Dynamic-programming-based subwaveform retracking with along-track constraints for inland river water level retrieval from Sentinel-3 altimetry

Sentinel-3 radar altimetry provides an important data source for inland water level monitoring, yet accurate retrieval over rivers remains difficult because contaminated waveforms complicate subwaveform identification. This study proposes a dynamic-programming-based subwaveform retracking (DPSR) method, which transforms conventional footprint-wise retracking into an along-track global optimization problem, enabling the selection of physically consistent subwaveforms from contaminated observations for river water level retrieval. DPSR is evaluated over the Amazon basin with 78 gauge-virtual station (VS) pairs and the Poyang Lake basin with 39 gauge-VS pairs, and is compared with several existing retracking methods. The results indicate that DPSR outperforms all benchmark methods, yielding a median RMSE of 0.41 m, which decreases by 0.02–1.2 m compared with other methods, while improving data availability by about 39% relative to benchmark retrackers. Moreover, DPSR performs particularly well in narrow rivers. The proposed method highlights the value of integrating spatial constraints into subwaveform interpretation and offers a robust solution for satellite altimetry retrieval of river water levels in complex environments.