DispFormer: A Pretrained Transformer Incorporating Physical Constraints for Dispersion Curve Inversion

AbstractSurface wave dispersion curve inversion is crucial for estimating subsurface shear‐wave velocity , yet traditional methods often face challenges related to computational cost, non‐uniqueness, and sensitivity to initial models. While deep learning approaches show promise, many require large labeled data sets and struggle with real‐world data sets, which often exhibit varying period ranges, missing values, and low signal‐to‐noise ratios. To address these limitations, this study introduces DispFormer, a transformer‐based neural network for profile inversion from Rayleigh‐wave phase and group dispersion curves. DispFormer processes dispersion data independently at each period, allowing it to handle varying lengths without requiring network modifications or strict alignment between training and testing data sets. A depth‐aware training strategy is also introduced, incorporating physical constraints derived from the depth sensitivity of dispersion data. DispFormer is pre‐trained on a global synthetic data set and evaluated on two regional synthetic data sets using zero‐shot and few‐shot strategies. Results show that even without labeled data, the zero‐shot DispFormer generates i