GFDL's SPEAR prediction system: MOM6 initialization and bias correction with data assimilation

Experimental operational seasonal predictions are successfully initialized in GFDL's new SPEAR prediction system, which is based on the newest generation of GFDL models including AM4 atmosphere model, LM4 land model, MOM6 ocean model and SIS2 sea ice model. The new initialization scheme is based on coupled model analysis with improved ocean data assimilation (ODA) in SPEAR's MOM6 ocean component and optional atmosphere. The new ODA system makes use of efficient parallelization and MOM6 library, and its speed enables an iterative approach to improve coupled model initialization. A new ocean tendency adjustment (OTA) scheme is explored with the climatological increments of ocean state variables from ODA in the ARGO era. OTA can serve as bias correction for both the reanalysis and prediction. Preliminary results of coupled seasonal predictions with OTA show reduced model drift and improved ENSO anomaly prediction. The use of OTA in control and transient coupled model simulations is also explored.