Researchers used a physics-informed neural operator to test how much Greenland ice loss can be inferred from sea-level patterns. Broad loss patterns emerged clearly, but different ice-loss shapes can create nearly identical signals, limiting recovery of finer details.
For satellite-altimetry reference data, the fingerprint-only accuracy ceiling was an estimated R-squared of about 0.79. The limit depends on where sea level is measured and what is assumed about ice loss. Independent ice-sheet observations and physical constraints could improve estimates.
✦ Sea-level patterns show Greenland’s broad ice loss but cannot reveal every local detail on their own.