Make your paper interactiveCreate

Text from the interactive explanation

Resolving short-term meteorological variability amplifies projected glacier loss and sea-level rise

Lander Van Tricht, Matthias Huss, Harry Zekollari, Rodrigo Aguayo, Janosch Beer, Marit van Tiel, Alexandra von der Esch, Daniel Farinotti

Original paper

Run one summer through two versions of the same glacier model. One reads the weather day by day; the other reads monthly averages. Watch where they part ways.

The question this page builds toward: can a few hot days change how much the sea rises?

Left: daily model · right: monthly model. Case: . Both versions are tuned to give the same result for this glacier’s typical summer (the monthly version’s melt factors 1.00), mirroring how the paper calibrates both versions to the same observed past.

Illustrative simulation with synthetic weather. It follows the model structure described in the paper’s Methods to show the mechanisms. It is not the paper’s model, and not a forecast for Svalbard, Iceland or the Alps. The paper’s own results start in chapter 4.

Same average, different summers

Global glacier models don’t simulate every ray of sunshine. They use a simple, well-tested rule called a temperature-index or degree-day model: on every day the air is above 0 °C, snow and ice melt in proportion to how warm it is. Add up those “degrees above freezing” over the days and you get positive degree-days. Multiply by a melt factor and you have melt, in millimetres of water.

A daily model simply counts them. A monthly model only knows the month’s average temperature. The standard version of the paper’s model, GloGEM, therefore estimates how the days were spread around that average, using a bell curve whose width comes from past weather.

The paper’s illustration uses a spread of 5 °C. The bell-curve method lets melt happen even in months that average below freezing, but it cannot know when a particular month had unusually extreme days.

One month at the top of a glacier

Melt here uses a snow melt factor of 3.5 mm per °C per day, within the range used in the paper (1.75–4.5). The bell curve keeps this month’s original spread.

Rain or snow? It’s decided near 0 °C

Snowfall is what builds a glacier; rain mostly runs straight off. In the model, precipitation falls as snow below 0.5 °C, as rain above 2.5 °C, and as a mix in between.

Schematic

A monthly model decides the phase once, from the month’s average temperature. A daily model decides it for each storm, using that day’s temperature. The difference is largest when storms come with mild air, as they often do in Iceland:

“Heavy precipitation events frequently coincide with relatively warm conditions, as mild south-westerly air masses typically deliver both heat and moisture from the North Atlantic.”

One stormy month, two ways to split it

When the snow is gone, the ice races

Fresh snow is bright and reflects most sunlight. Old, compacted snow (firn) is greyer, and bare glacier ice is darker still. The model captures this with melt factors in the ratio snow 1 : firn 1.5 : ice 2 . Once the snow is gone, the same warm day melts twice as much.

The daily model switches the surface on the day the snow disappears, or the day fresh snow falls. The monthly model keeps one surface for the whole month. In the paper this generally lowers the monthly model’s ice melt in early summer and raises it in late summer.

Inside the model a glacier is a stack of thin elevation bands, 10 m each in the paper’s model, and each band keeps its own snow, temperature and melt. The panel shows the same hot summer band by band. Each bar’s length is the share of the glacier’s area at that height.

Compare the two shapes. On a flat ice cap , a huge share of the area sits within a narrow range of heights. When the snowline climbs onto that plateau, a large area turns to bare ice almost at once. On a steep valley glacier , the area is spread over a wide range of heights and the snowline creeps up gradually.

“In these low-relief, ice-cap-dominated regions, the narrow elevation range strongly enhances the sensitivity to surface-type transitions and melt-albedo feedbacks.”

The hot summer, band by band

White = snow on top, cream = firn, blue = bare ice. Cyan ticks mark bands that are melting that day.

From one summer to the year 2100

The three effects you just explored are what the researchers set out to measure on the real world’s glaciers. Lander Van Tricht and colleagues built a daily version of the Global Glacier Evolution Model (GloGEM) and ran it side by side with the standard monthly version. It is the first time daily weather has driven a global glacier model to produce scenario-based projections for the 21st century.

Both versions were calibrated glacier by glacier to the same observed mass changes for 2000–2019. They agree about the recent past and differ in how they respond to extremes and to a warming climate.

Svalbard and Iceland pull in opposite directions Reported in the paper

Projected share of 2025 glacier ice lost by 2100 · SSP3-7.0 · Intermediate emissions · mean of five climate models ± 1 standard deviation

Bars run from 0 to 100% of the 2025 ice volume; white whiskers show ±1 standard deviation across the five climate models. The whiskers often overlap: the spread between climate models is large, and the daily–monthly difference is a shift within it.

All 19 glacier regions Reported in the paper

Gold = daily, lavender = monthly; right column = difference in percentage points of 2025 volume (positive: daily loses more). Select a region to compare it with Svalbard. Data: Extended Data Table 3.

