The Larsen C ice shelf - with an area about the size of Delaware and a volume near that of Lake Michigan - is poised to break away from Antarctica at any moment. The continual loss of sea ice (and glaciers) is contributing to climate change by a decrease of the Earth's albedo (reflectivity).
Climate models for the relationship between temperature, ice cover and albedo generate a result known as hysteresis. In a nutshell, hysteresis is a different path for change that depends on the starting conditions. Below is an example of a hysteresis plot for a warming Earth (red curve) and a cooling Earth (blue). The temperatures were calculated by an iterative process using a relationship between albedo and ice cover, and a relationship between ice cover and temperature (best-reasonable estimates from the scientific literature). Iteration is used in many scientific and engineering applications for calculations that cannot be solved directly.
The take-home message of the hysteresis plot below is that once the polar ice caps have melted (red curve), the loss of albedo will drastically increase the warming effect from further increases in greenhouse gases. The cooling curve (blue curve) suggests that an Earth beyond this tipping point must become much, much cooler than before to recover polar ice and increase the Earth's albedo. Thus, an ice-free Earth could only be reversed by the decrease in temperature of an ice age!
(The model I used to generate the hysteresis plot was written in the Python programming language, for the Climate Change Modeling class from Coursera offered by professor David Archer of the University of Chicago. I recreated the plot and added annotations in the graphics program GIMP).