For several weeks in early 2002, a slab of floating Antarctic ice larger than Rhode Island disintegrated into a moving mosaic of fragments. More than 3,250 square kilometers of the Larsen B Ice Shelf broke apart within about 35 days. By mid-April, the main event had ended, leaving open water, drifting icebergs, and exposed glacier fronts along the eastern Antarctic Peninsula.
The speed was extraordinary, but the collapse was not unprepared. Larsen B had been thinning, retreating, and developing fractures for years. Repeated warm summers supplied the final ingredient: dense networks of surface melt ponds whose water forced existing cracks through the shelf.
The Shelf Was a Floating Extension of Land Ice
An ice shelf forms where glaciers flow off land and remain attached as a floating plate. Because it already floats, its breakup adds little directly to sea level. The shelf can nevertheless restrain the grounded glaciers feeding it. Contact with embayment walls, islands, and slower ice along the margins produces back stress that acts like a brake.
Larsen B occupied an embayment south of the Larsen A shelf, which had largely collapsed in 1995. The northern Antarctic Peninsula had undergone strong atmospheric warming during the second half of the twentieth century. Summer melt became frequent enough for water to collect in surface depressions, while rifts and crevasses divided the shelf into increasingly vulnerable blocks.
Ocean-driven thinning may have helped weaken the system, but surface meltwater was crucial to the abrupt 2002 fragmentation. The event differed from the ordinary calving of one or several large icebergs. Thousands of narrow pieces appeared almost simultaneously, signaling failure throughout a broad area of the shelf.
Meltwater Opened Cracks From the Top
Liquid water is much denser than ice. When a pond fills a crevasse, its weight presses outward on the crack walls and concentrates stress at the tip. If enough water remains in the fracture, the crack can deepen through a process called hydrofracture. Neighboring ponds may drain suddenly as connected cracks open.
This creates a chain reaction. Water from one collapsing pond can enter other fractures, while the flexing of newly separated blocks produces fresh cracks. Satellite images before the breakup showed extensive ponding; during the collapse, the shelf transformed from a continuous surface into a field of long, capsizing slivers.
The blocks’ shape added another destructive mechanism. Tall, narrow fragments were unstable when floating freely. As they rotated, they pushed against adjacent ice and generated waves, helping the breakup propagate. The result was not a slow retreating front but a rapid mechanical disintegration across much of Larsen B.
Tributary Glaciers Lost Their Brake
The strongest evidence for shelf buttressing appeared after the floating ice vanished. Several glaciers that had fed Larsen B accelerated dramatically, commonly reaching several times their former speeds. Their surfaces lowered as ice was drawn more rapidly from inland toward the coast. Glaciers that retained substantial shelf support or encountered strong bed resistance changed less.
Observations during the first 18 months found speed increases of roughly threefold to eightfold among some affected glaciers. The exact response varied because each tributary had a different bed slope, width, thickness, and connection to the former shelf. Acceleration also evolved over time rather than remaining at its initial peak.
This aftermath clarified an important distinction. Floating shelf ice does not need to add much water directly to the ocean to affect sea level. By changing the rate at which grounded ice crosses the coastline, shelf loss can increase the ocean’s mass indirectly. Larsen B became a natural experiment demonstrating that connection on a scale satellites could observe.
Collapse Required Thresholds to Be Crossed
Not every melt pond triggers shelf failure, and not every ice shelf responds to warming in the same way. Thickness, fracture density, snow properties, firn air space, geometry, and the availability of drainage routes all matter. Dry, porous firn can absorb meltwater and allow it to refreeze. Once that pore space is reduced by repeated melting, water is more likely to remain at the surface and fill cracks.
An ice shelf can therefore change gradually for years before breaking rapidly. Thinning increases stress and makes fractures more consequential; surface ponding supplies pressure; a warm interval then links local failures into a shelf-wide event. The 2002 collapse was abrupt in the calendar but cumulative in its preparation.
A southern remnant of Larsen B survived the main breakup, protected partly by local geometry and persistent fast ice in front of it. Changes in that surrounding sea ice and continued fracturing show that the broader system has not returned to its earlier state. The remaining shelf and tributary glaciers continue to provide evidence about how support is lost in stages.
Larsen B does not serve as a direct timetable for the colder, much larger ice shelves of West Antarctica. Their dominant sources of melting and their bed settings differ. It does provide the clearest observed example of hydrofracture turning widespread surface melt into rapid disintegration—and of grounded glaciers responding once a floating restraint is removed.