A global surface drifter data set at hourly resolution

Abstract The surface drifting buoys, or drifters, of the Global Drifter Program (GDP) are predominantly tracked by the Argos positioning system, providing drifter locations with O (100 m) errors at nonuniform temporal intervals, with an average interval of 1.2 h since January 2005. This data set is thus a rich and global source of information on high‐frequency and small‐scale oceanic processes, yet is still relatively understudied because of the challenges associated with its large size and sampling characteristics. A methodology is described to produce a new high‐resolution global data set since 2005, consisting of drifter locations and velocities estimated at hourly intervals, along with their respective errors. Locations and velocities are obtained by modeling locally in time trajectories as a first‐order polynomial with coefficients obtained by maximizing a likelihood function. This function is derived by modeling the Argos location errors with t location‐scale probability distribution functions. The methodology is motivated by analyzing 82 drifters tracked contemporaneously by Argos and by the Global Positioning System, where the latter is assumed to provide true locations. A global spectral analysis of the velocity variance from the new data set reveals a sharply defined ridge of energy closely following the inertial frequency as a function of latitude, distinct energy peaks near diurnal and semidiurnal frequencies, as well as higher‐frequency peaks located near tidal harmonics as well as near replicates of the inertial frequency. Compared to the spectra that can be obtained using the standard 6‐hourly GDP product, the new data set contains up to 100% more spectral energy at some latitudes.