PSD_PhytoC_v2021: Ocean Color Algorithm for the Retrieval of the Particle Size Distribution and Size-Partitioned Phytoplankton Carbon: Algorithm Development and Operational Code

Files needed for the operational application of the PSD/phyto C algorithm to OC-CCI v5.0 merged ocean color remote-sensing reflectance data are shown below (applied in the given order). This code was used to produce the associated OC-CCI v5.0-based PSD/phyto C data set Kostadinov et al. (2022; https://doi.org/10.1594/PANGAEA.939863): 1) generate_OCCCIv50_bbp_operational_v0.m 2) generate_OCCCIv50_PSD_from_SAM_Coated_2pop_operational_v0.m 3) generate_OCCCIv50_C_from_SAM_Coated_2pop_operational_v2.m The following code and data files are also needed to run the above operational code: a) IOP_inversion_vectorized.m, b) lg.m, c) save_for_parfor.m, d) SAM_median_Ems_2pop_Cis_SeaWiFS_550.mat, e) NoOptimization_21032021.mat, f) Mie_coated_EAP_MC_input_20191101T234748.mat Of these, files b and f are also used in the model development phase. The rest of the provided files are used in the algorithm development phase and are not needed for operational application. Code files names and brief code comments should help explain the code purpose and how to apply it. The order of code execution for model development is as follows: 1) generate_coatedMie_EAP_like_sim_input.m & generate_homogen_NAP_sim_input.m; 2) generate_MC_Qs_EAP.m & generate_MC_Qs_NAP.m; 3) prep_Qbbs_EAP.m & prep_Qbbs_NAP.m; 4) integrate_IOPs_from_bandavged_MC_Qs_and_PSD_2pop.m; and 5) generate_SAM_median_EMs_SeaWiFS.m The code is authored by T.S.Kostadinov, except the following listed files, and except for contributions elsewhere as noted in the code itself. - ZhangMie.m – authored by Xiaodong Zhang - betasw_nsw_zhh2009.m - authored by Xiaodong Zhang - 501nm_extended_e1701000.mat – authored by Stewart Bernard - IOP_inversion_vectorized.m – original code was written by H. Loisel and/or their research team. Transliterated to MATLAB® and vectorized by T.S. Kostadinov - generate_coatedMie_EAP_like_sim_input.m and generate_homogen_NAP_sim_input.m - written by T.S. Kostadinov, based on code by Lisl Robertson-Lain and Stewart Bernard - Stéphane Maritorena developed the band-shifting method Regarding file Monte_Carlo_Qbb_BandAvg_EAP.mat - the variable 'Qbb_bandavg' has been purposefully degraded to single precision here for storage reasons. In the actual model development, the double precision variable was used. If the single precision variable provided here is used, user results can differ. In addition, the raw hyperspectral Qbb values for each Monte Carlo run are not included here (very large files). The double precision band-averaged Qbb values and the raw hyperspectral Qbb are available from the author upon request, or can be reproduced with the code provided here. Additional MATLAB® Toolboxes required to run the code: Statistics and Machine Learning Toolbox®, Parallel Computing Toolbox®, Signal Processing Toolbox®, and Wavelet Toolbox®, of which only the first two are required for the operational code. The code was tested and used in production using mostly MATLAB® R2020b (some part using earlier versions), and it may not work properly on earlier versions. See the accompanying publication Kostadinov et al. 2023 (https://doi.org/10.5194/os-19-703-2023) for more details.