> Much of the data on ocean temperatures currently relies on the Argo array — robotic devices that float at different depths, surfacing roughly every 10 days to transmit readings to satellites. There are about 3,800 such pieces of equipment in waters around the globe that provide the publicly available information...
> By comparison, Resplandy and Keeling calculated heat based on the amount of oxygen and carbon dioxide rising off the ocean. Filling round glass flasks with air from research stations in the Canadian Arctic, Tasmania and La Jolla, San Diego, researchers analyzed the samples to determine the aggregate temperature of the ocean.
On what grounds is three samples of a novel measurement preferable to a few thousand direct temperature measurements? I mean in terms of reliability, not ability to generate clicks.