Work

Optical Flow-based temporal displacement correction for Time-Consistent Multi-Satellite wind field reconstruction

Year: 2026

Type: article

Abstract: High-resolution ocean observations and numerical simulations are increasingly in demand. However, conventional multi-source remote sensing wind products often struggle to capture the spatiotemporal consistency and dynamic variability required for mesoscale and submesoscale oceanic studies. This paper proposes an Optical Flow-based Temporal Displacement Correction (OF-TDC) method to improve the temporal alignment and mitigate structural misalignment in global multi-satellite wind fields (MSWF). The method treats satellite wind fields as sequential images while utilizing ERA5 winds as a dynamic reference. To ensure consistent input quality for this process, a robust bias correction scheme combining hierarchical Bayesian shrinkage and dual-reference LOESS smoothing is introduced. OF-TDC then estimates advection trajectories of MSWF over time and projects satellite wind fields observed at different times to a unified reference time to complete temporal registration. Evaluations show robust performance under moderate winds of 4 to 20 m/s and effective correction over medium to long timescales, with error standard deviation (STD) reductions of 1° in direction and 0.03 m/s in speed. Tropical cyclone analysis shows error STD improvements of up to 0.19 m/s and 2.24°, respectively, indicating that stronger, compact cyclones enable more accurate OF tracking. Validations against independent GNSS-R data and heavy rainfall confirm significant robustness. OF-TDC enhances MSWF temporal consistency, especially in structured, moderate-evolution environments, facilitating integrated wind and air-sea interaction studies.

Source: International Journal of Applied Earth Observation and Geoinformation

DOI: 10.1016/j.jag.2026.105148