A Reconstructed dFe data product in the Ross Sea continental shelf

This dataset provides a monthly three-dimensional reconstruction of dissolved iron concentrations over the Ross Sea continental shelf for the period 2005–2019. The product was generated using an XGBoost machine-learning model trained with historical in situ dissolved iron observations and paired environmental variables. The trained relationship was subsequently applied to temporally and spatially continuous environmental fields from the MOM6 ocean model to reconstruct dissolved iron distributions across the study region. The reconstruction covers the Ross Sea continental shelf between approximately 69°S and 79°S and 162°E–202°E. The dataset contains monthly fields on multiple vertical levels extending from the surface to the deep water column, with the continental shelf defined using a bathymetric threshold of 1500 m.

The reconstructed product reproduces the major spatial and vertical patterns of dissolved iron observed in the Ross Sea, including relatively elevated concentrations in the western coastal region and north of Franklin Island, and widespread surface iron depletion across the central shelf during the growing season. The dataset was designed to support analyses of dissolved iron variability, vertical iron supply, ferricline–mixed-layer relationships, and their links to phytoplankton dynamics on seasonal and interannual timescales.