Near-Real-Time Inverse Current and Wave Estimation from Multiple Sensors

This paper presents an ANN-based methodology to inversely estimate the significant wave height and near-surface current velocities using field data collected from a buoy (TABS-K) in the Gulf of Mexico. The buoy integrates multiple sensor systems—including an Acoustic Doppler Current Profiler (ADCP), directional current sensor, compass, and inertial measurement unit (IMU)—to capture a variety of signals such as heading, pitch, roll, and accelerations. Since each sensor operates with different sampling intervals and occasionally yields missing data, the raw measurements are first processed on an hourly basis, ensuring consistency and quality control. The proposed approach focuses on predicting the significant wave height and near-surface currents. To improve the estimation accuracy of significant wave height (Hs), the measured accelerations and angular velocities from the IMU are converted into displacement signals via frequency-domain filtering, removing high-frequency noise and low-frequency drift. The proposed ANN model takes these preprocessed signals as inputs and shows a marked improvement in capturing the actual wave conditions, particularly for moderate sea states (Hs < 3 m). In addition, inversely estimating the ocean-surface current velocity using roll, pitch, heading, and other derived variables demonstrates that relatively inexpensive floater-mounted motion sensors can reproduce meaningful current velocity data. Overall, the results underscore the feasibility of near-real-time inverse estimation of the significant wave height and near-surface current velocity by combining cost-effective motion sensors, systematic data preprocessing, and robust ANN-based modeling. This approach promises broader accessibility for oceanographic monitoring, offering an efficient alternative when direct measurement instruments are unavailable or cost-prohibitive.