Poster Title	Quality Control of the CICE Consortium Sea Ice Model
Authors	Andrew Roberts, Elizabeth Hunke (Unlicensed), Richard Allard, David Bailey, Anthony Craig, Jean-François Lemieux, Matthew Turner
First Author	Andrew Roberts
Session Type	E3SM/Integrated Session
Session ID	I3
Submission Type	Poster
Group	Ocean/Ice
Experiment	all
Poster Link

Abstract

The CICE Consortium is leveraging sea ice model development across the Department of Energy, Navy, National Center for Atmospheric Research, Environment Canada, and NOAA for scientific research applications and medium-range to centennial prediction. As part of this work, we have devised quality control procedures to maintain the integrity of the dynamical core and column physics package - Icepack - within the Los Alamos Sea Ice Model, CICE. This work is designed to help facilitate contributions to CICE from within the Consortium as well as from groups using our open software development environment beyond the Consortium. We present a quality control framework for CICE as described in a recent paper published in the Philosophical Transactions of the Royal Society. Using results from five coupled and uncoupled configurations of CICE, we have devised quality control methods that exploit common statistical properties of sea ice thickness, and test for significant changes in model results in a computationally efficient manner. New additions and changes to CICE are graded into four categories, ranging from bit-for-bit amendments to substantial, answer-changing upgrades. These modifications are assessed using criteria that account for the high level of autocorrelation in sea ice time series, along with a quadratic skill metric that searches for hemispheric changes in model answers across an array of different CICE configurations. These metrics also provide objective guidance for assessing new physical representations and code functionality. An essential outcome of this work is that it facilitates early identification of frailties in new model physics and changes to existing CICE configurations ahead of their implementation in coupled frameworks including E3SM. This approach is computationally efficient by way of early detection of component model problems that may otherwise go undetected until implementation in coupled model configurations, and it serves as an example for a more general strategy to manage community codes used in E3SM.

Browser not supported