Paper submitted to EAGE Annual 2026

Summary

Full Waveform Inversion (FWI) has evolved to a robust and reliable technique to build accurate velocity models for depth imaging which has been routinely applied in seismic processing for both land and marine surveys. By incorporating more accurate physics, elastic FWI further enhances this capability, delivering highly precise models even in structurally complex regions with strong impedance contrasts such as salt boundaries (Macesanu et al., 2024, Liu et al., 2025).  FWI‑derived reflectivity (FDR), computed as the directional derivative of the velocity model following high‑frequency FWI, produces a high‑resolution image comparable to least‑squares RTM results. (Wang et al., 2021).  

Conventional single‑parameter elastic FWI focuses mainly on P‑wave velocity inversion, with shear‑wave velocity and density updated passively through analytical relationships linked to Vp, often calibrated using well information. Although this approach has proven robust and effective, in geological settings exhibiting Class IIp AVO anomalies, which are typically spatially localized, density often exhibits trends opposite to those of P‑ and S‑wave velocities, such that a single empirical relationship cannot adequately represent the entire model.  

Alternatively, multiparameter elastic FWI (MP-EFWI) simultaneously inverts for P-wave velocity and reflectively (Huang et al., 2025, Macesanu et al., 2025). In this approach, the elastodynamic equations are formulated in terms of velocities and P-wave reflectivity. Density is not a required parameter for modelling and inversion although its relative changes can be estimated from reflectivity, which is iteratively updated within the same FWI framework. To minimize the crosstalk between background velocity and reflectivity in our MP-EFWI framework, we applied an effective scale separation of the FWI gradient.

 In this work, we apply MP-EFWI workflow to narrow‑azimuth data containing a Class IIp AVO anomaly identified at a well location. Results show that both background velocity and reflectivity can be properly inverted, allowing the derivation of a relative density profile that agrees well with the well log, which demonstrates the potential of our MP-EFWI for reservoir interpretation.