Yanfeng Wen, Peng Chen, Delu Pan
Oceanic chlorophyll-a (Chl-a) 3D distribution is critical for quantifying marine primary productivity, ecosystem dynamics, and the oceanic carbon cycle. However, existing global 3D Chl-a datasets have coarse spatial resolution (≥0.25°), limiting their ability to resolve mesoscale and sub-mesoscale processes. We present MEGO-C-M (Multi-model Ensemble for Global-Ocean Chlorophyll-a - Monthly Gridded Dataset), a high-resolution global 3D Chl-a dataset on a 0.0416° (~4.6 km) grid with monthly coverage (October 1997-November 2024). Derived via the MEGO-C machine-learning framework (trained on 144,936 quality-controlled BGC-Argo profiles and co-located satellite/physical reanalysis predictors), the model is applied to monthly predictor fields to generate gridded Chl-a across 27 vertical levels. Validation against independent BGC-Argo profiles shows strong agreement (R² = 0.818, RMSE = 0.254 log₁₀(mg·m-3), bias = -0.001 log₁₀(mg·m-3)). The dataset captures vertical Chl-a structures and large-scale seasonal variability, per-grid-cell ensemble spread. It is distributed in NetCDF format and assigned a digital object identifier (DOI) for persistent access: https://doi.org/10.57760/sciencedb.30693 .