Amplitude-invariant phase masking for coherence recovery in scattered wavefields

Coherent summation of multichannel recordings relies on phase stability across measurements. In scattered wavefields, near-surface heterogeneities introduce record-dependent phase distortions that decorrelate otherwise coherent signals and degrade summation quality. Conventional approaches to recovering phase coherence implicitly tie phase estimation to signal amplitude, introducing bias where amplitude variability is large. We present a framework based on circular statistics that separates phase estimation from amplitude entirely, ensuring all records contribute equally to the representative phase regardless of their energy. Synthetic experiments confirm that the proposed method outperforms conventional approaches when amplitude variability is present, offering a practical path toward more robust coherence recovery in seismic imaging and other wave-based sensing applications.