Four ways the gap behaves over the century Schematic

Sketches of how the daily-minus-monthly ice volume difference evolves from 2025 to 2100 in each group of regions (downwards = daily loses more). Shapes are schematic; the numbers in the text are from the paper.

Does reading every day match the past better?

Before trusting the projections, the authors compared both versions with 4037 independent yearly measurements of glacier-wide mass balance on 175 glaciers (1980–2024, World Glacier Monitoring Service). Mass balance is the yearly gain or loss of ice, expressed as metres of water spread over the glacier.

Root-mean-square error, m of water equivalent per year (lavender monthly, gold daily); bar scale 0–1. Overall the gain is modest: R² rises from 0.71 to 0.75, and measurements themselves carry ~0.2–0.4 m uncertainty. In extreme years, outside ±2 standard deviations, the gain is clear: the typical error (MAE) falls from 0.81 to 0.64.

RMSE for measured glaciers in each region (Extended Data Table 2); bar scale 0–0.8. Improvements are not uniform: some regions change little or get slightly worse, and some have very few measured glaciers.

Record years are the test. In the record Alpine melt year 2022, the daily version simulated a mass balance 0.31 m of water more negative than the monthly version, “substantially closer to observations”. Similar gaps appeared in other recent extremes: 0.15 m in Western Canada and the USA in 2023, and up to 0.98 m in Svalbard in 2024.

Schematic Glacier meltwater feeds rivers, and hydrological models usually work day by day. Daily forcing turns monthly steps into a continuous curve, so the timing of high flow can be tracked. By 2100 the daily model projects high flow in Western Canada and the USA about 46 days earlier than in 2001–2020 (SSP3-7.0 and SSP5-8.5); the monthly model misrepresents such shifts by about 15 days. In Central Europe, simulated runoff in the extreme year 2022 was 8% higher with daily forcing, and 12% lower in the cool year 2014. In High Mountain Asia, the daily model has ice melt making up about 70–80% of present-day runoff by September, against 44% in the monthly model. The authors describe these splits as model diagnostics, not validated estimates.

The world’s glaciers, and the sea

Added up over all modelled glaciers, the small yearly differences grow steadily through the century. The daily version projects 4–5% more global ice loss than the monthly version by mid-century, and 5–10% more by 2100, depending on the scenario. That leaves 7–15% less glacier ice at the end of the century. The gap is still widening in 2100, and because glaciers respond over centuries, the authors expect the effect to persist beyond it.

Mean of five climate models ± 1 standard deviation (Extended Data Table 3). Gold = daily version, lavender = monthly version.

Solid = lower end of the range, hatched = up to the upper end across scenarios. The extra contribution is substantial but remains within the spread between climate models.

Can a few hot days change how much the sea rises? In these simulations, yes. Letting the model feel each day adds 20–33 mm of sea-level rise from glaciers by 2100.

The authors point out that a few extra centimetres matter for coastal flooding, erosion and saltwater intrusion. Their conclusion is that temporal resolution “is not a technical detail but a modelling choice with direct consequences for impact assessments.” Leaving it out may systematically underestimate future glacier loss.

What this does, and doesn’t, show

Polar magnitudes are upper-end

A deliberate, simplified setup

Future weather has limits too

The cost of seeing every day

Projections, not predictions

About this page’s simulations

Source

Van Tricht, L., Huss, M., Zekollari, H., Aguayo, R., Beer, J., van Tiel, M., von der Esch, A. & Farinotti, D. (2026). Resolving short-term meteorological variability amplifies projected glacier loss and sea-level rise. Nature Communications 17, 10696. https://doi.org/10.1038/s41467-026-78260-y

Open access under a Creative Commons Attribution 4.0 International License. Model code: GloGEM on GitHub · data and simulation outputs: Zenodo.

All graphics on this page are original drawings made with code; no figures from the article are reproduced. Projection and evaluation numbers are transcribed from the article and its Supplementary Information (Extended Data Tables 2–3 and Extended Data Fig. 1). Quotations are from the article. The glacier simulations are simplified illustrations of the model structure described in the paper’s Methods, run on synthetic weather. Explanatory text is adapted and paraphrased from the article.

Topics in this iPaper

Earth & environment
Global glacier models typically rely on temperature-index mass balance calculations forced with aggregated meteorological data

Abstract

glaciology
an elevation-band model designed to compute glacier mass balance, geometry and volume evolution at regional to global scales

Methods, Glacier modelling

sea level rise
resulting in an additional 5-10% mass loss (20-33 mm of global mean sea-level rise) relative to the monthly implementation

Abstract

climate change
As climate change increases the frequency and intensity of extremes

Introduction

temperature index model
Ablation is computed using a temperature-index (degree-day) approach

Methods, Mass balance model

glacier runoff
the daily model explicitly resolves sub-monthly variability in glacier runoff, producing a continuous seasonal cycle

Results and discussion, The role of daily modelling in glacier runoff timing and composition

Resolving short-term meteorological variability amplifies projected glacier loss and sea-level rise — Interactive explanation | iPaper.ai