wavefield Inversion: An Extreme-Scale PDE-Constrained Optimization Problem

Multidimensional Petrophysical Seismic Inversion Based on Knowledge-Driven Semi-Supervised Deep Learning

Tunnel resistivity deep learning inversion method based on physics‐driven and signal interpretability

ContrasInver: Ultra-Sparse Label Semi-supervised Regression for Multi-dimensional Seismic Inversion

Improved methodology for deep aquifer characterization using hydrogeological, self-potential, and magnetotellurics data

Azimuthal Amplitude Difference-Based Multidomain Seismic Inversion for Fracture Weaknesses

Joint Inversion of Potential Field Data with Adaptive Unstructured Tetrahedral Mesh

3-D Magnetotelluric Deep Learning Inversion Guided by Pseudo-Physical Information

Physics-Informed Deep Learning Inversion with Application to Noisy Magnetotelluric Measurements