Interferogram Formation, Flat-Earth Removal, Filtering, and Phase Unwrapping using InSAR Technology

InSAR (Interferometric Synthetic Aperture Radar) technology is a remote sensing technique that measures surface deformation or elevation changes through radar signal phase differences. When processing InSAR data, a series of steps are typically required to obtain accurate phase information, ultimately generating interferograms or deformation maps.

Interferogram Formation Interferograms are generated by conjugate multiplication of complex data from master and slave images, containing phase difference information. This step requires precise registration of two SAR images to ensure consistent positioning of identical ground features in both images. Through complex multiplication operations implemented in MATLAB using functions like `conj()` and element-wise multiplication (`.*`), phase difference information can be extracted to form initial interference fringes. The registration accuracy is critical and often achieved through cross-correlation algorithms.

Flat-Earth Removal (Eliminating Flat-Earth Effect) Due to Earth's curvature and radar side-looking imaging geometry, interferograms contain fringes caused by flat-earth phase. These fringes do not represent surface deformation or elevation changes and must be removed using reference DEM (Digital Elevation Model) or geometric models. In code implementation, this involves calculating the expected flat-earth phase using radar parameters and satellite orbits, then subtracting it from the original interferogram using phase subtraction operations, leaving only true topographic or deformation signals.

Filtering (Noise Suppression) Interferogram phases are typically affected by noise such as thermal noise and temporal decorrelation. To improve signal-to-noise ratio, phase filtering is necessary. Common filtering methods include Goldstein filtering and Boxcar mean filtering, which reduce phase noise through local smoothing or adaptive approaches. In MATLAB implementations, Goldstein filter can be applied using frequency-domain multiplication with adaptive filter kernels, while Boxcar filtering uses spatial convolution with uniform kernels to enhance phase quality for subsequent unwrapping.

Phase Unwrapping Interferometric phases are wrapped modulo 2π, requiring phase unwrapping algorithms to recover true absolute phases. Common unwrapping methods include Minimum Cost Flow (MCF), Branch Cut, and Snaphu algorithms. The unwrapping process must maintain phase continuity and prevent error propagation. Code implementations typically involve creating quality maps to guide unwrapping paths, using network flow optimization in MCF or residue identification in Branch Cut methods, ultimately producing continuous phase fields through integration-like algorithms.

These steps constitute the core workflow of InSAR data processing. Using tools like MATLAB, automated processing can be implemented through customized scripts combining image processing toolbox functions and specialized InSAR algorithms, ultimately applied to topographic mapping and surface deformation monitoring applications.