top of page
  • Writer's pictureXia Lin

A SCIENTIFIC PAPER MADE EASY: SITool – a new evaluation tool for large-scale sea ice simulations

Updated: Feb 23, 2023

Are you interested in polar sea ice simulation and prediction? Would you like to evaluate your sea ice model simulations smoothly and efficiently? Do you plan to use python for your research work? Have a look at the SITool we developed at UCLouvain!


WHY SITOOL?


Most regional and global climate models now include an interactive sea ice model, reflecting the reality that sea ice plays a fundamental role in the polar environment, by influencing air–ice and ice–sea exchange, atmospheric and oceanic processes, and climate change. However, sea ice projections and evaluations are still not systematic, and, to date, no tool allows precise tracking of sea ice model performance through time from one version to the next.


Figure 1. Schematic overview of SITool (v1.0) and its application to the CMI6 OMIP model evaluation.


WHAT IS SITOOL?


The Sea Ice Evaluation Tool (SITool) is a performance metrics and diagnostics tool. It is developed to evaluate the skill of Arctic and Antarctic model reconstructions of sea ice. The general idea of SITool is described in Fig. 1. We applied it to evaluate the performance of Arctic and Antarctic historical sea ice simulations under the experimental protocols of the Coupled Model Intercomparison project phase 6-Ocean Model Intercomparison Project (CMIP6-OMIP).


Figure 2. The ice concentration metrics of 14 model outputs and model mean compared to satellite-based observational data (a) NSIDC-0051 and (b) OSI-450. The six columns correspond to model performance metrics on the mean state, standard deviation (SD Ano), and trend (Trend Ano) of monthly anomalies of the Arctic and Antarctic ice concentration during 1980–2007. Lower values indicate better skill.


SITool is written in Python andconsists of well-documented functions used to derive various sea ice metrics and diagnostics on the ice concentration, extent, edge, thickness, snow depth and ice drift. These sea ice metrics give a detailed view of sea ice state and highlight major deficiencies in the sea ice simulation. Take the ice concentration metric for example (Fig. 2), we can easily identify which model is better on the mean state, variability and trend of ice concentration simulations.


We also provide supporting maps (such as Fig. 3) to help understand why these metrics vary from one data set to another. The SITool is useful to describe inter-model differences quantitatively and to help teams managing various versions of a sea ice model, detecting bugs in newly developed versions, or tracking the time evolution of model performance.


Figure 3. The 1980–2007 September mean Arctic ice concentration differences between OSI-450/model outputs and the NSIDC-0051 data (colors), and contours of 15 % concentration of the NSIDC-0051 data (green lines) and OSI-450/model outputs (magenta lines).


HOW TO GET STARTED WITH SITOOL?


The source code of SITool (v1.0) is developed fully based on freely available Python packages and libraries and is released on a GitHub repository available at my GitHub. Here you can find a user guide providing technical information for using this tool.


Interested in the full details of the idea and the results? Check out the paper published in Geoscientific Model Development.


Written by Xia Lin



88 views0 comments

Comments


bottom of page