A cost-effective data assimilation system for improving ROMS simulations in the Southwest Atlantic

Integrating the low-cost REMO Ocean Data Assimilation System (RODAS) with the Regional Ocean Modeling System (ROMS) reduced sea-surface temperature root-mean-square deviation by 70% relative to a free-running simulation in a four-year Southwest Atlantic experiment. The study reports the first adaptation of RODAS, an Ensemble Optimal Interpolation-based system originally developed for the HYCOM model, for assimilating sea-surface temperature observations in ROMS. The evaluated application focused on short-term ocean hindcasts rather than forecasts across other regions or forecasting systems.

The researchers compared two four-year simulations. The RODAS_SST experiment assimilated sea-surface temperature observations, while the Free experiment ran without observation assimilation. In addition to its 70% reduction in sea-surface temperature root-mean-square deviation relative to Free, RODAS_SST produced a 55% reduction compared with 72-hour hindcasts generated from previous analysis cycles. Against the GLORYS reanalysis, the assimilated simulation also had a 64% reduction in sea-surface temperature root-mean-square deviation. These results show improved performance for the tested regional-model configurations, not a general guarantee that the same gains will occur elsewhere.

Ocean heat-content estimates also improved when sea-surface temperature observations were assimilated. The mean absolute error decreased by 15% during summer and by 25% in winter. Together with the sea-surface temperature results, these findings indicate that the assimilation system improved the modeled thermal structure in the study domain and produced more accurate heat-content estimates under the evaluated conditions. The magnitude of the improvement varied by season, with the larger reported reduction in mean absolute error occurring during winter.

Comparisons with Argo profiles showed that assimilation generally improved surface and subsurface thermal structure, but some conditions deteriorated near the thermocline. The researchers identified the representation of multivariate covariance in the EnOI scheme as an area requiring further improvement, demonstrating that reduced sea-surface temperature error did not translate into uniform gains throughout the water column. Planned extensions include assimilating Argo temperature and salinity profiles, sea-level anomaly data, and observations from the SWOT satellite. The authors also plan an open release of the code, with the stated aim of widening access to ocean data assimilation and supporting robust performance on limited computational resources.

Subject of Research: Evaluation of low-cost sea-surface temperature assimilation in Southwest Atlantic ROMS simulations

Article Title: A cost-effective data assimilation system for improving ROMS simulations in the Southwest Atlantic

Article References: Dorfschäfer, G. S., Tanajura, C. A. S., De Paula, T. P., Costa, F. B., & Pereira, J. (2026). A cost-effective data assimilation system for improving ROMS simulations in the Southwest Atlantic. PLOS One, 21(10), e0358351. https://doi.org/10.1371/journal.pone.0358351

Image Credits: AI Generated

DOI: 10.1371/journal.pone.0358351

Keywords: RODAS, ROMS, ocean modeling, data assimilation, sea-surface temperature, Southwest Atlantic, ocean heat content, Argo profiles, Ensemble Optimal Interpolation, regional simulations, cost-effective, data