Large Scale Modelling of Sea Ice Thickness Distributions
A sea-ice forecast model has been developed at the Antarctic Climate and Ecosystems CRC for research into the requirements of an operational sea-ice forecasting system for Antarctica. The purpose of the model is to investigate ways to aid icebreaker navigation through sea ice. It is also intended as a tool for developing methods of combining satellite-observed sea ice cover with modelled sea-ice state. This will contribute to understanding the role of Antarctic sea ice in Earth's climate system. The model code is designed for parallel computers and has been run on the APAC National Facility AlphaServer SC system, the TPAC SGI system and small Linux clusters. It is hoped that outcomes from this research will be incorporated into high resolution Antarctic weather forecast models to be made operational by Australia in 2007. An operational version of the model can be expected to have a resolution of less than 10km. For the first time, it will provide ships with probabilities of encountering up to 11 different thickness classes of ice together with sea-ice drift speeds in forecasts offering five-day lead time for planning and operations. It is hoped the operational model will also incorporate an assimilation scheme combining information from polar orbiting satellites with model output to generate sea-ice analyses.
|
Principal Investigator Andrew RobertsAntarctic CRC / IASOS University of Tasmania |
Project e39 |
|
Co-Investigators William BuddTPAC University of Tasmania Petra Heil Australian Antarctic Division University of Tasmania Roger Hughes Dept of Civil and Environmental Engineering University of Melbourne |
RFCD Codes 260299 |
Significant Achievements, Anticipated Outcomes and Future Work
The final version of the sea ice forecast model development has been completed, and the results have been presented in a Ph.D. thesis entitled "Medium range numerical prediction of Antarctic sea ice" to be submitted on March 19, 2004.
There are three key outcomes from this work: Firstly, a new physics scheme has been devised to provide deterministic sea ice thickness forecasts. The scheme is based on a new sea ice mechanics theory that has been developed as part of this research. It assumes the scaling features observed in the fractal geometry of sea ice are replicated in the mechanical properties of sea ice, and the model is able to closely match some observed Antarctic thicknesses using the scheme. Secondly, a simple assimilation scheme has been constructed to combine satellite-observed sea-ice cover with the modelled sea-ice state. This has enabled the generation of sea-ice analyses that provide a relatively close estimate of the current state of the Antarctic pack and serve as an initial condition for numerical sea ice predictions. Finally, the model greatly improves forecast accuracy over simply projecting sea ice state forward from satellite imagery, as is the currently accepted method used in Antarctic operations.
An example of a 4-day forecast is provided in Figure 1. The sea ice thickness and cover analyses in Figures 1(a) and (b), respectively, have been generated by assimilating numerical model results with satellite-derived sea ice cover. The forecast change in sea-ice thickness and cover 96 hours into the future from the analysis is shown in Figures 1(c) and (d), respectively using a 50km resolution version of the forecast model. The forecasts have been driven with atmospheric analyses from the European Centre for Medium Range Forecasts. These are a relatively accurate estimate of the state of the atmosphere above sea ice, so that the "forecasts" provided here represent how the model changes when not guided by satellite imagery, guided only by the weather above sea ice. In future, the guidance will be provided by a high resolution weather prediction model.
Computational Techniques Used
The model has made use of an Elastic-Viscous-Plastic sea ice rheology which has been especially designed at Los Alamos Laboratory for solving the sea ice momentum equation on scalar parallel machines. This method makes a small change to observed sea ice physics in order to facilitate a considerable speed-up over existing methods which are not well suited to parallel execution.
Publications, Awards and External Funding
External Funding and Awards
None.
Publications
None.