Computational Performance Analysis of JEDI-based Sea Ice Ocean Coupled Assimilation (SOCA) System: CICE6

Under joint effort for Data assimilation Integration (JEDI) at JCSDA, the MOM6 and CICE5 models are utilized for the SOCA data assimilation system development. To update the sea ice modeling component of the JEDI-SOCA system, we have implemented the JEDI-based CICE6 model interfaces and focus on initializing the CICE6 model with various satellite-observed sea ice thickness and fraction data in the data assimilation framework of the JEDI-SOCA system. Preliminary 3D-Var data assimilation experiment result for the use of ICESat and CryoSat sea ice thickness and fraction data is presented with an overview of the JEDI-SOCA system structure. Along with the analysis outcome of the JEDI-SOCA system, computational efficiency is another key requirement for the operational application at NOAA-NCEP-EMC. We analyzed the computational performance of the JEDI-SOCA system with the TAU performance analysis toolkit. Different options of TAU profiling methods such as event-based sampling and source code instrumentation methods were used in the JEDI-SOCA performance evaluation. In this study, detail computational cost analysis results are presented for each of model forecasting, observation operator, and data assimilation solution steps of the JEDI-SOCA system.