Estimating Argo Float Trajectories Under Ice.

Since the Argo program began, 568 floats returned almost 31,000 profiles, at high‐southern latitudes, with no measured position. These data are either disseminated with positions linearly interpolated between known positions, or with no geographic positions. Here, we present a simple method for estimating unknown Argo float trajectories. We try to identify trajectories that approximately follow contours along three different properties: potential vorticity (f/H), sea‐level, and density at 1,000 m. No single property‐constraint can be used to estimate trajectories for all position‐gaps. Each constraint fails for 9%–18% of gaps, where no continuous contour between the end‐points exists. But all constraints fail for the same position‐gap, for fewer than 1% of cases. For a given position‐gap, when a trajectory is identified using two or three different constraints, we select the shortest trajectory to be used to "fill the gap". This selection process could be performed better by an Expert Operator, inspecting each estimated trajectory, and selecting the trajectory that is most consistent with a priori knowledge of the circulation in the vicinity of the position‐gap. Nonetheless, using the objective metric for selection, we find that 41.2% of position‐gaps use the f/H‐constraint, 32.1% use density, and 25.8% use sea‐level. We assess the estimated trajectories for consistency, by comparing bottom depths beneath trajectories to the deepest measurements in each profile. We find inconsistencies for 11.6% of position‐gaps using our method, compared to 28.0% using linearly‐interpolated trajectories. Adoption of the estimated trajectories for measurements under ice may yield benefits to many applications. Plain Language Summary: Argo floats measure ocean properties between the surface and 2,000 m depth, providing an estimate of their position using satellite fixes when they're at the surface. When a float samples under ice, it cannot measure its position. When that happens, the float continues to sample, collecting profiles over time, and later sends data if and when it finally surfaces. While under ice, the geographic positions of the float's observations are unknown, making it hard for scientists and marine users to exploit the data. Here, we present a method for estimating the float's position, while under ice, by assuming that the float's path followed contours of some climatological ocean properties. The method won't always be accurate, because floats mostly move with ocean properties that change with time. But sometimes, we expect the new estimates will be accurate enough to make the data more useful. Key Points: Almost 31,000 Argo profiles have sampled under ice at high‐southern latitudes, and are disseminated without measured positionsWe present a method to estimate missing trajectories, constrained by multiple different oceanic propertiesNo single property works for all cases. But using multiple constraints, realistic trajectories are estimated for almost all position‐gaps [ABSTRACT FROM AUTHOR]