Elena Ceballos Romero , Ken Buesseler, Erik Fields, Rainer Kiko, Margaret Estapa , Lee Karp-Boss, Samantha Clevenger, Laetitia Drago, David Siegel

The biological carbon pump plays a crucial role in the global carbon cycle, particularly through sinking particles carrying carbon to deep waters. The Underwater Vision Profilers (UVP) is widely used for studying particle properties. UVP-based particulate organic carbon (POC) flux is typically derived from particle size distributions (PSDs) assuming size dependent sinking rates and carbon content. This approach, the “classic UVP method”, calibrates PSD-based flux against sediment trap flux data that are not necessarily co-located in space or time. We put forth a “modified UVP method” that combines a large data set of UVP measurements calibrated against POC flux from co-located and simultaneously collected sediment traps and thorium-234 measurements. Data were collected in the North Pacific (50°N, 145°W, August 2018) and the North Atlantic (49°N, 16.5°W, May 2021) as part of EXPORTS (EXport Processes in the Ocean from RemoTe Sensing), covering a wide range of environmental conditions. We find that our modified UVP methods explain 80% of the variance in POC flux when applied across sites, where flux values vary over orders of magnitude. However, the method fails to account for smaller flux variations within a single site or across depths. Reasons include undersampling rare large particles, mismatch in time and spatial scales of UVPs calibrated against fluxes in traps and 234Th, and difficulties in interpreting particle stock and flux changes within non-steady state conditions. To use UVP as a high-resolution POC flux tool, it is recommended not to rely on a few profiles for calibration.