A spatio-temporal model for Antarctic sea ice formation

The temporal variability of polar sea ice is complex and closely linked to global climate. The amount of sea ice over an area can have a significant effect on the energies transferred between the atmosphere and the ocean. Statistics derived from sea ice observations are therefore of great interest to scientists. We showcase a new method of analysis which may be used to examine the annual variability in sea ice formation. We demonstrate our method using sea ice concentration images derived from Earth‐orbiting satellites that span several decades. The growth and melt of Antarctic sea ice is modelled using a simple two parameter spatial nearest neighbour process that treats ice and water as two species competing for territory. Simulations of the model are used to estimate two time series representing rates of competition. With techniques of functional data analysis, these series may be used to detect both amplitude and phase variation, and to isolate major modes of annual variation in sea ice formation. Copyright © 2008 John Wiley & Sons, Ltd.