Dataset: Global reconstructions of particle biovolume, size distribution, and carbon export flux from the seasonal euphotic zone and maximum winter time mixed layer from particle profiles conducted during cruises from 2008 to 2020

Methodology:

This work is based on the compilation of over 6800 profiles of particulate matter observations from Underwater Vision Profilers (UVP5) (Rainer, 2021). The biovolume of the particle size distribution is calculated as the equivalent spherical volume of the particle size distribution  (PSD), by summing the product of particle counts time particle volume in each size class. The slope of the PSD is calculated assuming a power law distribution for the particle abundance, by linear least square fit of the log of particle counts vs. the log of particle size. These quantities are estimated from two different depth horizons, the mixed layer depth (MLD_Export.nc) and the euphotic zone depth (Euphotic_Export.nc). To convert sparse observations to a global climatology, we trained 100 ensembles of regressions trees (Random Forests, RF) to predict biovolume and slope based on their relationship to well-sampled physical and biogeochemical predictors.

We calculate the particle sinking speed and carbon content by combining PSD reconstruction (biovolume and slope) with an empirical relationship between particle size, carbon content and sinking speed, the parameters of which are optimized to match in situ particle flux observations (Bisson et al. 2018). The flux values are calculated as the sum of the PSD time the sinking carbon parameters, for each grid cell. The error for reconstructed quantities is given by the standard deviation of 100 independent realizations of the RF reconstructions.

Sampling and analytical procedures: 

This dataset contains a compilation of data from multiple sources.  A list of all datasets and the associated information, including cruise name, is included.

Observations of Particle size and biovolume are made via the UVP5 camera, which is lowered in the water column on a CTD rosette. Images are captured at up to 30 images per second while the instrument is lowered at 1m/s. Images are analyzed and the pixel size of each particle is translated into a particle size, and the abundance calculated, following the method described by Picheral et al. (2010). 


The data is compiled from multiple sources published and unpublished, and was accessed from EcoPart, the particle module of EcoTaxa https://ecotaxa.obs-vlfr.fr/part/ (Picheral et al., 2017).

Instruments: 

Observations of particle abundance and biovolume were made with the Underwater Vision Profiler, version 5 (UVP5).