Selective extended Kalman filter for enhancing low-velocity flow measurements in large rivers

Introduction Reliable velocity measurement under low-flow conditions remains a major challenge in automatic discharge monitoring using Horizontal Acoustic Doppler Current Profilers (H-ADCPs). Accurate measurements under such conditions are essential for improving the reliability of real-time river monitoring systems.Objectives This study aims to develop and evaluate a post-processing framework that combines interquartile range (IQR)-based outlier removal with selective application of the extended Kalman filter (EKF) to enhance the quality and reliability of H-ADCP time-series velocity data, particularly under low-velocity conditions.Method Field measurements were conducted at three bridge-based monitoring sites on the Han and Nakdong Rivers (Gwangjin, Yeoju, and Hoguk) in October 2024. Following IQR-based outlier removal, the EKF was selectively applied to segments where the coefficient of variation (CV) exceeded 0.1, indicating unstable velocity fluctuations. In the low-velocity regime (<0.10 m/s), the CV decreased by an average of 22.8% and a maximum of 33.7%, while higher-velocity segments (≥0.10 m/s) improved by up to 16.7%. Comparison with official reference velocities confirmed a reduction in root mean square error (RMSE) and an increase in the coefficient of determination (R2) at most sites. The proposed framework effectively suppresses short-term noise without distorting long-term flow patterns, demonstrating its practical applicability for improving H-ADCP measurement reliability in real-time river monitoring systems.