A Deformation-Prior-Constrained Unified Error Correction Method for Wide-Area ALOS/PALSAR InSAR Processing

Abstract

Separating trend-like errors from long-wavelength deformation signals remains challenging in wide-area ALOS/PALSAR InSAR processing. This letter proposes a unified error correction method incorporating deformation priors. The prior information is used to generate an initial deformation mask for identifying intraframe stable points, while corresponding points in overlapping regions provide interframe constraints. Based on these constraints, an iterative weighted least-squares adjustment model is established to jointly estimate polynomial error parameters for multiple frames and unify the deformation reference datum. Validated over the North China Plain (NCP), the proposed method is compared with conventional error correction approaches. Experimental results show that the proposed method converges within two iterations, reducing the P95 of adjustment accuracy for all frames from 16.7 to 8.0 mm/yr, an improvement of 52.1%. It effectively suppresses systematic trend errors while preserving the integrity of long-wavelength deformation signals, significantly improving the interframe consistency across overlapping zones. The proposed method provides an efficient and scalable technical solution for automated wide-area ALOS/PALSAR error correction.