History
Development of the CICE model began at Los Alamos National Laboratory (LANL) in 1994 with a translation of the Semtner (1976) 3-layer thermodynamic code into Fortran 77. Development of the Elastic-Viscous-Plastic (EVP) sea ice dynamics model followed, which attracted the first community user for the code, the Naval Postgraduate School (NPS). CICE v1 was released in 1998 and initial attempts to validate the model ensued, in collaboration with researchers at the Naval Research Laboratory – Stennis Space Center (NRL-Stennis). The Naval Postgraduate School again supported the EVP model’s advance by implementing it in the Department of Energy (DOE)-supported Parallel Climate Model (PCM), based at the National Center for Atmospheric Research (NCAR). Development efforts for PCM and NCAR’s other climate model, CSM, later merged to become CCSM and adopted CICE v3 in 2002. A formal Memorandum of Understanding between LANL and NCAR formed the foundation for a strong and lasting collaborative relationship for sea ice model development, verification and validation, and scientific research. NOAA’s Geophysical Fluid Dynamics Laboratory (GFDL) later adopted EVP for use in their CM2 climate model, leading to its use in the National Centers for Environmental Prediction (NCEP) numerical weather prediction model CFS in 2011. The UK Met Office (UKMO) adopted CICE v3.1 in 2005 for use in its climate model and later for the full range of its “seamless prediction system” climate and weather prediction models. The coupling scheme used by UKMO is different than that used in other climate and operational models, requiring oversight for subsequent community contributions to the code to ensure that both coupling mechanisms continue to work properly. Canadian research groups began adopting the CICE model during the late 2000s, and in 2009 NRL-Stennis implemented CICE v4 in their Arctic Cap Nowcast-Forecast System (ACNFS) for operational use by the U. S. Navy and other federal centers. Following several years of validation and acceptance testing, ACNFS became operational in 2013, incorporating CICE v4.0. NASA also uses CICE as part of its GOES-S2S-1 seasonal-to-interannual forecast system.
Meanwhile, the Model for Prediction Across Scales (MPAS) framework for climate model components was developed, spurring discussions about an MPAS-seaice model in 2011. MPAS provides a framework of operators and code infrastructure for component model development on variable resolution, unstructured grids. A design document for MPAS-seaice was prepared in 2012, with funding for the project included in DOE’s Accelerated Climate Model for Energy (ACME) project beginning in 2014. Because MPAS-seaice utilizes an unstructured grid, the grid- dependent portions of CICE had to be recoded. Physical parameterizations that are limited to single grid-cell columns could be shared between the models, however, and these modules (the “column package”, now renamed “Icepack”) were rewritten to make them independent of the grid and other CICE infrastructure. CICE v5 was released in 2015 without the column package changes; a formal agreement between the ACME project and CICE’s community users enabled them to access and use the column-package version of the code prior to its release.
With the exception of the LANL-NCAR Memorandum of Understanding, collaborations among other members of the CICE community have been informal. Updated versions of CICE have been released every 2 to 3 years, enabling other research groups to utilize and develop the model. The collaborating institutions listed above and many of the other community research groups have contributed code (bug fixes, new parameterizations, infrastructure, etc.) back to the central CICE repository at LANL, coordinated and overseen by Elizabeth Hunke with support from DOE.
A more formal arrangement for sea ice modeling collaboration is needed for several reasons. First, significant multi-decadal changes in both the Arctic and the Antarctic require improved understanding of the role of sea ice in the earth system. CICE is also used to meet civil and military operational prediction demands and to provide seasonal to centennial climate predictions. Sea ice modeling efforts at DOE and in the larger community have benefitted greatly from many CICE modeling collaborations, a community success story that should be sustained. Finally, DOE’s shift to the MPAS framework for ocean and sea ice models, which led to creation of the column package (Icepack), opens up additional opportunities for other groups to utilize this portion of CICE in their own sea ice models. DOE is interested in sharing responsibility, coordination and support of community sea ice modeling efforts with the community, and is therefore responding to the community’s suggestion for the formation of a sea ice modeling consortium.