Development of Ocean Reanalysis for Coupled Physical-Biogeochemical Prediction Using PDAF with GFDL-ESM4

Earth System Models (ESMs) are sophisticated tools for simulating the global ocean and its ecosystems through tightly coupled physical and biogeochemical (BGC) processes. A key factor in improving ESM predictive skill is data assimilation, which incorporates observations to produce reanalyses used as initial conditions. Despite its crucial role, ocean data assimilation is challenged by limited BGC observations and physically unrealistic diapycnal mixing, which hinder progress in coupled physical-biogeochemical prediction. Here we develop an ocean reanalysis production system using the Parallel Data Assimilation Framework (PDAF) within GFDL-ESM4 to initialize coupled physical-biogeochemical prediction. The system produces a 27-year reanalysis (1991–2017) by assimilating only physical observations (e.g., temperature and salinity) and does not exhibit spurious diapycnal mixing. Comparisons with observations and existing reconstructions indicate that our reanalysis similarly represents the physical mean state and climate variability. Notably, the reanalyzed BGC variables show consistency with observations—carbon flux exhibiting similar spatial patterns, and chlorophyll anomalies showing significant correlations (~0.67) across several ocean regions—even without assimilating BGC data. These findings highlight the potential of our developed system to initialize coupled physical-biogeochemical predictions.