River Surface Velocity Estimation Using UAS-Borne Doppler Radar and Continuous Wavelet Transform

Reliable estimation of river surface velocity is essential for hydrological monitoring, flood forecasting, and water resource management, yet conventional in-situ measurements remain spatially limited and operationally demanding. Unmanned aerial systems (UAS) equipped with Doppler radar provide a promising remote sensing solution for rapid, contactless velocity measurement. However, mixed ambient wind and propellerinduced airflow (propwash) introduce significant distortion in radar returns, leading to composite velocity estimates that mix true river flow with locally induced air motion. To address this challenge, we develop a bidirectional observation framework for UAS-borne Doppler radar measurements. The same surface footprint is observed from two opposing viewing directions, and Doppler spectra from each acquisition are analyzed using the continuous wavelet transform (CWT) to resolve mixed velocity components. An analytical integration of the bidirectional measurements is then applied to remove the propwash-induced interference, enabling retrieval of the true river surface velocity under light wind conditions. The proposed approach was evaluated using field experiments conducted across more than 50 cross-sections in four rivers, spanning a wide velocity range from 20 to 250 cm/s. Retrieved velocities were validated against independent in-situ measurements obtained with an OTT-MFPro current meter, acoustic Doppler current profiler (ADCP), and flow tracker. Results demonstrate strong agreement with reference observations, yielding normalized root mean square errors below 20%. These results highlight the capability of UASborne Doppler radar combined with bidirectional measurements and wavelet-based spectral analysis to provide accurate and efficient river surface velocity retrieval, offering a scalable remote sensing approach for high-resolution river monitoring.