Velocity estimation in GPR data based on diffraction analysis: Methodology and application to Antarctic firn

Abstract This study presents a methodological framework for estimating electromagnetic wave velocities in ground‐penetrating radar (GPR) data based exclusively on the analysis of diffractions. The approach integrates diffraction separation using the plane wave destruction algorithm and subsequent velocity refinement through the residual diffraction moveout (RDM) technique. The methodology was initially validated on synthetic datasets with controlled geological complexity, including models with lateral and vertical velocity variations. These synthetic models were intentionally designed to simulate environments rich in diffracted events, replicating conditions similar to those observed in real data. Results from synthetic tests demonstrated the capacity of the methodology to isolate diffraction hyperbolas effectively, estimate local velocities and generate coherent velocity models suitable for migration. The workflow was then applied to real GPR data acquired on the Detroit Plateau, Antarctic Peninsula. The resulting velocity model aligned well with previous studies based on common midpoint analysis, highlighting structural variations associated with firn densification and the firn–ice transition zone. Kirchhoff migration using the RDM‐derived velocity model led to more precise reflector imaging and improved subsurface structural interpretation. This study demonstrates that diffraction‐based velocity modelling can provide accurate and high‐resolution subsurface characterization, particularly in environments with low lateral velocity contrast where conventional methods may be limited.