Evaluating ocean forecasting systems

OceanBench is a benchmarking tool to evaluate ocean forecasting systems against reference ocean analysis datasets (such as 2024 GLORYS reanalysis and GLO12 analysis) as well as observations.

Evaluating ocean forecast performance is a complex task, as different users have varying priorities. Some may focus on accurate predictions of currents, while others prioritize sea level or salinity forecasts. To accommodate these diverse needs, OceanBench provides a comprehensive set of metrics. Please keep in mind that these metrics capture key aspects of forecast quality rather than the complete picture.

Want to evaluate your system or get involved in OceanBench? Check out the GitHub project!

Comparison to reanalysis

Absolute RMSE scores against 2024 GLORYS reanalysis. These are the deterministic scores for the forecasted variables and physically-consistent diagnostic variables. The colors denote % difference to the GLO12 baseline.

Comparison to analysis

Absolute RMSE scores against 2024 GLO12 analysis. These are the deterministic scores for the forecasted variables and physically-consistent diagnostic variables. The colors denote % difference to the GLO12 baseline.

Comparison to observations

Absolute RMSE scores against 2024 observations. These are the deterministic scores for the forecasted variables and physically-consistent diagnostic variables. The colors denote % difference to the GLO12 baseline.

Implemented by Mercator Ocean.