Real-time Ocean Monitoring at the Joint Center for Satellite Data Assimilation: a Testbed for Ice/Ocean Data Assimilation Development and Evaluation.

Travis Cole Sluka, Joint Center for Satellite Data Assimilation, Boulder, United States, Guillaume Vernires, NOAA/NWS/NCEP/EMC, College Park, United States and Rahul Mahajan, NASA/GMAO, Greenbelt, MD, United States
Abstract:
The multi-agency partnership Joint Center for Satellite Data Assimilation (JCSDA) is working with NOAA and NASA to develop improved ocean data assimilation methods for next-generation seasonal to sub-seasonal prediction systems. These DA systems are to be based on JCSDA’s Joint Effort for Data Assimilation Integration (JEDI), a unified DA framework designed to work across various domains and models. As a demonstration and development testbed for the ocean component, JCSDA is creating a real-time ocean monitoring system updated daily. An expansive set of observations are ingested including satellite sea surface temperature (from VIIRS, AVHRR, MODIS, AHI, ABI), sea surface salinity (SMAP), absolute dynamic topography (various satellite altimeters), and insitu temperature and salinity profiles (ARGO, XBT, CTD, ...).

The JCSDA real-time ocean monitoring system leverages ideas commonly used in the software development community such as automated unit testing, agile development, continuous deployment, and use of cloud computing. These practices allow for rapid deployment of new code to the baseline version, providing immediate turnaround of performance metrics and quickly exposing unforeseen issues with “operational” cycling. The system is based on the GFDL MOM6 ocean model currently at 1 degree resolution. Planned improvements will slowly increase the capabilities of the system, increasing resolution to ¼ degree and adding a 1/25th degree regional domain over the Gulf of Mexico. The flexibility provided by the JEDI system will allow for rapid transition to more sophisticated DA methods: 3DVAR is currently being used but the real-time ocean monitoring will expand to 3DVAR-FGAT, EnKF, hybrid-4DEnVAR, in the near future.