Why does an ice shelf already floating in the sea still matter for sea level?

Because its importance is mechanical, not a matter of its mass. A shelf wedges between coasts and islands and buttresses them, resisting and slowing the flow of land ice into the sea. It is a plug, not a cargo. When a shelf breaks up the glaciers behind it speed up, sometimes several times over, and that ice is on land today and does add water to the ocean when it arrives.

The confusion here is common and understandable, and it starts from a thought experiment that is perfectly correct. An ice cube melting in a full glass does not spill it, because the ice was already displacing a volume of water equal to its own weight. That is Archimedes’ principle, and an ice shelf floating on the sea obeys it too.

So the importance of a shelf lies not in its mass but in its mechanical role. It is squeezed between the walls of valleys and snags on submerged islands, so it resists movement and pushes back against the land ice feeding it. That buttressing is what makes it a plug and not a cargo.

When a shelf breaks up, the glaciers behind it accelerate, sometimes several times over. That ice is on land today, and when it reaches the sea it does add water to the ocean. The break-up itself barely changes the level, and what follows it changes a great deal.

The measurement comes from space: satellites track the speed of ice flow and its thickness over years, and after shelves broke up the glaciers that fed them were seen to accelerate measurably. The remaining uncertainty is mostly about what happens under the ice, where warm water eats at the shelf from below and direct measurement is hard. So the range of estimates for future rise is still wide, and that is precisely the number coastal engineers need.