Flexible Automatic Differentiation‐Based Framework for Seismic Data Q Compensation

ABSTRACT Seismic attenuation significantly degrades the resolution and fidelity of seismic data and remains a key challenge for high‐resolution subsurface imaging. Although numerous attenuation compensation methods have been developed, most rely on fixed optimization models and analytical derivations. When applied to seismic data from different geological settings or with varying complexity, these methods often require reformulation of objective functions or rederivation of gradient expressions. This dependency limits the flexibility and scalability of existing approaches. Automatic differentiation (AD) enables accurate gradient computation by tracking the computational graph of the forward modelling process, thereby decoupling the optimization procedure from explicit analytical derivations. On the basis of this concept, we propose an AD‐based Q compensation flexible framework that provides a modular and generalized formulation of the compensation workflow. The proposed method achieves stable, accurate and customizable Q compensation and remains robust under low signal‐to‐noise ratio conditions. Applications to both synthetic and field seismic data demonstrate effective high‐frequency energy recovery, noise suppression and preservation of structural integrity.