Abstract: Current speed is one of the most fundamental physical variables in the ocean, and its observation constitutes the basis for studies of oceanic dynamical processes. Conventional acoustic current meters suffer from observational blind zones, and their measurement quality is strongly affected by the concentration of suspended particles in the water column. In many environments—particularly in regions close to the seafloor—current observations are therefore subject to significant limitations. In this study, high-resolution temperature measurements obtained from a mooring deployed near the Dongsha Islands from 19 to 24 June 2025 were used to infer current speed at approximately 1 and 23 m above the seabed based on the maximum cross-correlation method. By performing correlation analyses with varying window sizes ranging from 0.5 to 240 minutes and comparing the inferred speed with those measured by an acoustic Doppler current profiler (ADCP), we demonstrate that the inferred current speed is sensitive to the window size. When the window size is set between 2 and 60 minutes, the speed derived from temperature cross-correlation agree well with the measurements, yielding a correlation coefficient of up to 0.67 and a root mean square error (RMSE) of approximately 0.02 m·s^-1. This study provides a new methodological perspective for measuring ocean current.

Key words: speed estimation, maximum cross-correlation, temperature observation, window size