What the study found
A statistical semi-empirical bias-adjustment model reduced systematic errors in Greenland ice-sheet snow accumulation estimates from several regional climate model and reanalysis products. After adjustment, mean point-wise biases were reduced to ±0.3%, and the authors report improved agreement among models in the observation-rich accumulation zone.
Why the authors say this matters
The authors say accurate accumulation estimates are essential for reliable projections of sea-level rise. They conclude that the framework offers a scalable and transferable way to improve accumulation estimates by better integrating observational data, with the potential to improve input for ice-sheet models and reduce uncertainty in future sea-level rise projections.
What the researchers tested
The researchers developed a statistical-semi-empirical model that uses Empirical Orthogonal Function analysis, a method for identifying dominant patterns of variability, to decompose gridded accumulation output and fit adjustment coefficients from the SUMup observational dataset. They applied it to monthly output from HIRHAM5, MAR3.14, RACMO 2.4p1, and CARRA over periods from 1960 to 2022 depending on the product.
What worked and what didn't
Initial mean point-wise biases were reported as -7.4% for HIRHAM, -0.5% for MAR, 0.0% for RACMO, and +10.1% for CARRA for 1991–2022, with all but RACMO statistically significant. After adjustment, these biases were reduced to ±0.3%. The bias-corrected mean annual accumulation rates for 1991–2022 were estimated at 469 mm yr−1 for HIRHAM, 412 mm yr−1 for MAR, 435 mm yr−1 for RACMO, and 408 mm yr−1 for CARRA; inter-model agreement improved by 68% in the accumulation zone but worsened by 27% in the sparsely sampled ablation zone.
What to keep in mind
The abstract notes that model bias is strongest in the southern ice sheet and that the ablation zone remains sparsely sampled, which limits agreement there. The summary provided does not describe additional limitations beyond the need for more observational constraints in undersampled areas.
Key points
- The model reduced mean point-wise bias across Greenland accumulation estimates to ±0.3%.
- Bias-corrected annual accumulation estimates were reported for four products: 469, 412, 435, and 408 mm yr−1.
- Inter-model agreement improved by 68% in the observation-rich accumulation zone.
- Agreement worsened by 27% in the sparsely sampled ablation zone.
- The largest statistically significant bias contributions were in southern Greenland.
Disclosure
- Research title:
- Bias-corrected Greenland accumulation maps align more closely with observations
- Authors:
- Josephine Lindsey-Clark, Aslak Grinsted, Baptiste Vandecrux, Christine S. Hvidberg
- Institutions:
- Niels Brock, Geological Survey of Denmark and Greenland
- Publication date:
- 2026-03-05
- OpenAlex record:
- View
Get the weekly research newsletter
Stay current with peer-reviewed research without reading academic papers — one filtered digest, every Friday.