Multitrack Adjustment for Wide-Area SAR Interferometry in Sparse GNSS Scenarios: A Hydrological Case Study Over North China Plain

Interferometric Synthetic Aperture Radar (InSAR) is widely used for mapping large-scale surface deformation and capturing long-wavelength geophysical signals. Mosaicking multiple frames and tracks remains a major challenge for obtaining accurate vertical and horizontal velocity measurements over wide areas. This study reviews conventional adjustment-based mosaicking methods, with a focus on ill-conditioning problems under sparse GNSS scenarios. Particularly, we propose a two-step mosaicking framework for multi-track scenarios that treats biases between adjacent frames and between InSAR and GNSS as linear and nonlinear components. The linear component is addressed using a polynomial-model-based block adjustment, which connects along-track and across-track frames to construct a globally constrained network that mitigates ill-conditioning issues. The nonlinear component is handled through a constant-model-based adjustment within quadtree-decomposed local tiles, guided by at-test for adaptive segmentation. This approach serves as an alternative to high-degree polynomial models that are susceptible to overfitting. The rationale behind the proposed framework is to use a stable spatial network and low-degree models to enhance mosaicking accuracy while ensuring computational stability. Comprehensive experiments using synthetic data and Sentinel-1 imagery over parts of the North China Plain (NCP) demonstrate the effectiveness of the proposed approach in measuring vertical and horizontal deformation rates induced by groundwater extraction.