Spaceborne SAR Time-Series Radiometric Cross-Calibration via Spatiotemporal Screening-Regression-Modeling Optimization

Maintaining temporal radiometric consistency in synthetic aperture radar (SAR) satellite imagery requires advanced time-series radiometric calibration. Conventional approaches that rely on artificial calibration targets incur high operational costs. Capitalizing on the cost-efficiency of natural-target-based radiometric cross-calibration, this study extends cross-calibration principles to time-series calibration through three innovations: (1) A multi-criterion calibration target screening method through integrated analysis combining satellite revisit frequencies, target spatial histogram characteristics, and temporal scattering stability to overcome spatio-temporal constraints. (2) An enhanced calibration constant calculation approach that iteratively applies robust linear regression and removal of spatial pixel outliers to improve radiometric accuracy. (3) An inter-temporal calibration constant optimization mechanism by implementing Kalman filtering and sensor-attenuation modeling to ensure long-term stability. Experimental validation using Sentinel-1 data reveals that the maximum root mean square error (RMSE) across 69 temporal points is 0.36 dB, representing a 62% reduction compared to conventional methods. Furthermore, the Kalman-filtered results achieve a temporal stability of 0.0269 dB, an 89% improvement over unprocessed calibrations. This methodology enables continuous monitoring of SAR sensor performance with improved temporal consistency, which is crucial for the long-term harmonization of radiometric data.