The Greenland Ice Sheet, the world’s second-largest ice sheet, is losing mass at an increasing rate and accounting for about 25% of the current sea-level rise. Scientists use advanced climate models to predict the rate of mass loss but face challenges since the results vary greatly. Researchers analysed how the models account for meltwater, the expansion of the melt zone and how much solar heat the ice absorbs to understand the substantial differences in the predictions. The study emphasises that precise climate models are needed to accurately predict future sea-level rise.
The Greenland Ice Sheet is melting, and the meltwater is flowing into the sea, causing sea-level rise.
This melting accounts for 25% of the sea-level rise, and scientists are working to determine how rapidly the Ice Sheet will melt. They use advanced computer models to make predictions, but the results vary.
To understand the discrepancies, researchers compared three climate models that predict trends in the surface of the Ice Sheet. Their study indicates that the differences mainly result from how the models account for meltwater, how rapidly the melting zone expands and how much solar heat the ice absorbs.
“We used three climate models to analyse how much the Greenland Ice Sheet will melt in warmer conditions. All the models started with the same initial data but produced different results. The variation primarily stemmed from the amount of water that refreezes and the extent to which the melting zone expands. Two of the models, MAR and HIRHAM, found that the melting zone grew more rapidly than the third model, RACMO, which means more runoff water flowing into the sea. Despite these differences, the Ice Sheet continues to melt, and sea level continues to rise as temperatures increase,” explains Nicolaj Hansen, climate researcher, Danish Meteorological Institute, Copenhagen, Denmark.
Evaporation, sublimation and runoff from melting
The Greenland Ice Sheet strongly affects sea level. When the Ice Sheet melts and the runoff flows into the sea, it contributes to rising sea level, affecting coastal communities worldwide.
Projecting Sea-Level Rise: from Ice Sheets to Local Implications (PROTECT) is a project funded by the European Union investigating how sea level will change and how the Greenland Ice Sheet affects this. This study is part of PROTECT and focuses on the physical processes, especially the surface mass balance of the Greenland Ice Sheet.
To understand how much mass the Ice Sheet will lose in the future, researchers examine the surface mass balance, which describes only the changes to the surface of the Ice Sheet and not the total loss of mass. The changes comprise the difference between the precipitation falling on the Ice Sheet and the total loss of mass from evaporation (liquid phase to gas phase), sublimation (solid phase to gas phase), melting (solid phase to liquid phase) and liquid runoff into the sea.
“Snow and rain add mass to the Ice Sheet. The surface mass balance is calculated by taking all the precipitation that falls and subtracting evaporation, sublimation and runoff from melting,” says Nicolaj Hansen.
The Ice Sheet also loses mass in another way: discharging calving icebergs into the sea. This process is not included in the surface mass balance. Understanding this distinction is important because the surface mass balance shows how climate change is affecting the melting processes on the surface.
“The surface mass balance alone does not tell the full story of how rapidly the Greenland Ice Sheet is shrinking. We investigated surface mass balance using three climate models to better predict how the Ice Sheet will change in the future.”
Climate models diverge
Researchers use regional climate models to make predictions by simulating how temperature, wind, snowfall and melting affect ice.
“Regional climate models cover only a limited area – in this case, Greenland. However, these models require external data to function accurately. In this study, we provided them the same background data from a larger global model, CESM2, and applied a high-end warming scenario, SSP5-8.5, an extreme scenario. This assumes that we continue to add considerable energy to the atmosphere,” notes Nicolaj Hansen.
The researchers compared three models – RACMO, MAR and HIRHAM – all driven by the global climate model CESM2. These models were tested under SSP5-8.5, which assumes high greenhouse-gas emissions.
However, despite being fed the same data, the models produced very different results.
“The models start with the same data but diverge when we run them. This results in large differences, especially when we project towards the end of this century.”
Albedo effect: when melting creates more melting
The biggest source of the differences in the models is how they account for meltwater runoff. When the ice melts, some of the meltwater seeps into the snow and refreezes and the rest runs off into the sea. The models vary in the amount of water they project refreezing.
“The primary source of differences in the surface mass balance between RACMO, MAR and HIRHAM is the runoff component. The more rapid expansion of the melt zone MAR and HIRHAM project – 70% and 86% of the Ice Sheet by the end of this century versus 47% for RACMO – contributes to the differences in the runoff forecasts,” explains Nicolaj Hansen.
Another important phenomenon contributing to the differences in the models is the albedo effect. Albedo describes how much sunlight a surface reflects. White snow reflects the most sunlight, and dark ice absorbs more sunlight and therefore heat.
“Newly fallen white snow reflects most of the sunlight, but if this snow melts and darker ice is exposed to solar radiation, more heat is absorbed. This makes the ice melt even faster,” observes Nicolaj Hansen.
As the melt zone expands, more ice is exposed, further exacerbating the melting process.
Artificial intelligence and machine learning can improve the models
Another reason for the differences between the models is how detailed the grid resolution is.
“Climate models divide Greenland into grid cells, and the resolution varies between models. For example, one model uses a 5 km grid and another uses a 10 km grid. The higher the resolution, the more accurate the results, especially for the surface mass balance in coastal areas,” says Nicolaj Hansen.
Researchers are continually working to improve the models by refining their descriptions of snow properties, melting processes and the albedo effect.
“The models are constantly being improved. For example, at the Danish Meteorological Institute, we have developed HIRHAM through several versions, now reaching HIRHAM5. This also applies to the RACMO and MAR models. When researchers discover a more accurate way to describe cloud formation, the model is adjusted, tested and updated,” notes Nicolaj Hansen.
One new method being investigated is artificial intelligence, which can make climate models more rapid and accurate.
“We are investigating how artificial intelligence and machine learning can improve climate model calculations. They can potentially reduce computation time and increase the accuracy of projections of surface mass balance,” says Nicolaj Hansen.
Uncertainty does not change the conclusion
Despite differences in the exact figures, the conclusion remains the same: the Greenland Ice Sheet is losing mass, and this will continue as temperatures rise.
The models show different absolute values but all indicate a clear negative trend in surface mass balance.
“The study underscores the need to improve climate models to more accurately predict the rise in sea level. This is crucial for protecting coastal communities and developing effective strategies to address future climate change,” concludes Nicolaj Hansen.
