Mesozooplankton Abundance, Biomass, and Diel Vertical Migration via Passive Ocean Color Retrievals of the Particle Size Distribution (PSD).

Data description Monthly global merged passive ocean color satellite retrievals (spanning September 1997 to December 2024) of the following variables (at 0.25 degree spatial resolution): mesozooplankton numerical abundance, biomass, and diel vertical migration; particle size distribution power-law slope (not used in the derivation of the mesozooplankton data, included here for completeness), particle size distribution scaling factor N0 (differential numerical concentration at reference diameter of 2 micrometers) - original and tuned versions. The tuned version is used in the derivation of mesozooplankton data. Associated estimates of partial uncertainty are also included, as well as relevant ancillary data such as the tuning coefficients for N0, and latitude/longitude grids. For details on the PSD data retrieval algorithm and N0 tuning, see Kostadinov et al. (2023; https://doi.org/10.5194/os-19-703-2023) and associated manuscrupt assets (data set and scientific code linked therein). Extensive metadata is included in the netCDF files. Also included is browse imagery - these maps are produced using the original resolution 4 km data. The maps use the 'cividis' colormap (Nuñez et al., 2018; https://doi.org/10.1371/journal.pone.0199239; https://doi.org/10.1371/journal.pone.0199239.s002). The MATLAB(R) code adapted here for 'cividis' is due to Ed Hawkins (http://www.met.reading.ac.uk/%7Eed/viridis.m). The data set is based on input monthly ocean color remote-sensing reflectance (Rrs) from the ESA/PML OC-CCI v6.0 data set (Sathyendranath et al., 2019; https://doi.org/10.3390/s19194285). The data set citation specific to v6.0 is Sathyendranath et al. (2023), https://doi.org/10.5285/5011d22aae5a4671b0cbc7d05c56c4f0 Modeling, analysis and data production (including climatologies and uncertainties) were performed using the original 4 km resolution data set in sinusoidal projection. The data set provided here (including the original uncertainty imagery) has been re-projected to equidistant cylindrical projection (unprojected lat/lon) and downsampled to 0.25 degree spatial resolution. Important Notes: The mesozooplankton variables are computed using a global constant value of the PSD power-law slope of 4.0. The variable retrieved PSD slopes are included for completeness. The extrapolation of the PSD is discussed, for example, in Jennings et al. (2008; https://doi.org/10.1098/rspb.2008.0192). See also Sheldon et al. (1972; https://doi.org/10.4319/lo.1972.17.3.0327), Hatton et al. (2021; https://doi.org/10.1126/sciadv.abh3732) and Tekwa et al. (2023; https://doi.org/10.1371/journal.pone.028302). Note that the retrievals of mesozooplankton variables presented here are based on theoretical ecosystem size structure assumptions and the PSD retrieval based on phytoplankton modeling (Kostadinov et al., 2023; https://doi.org/10.5194/os-19-703-2023). A time lag can thus be expected between abundances of phytoplankton to which the PSD retrieval is assumed to be sensitive here, and the resulting mesozooplankton variables. Preliminary investigations indicate that certain locations can have time lag as measured against the Behrenfeld et al. (2019; https://doi.org/10.1038/s41586-019-1796-9) satellite CALIOP lidar DVM data (e.g. 2 months at BATS), but no consistent time lag was found globally. No time lag has been applied here at the monthly scale. In addition, preliminary validation results against the in-situ BATS zooplankton data (https://simonscmap.com/catalog/datasets/BATS_Zooplankton_Biomass and Steinberg and Cope (2025; https://doi.org/10.26008/1912/bco-dmo.881861.5)) are quite satisfactory, but further global validation/comparison with these and additional data sets (MAREDAT mesozooplankton (O'Brien and Moriarty (2013; https://doi.org/10.5194/essd-5-45-2013)), satellite CALIOP lidar DVM (Behrenfeld et al. (2019; https://doi.org/10.1038/s41586-019-1796-9)) is ongoing/pending. Scientific Code Scientific code in MATLAB(R) is also included. Function/script and variable names are mostly self-explanatory. The operational code to generate the mesozooplankton data needs only MATLBA(R) and no additional toolboxes. Also included is code for the optimization used to compute the empirical correction(tuning) for the N0 parameter, as well as code to compute climatologies. This additional code also needs the following MATLAB(R) toolboxes: Statistics and Machine Learning Toolbox (R), Optimization Toolbox (R), and Global Optimization Toolbox (R). Code to generate the PSD products themselves is provided elsewhere (Kostadinov et al., 2022): https://doi.org/10.5281/zenodo.6354653 Corresponding author Dr. Tihomir S. Kostadinov ([e-mail]) Acknowledgements Funding: NASA grant #80NSSC22K0284 and NASA grant #80NSSC19K0297. Support from California State University San Marcos is also acknowledged. Mike Behrenfeld and his team (Robert O'Malley in particular) are acknowledged for useful discussions and sharing lidar DVM data and offering support with the data. Camila Serra-Pompei and Nils Haentjens are acknowledged for the very useful discussions. We also acknowledge ESA, PML and the OC-CCI team for creating the OC-CCI v6.0 data set, and we thank all data providers and algorithm developers/providers, individuals and institutions contributing to the input satellite data sets (NASA, NOAA, ESA and/or EUMETSAT) and the creation of the OC-CCI v6.0 merged satellite data set. DigitalGlobe(R) and predecessor companies GeoEye, Inc.(R) and ORBIMAGE(R) are also acknowledged for their role in SeaWiFS data acquisition. We acknowledge the BATS zooplankton and MAREDAT zooplankton in-situ data set providers, contributors and authors, as well as the satellite CALIOP lidar DVM data providers, contributors and authors. Links to relevant manuscripts/data sets are provided above. Modeling, analysis, data processing and computations, as well as visualization and file production were performed in MATLAB(R). Erik Fields, ESA, BEAM (Brockmann Consult GmbH) and NASA are acknowledged for the re-projection algorithm. We thank Alois Schlögl for ideas how to deal with NaN values and downsampling using 2D convolution. Disclaimer The views and opinions of authors expressed here do not necessarily state or reflect those of the U.S. Government or NASA. The views and opinions of authors expressed here do not necessarily state or reflect those of the author's institutions. This is a highly experimental, research data set. No warranty or guarantee of any kind is given, express or implied, of fitness for any purpose or of any level of accuracy. Note the large uncertainties in some of the retrievals, and note that the provided uncertainties are partial. Under no circumstances shall the authors or their institutions be liable to anyone for direct, indirect, incidental, consequential, special, exemplary, or any other kind of damages (however caused and on any theory of liability, and including damages incurred by third parties), arising from or relating to this data set, or user's use, inability to use, or misuse of the data set, or errors of the data set. The data set is not guaranteed to be error-free, and is not meant to be used in any mission-critical applications. Use at your own risk. NASA, ESA or other institutions have not formally or informally reviewed these data.