Initialization of Upper Ocean Temperature towards Forecast Improvements in GFSv17

Upper ocean temperature plays a critical role in the coupling of atmospheric and oceanic processes in numerical weather prediction (NWP). For instance, sea surface temperature (SST) is used to compute air-sea heat fluxes that can influence the atmospheric circulation. Since current NOAA operations do not include an ocean model, SST is prescribed as the persistence of the analysis with a correction for seasonality, and this has led to large differences between the forecasts and observed SST. The addition of an ocean model has been shown to improve the state estimate of SST which in turn further improves the atmosphere, e.g., by reducing the observation-minus-background (OMB) statistics for SST-sensitive radiance channels from AVHRR and other satellite-based observing instruments.

The next major update to the Global Forecast System (version 17) (GFSv17) will include ocean modeling and accompanying data assimilation (DA) schemes for marine (ocean and sea ice) observations. To support this effort, this presentation highlights the expected impact of marine DA on global forecasts. Three experiments were conducted for July-August 2021: the first has a configuration similar to current NOAA operations (atmosphere model with atmospheric DA); the second has a model configuration similar to GFSv17 (atmosphere, ocean, and sea ice models) but with atmospheric DA only; and the third is identical to the second except that marine DA is also included. Comparisons of forecasts out to two weeks are assessed, as well as the impact of assimilated SST observations from geostationary (GEO) and low-Earth orbit (LEO) satellites on both the ocean and atmosphere. The results emphasize the importance of marine DA to the subseasonal prediction of SST and near-surface atmospheric fields.