Development of a weakly-coupled atmosphere-ocean data assimilation system for the KIM coupled model

It has long been known that coupling between the various Earth system components (the ocean, atmosphere, sea ice, and land) produces improved forecasts on seasonal and longer time scales (Neelin et al. 1994). KIAPS (Korea Institute of Atmospheric Prediction Systems) has developed a global atmospheric model called KIM (Korean Integrated Model) that has been running operationally at KMA (Korea Meteorological Administration) since April 2020, with deterministic analyses produced by the KIAPS hybrid-4DEnVar system, and ensemble analyses produced by an LETKF. KIAPS is now developing a coupled atmosphere-land-ocean-sea ice model aimed at extended-range forecasts, using NEMO as the ocean model. Here, we describe our progress in developing an ocean data assimilation (DA) component for this new coupled model.
When atmospheric and ocean initial conditions are produced by independent DA systems, inconsistencies at the ocean surface can lead to imbalances in the coupled model (Lea et al., 2015). Therefore, we are developing a weakly-coupled atmosphere-ocean DA system to provide more balanced initial conditions. After investigating possible DA cycling strategies – including issues such as assimilation window lengths, observation cut-off times, and use of “catch-up” cycles – we decided to develop an atmosphere-ocean DA system where the ocean DA system uses the same cycling strategy as the atmospheric DA system, with 6-hour DA windows and a 6-hour cycle.
KMA operates an ocean-only DA system to initialize ocean states for seasonal forecasts, which are based on a coupled model that combines the Met Office Unified Model (UM) atmosphere with an older version of the NEMO ocean model, and a different sea ice model (CICE rather than SI3) to that used by the KIM coupled model. The DA system is based on NEMOVAR, using a 3DVar-FGAT approach with 24-hour DA windows. We are now adapting this system to produce an ocean-only DA system that will eventually be coupled to the KIM atmospheric DA system. The key steps are:
1. Change the surface forcing model from the UM to KIM.
2. Change to the 6-hour DA windows used by the atmospheric DA system.
3. Upgrade to the NEMO version and sea ice model used by the new KIM coupled model.
We will present results illustrating the impact of these changes, and then summarize our progress in coupling the new ocean DA system with the existing KIM atmospheric DA system, within a single NWP suite.