Density assumptions for converting geodetic glacier volume change to mass change

When scientists measure how much a glacier has shrunk, they don't actually measure mass. They measure volume. Satellites, aircraft, and ground surveys all compare digital elevation models taken years apart and calculate how much the surface has dropped. That gives you cubic meters. But the number the world needs for sea level projections and for water resource planning is kilograms. Getting from one to the other requires knowing the density of whatever was lost. That turns out to be the problem Matthias Huss set out to solve rigorously. The conversion looks simple on paper: mass change equals a density factor multiplied by volume change. Call that factor f. Most geodetic studies just plug in a constant, often the density of ice, around 900 kilograms per cubic meter, and move on. Some have argued for a slightly lower value of 850 kilograms per cubic meter. Huss shows that the choice matters. Using ice density systematically overestimates mass loss by roughly 2 to 15 percent in many real cases. That might sound modest, but applied across every glacier on Earth over decades, it compounds. The reason the factor isn't simply ice density comes down to firn. Firn is the intermediate material between fresh snow and glacial ice — a compacting, low-density layer that can be tens of meters thick on mountain glaciers. The classical assumption, known as Sorge's Law, says the vertical firn density profile stays constant over time as long as climate is steady. If that held, the mean density of volume change would just be ice density, and the problem would be solved. But climate is not steady, and glaciers are not in equilibrium. Here's what actually happens: when a glacier tips into sustained negative mass balance, the equilibrium line altitude, the elevation separating where snow accumulates from where it melts, rises. The firn zone shrinks. Low-density firn at the surface, with a typical initial density around 490 kilograms per cubic meter, gets replaced by exposed ice or by melt and refreezing. So the material being lost is a mix: some ice and some firn, in proportions that depend on the glacier's recent history. The effective density of the volume change ends up somewhere between ice density and firn density, but exactly where depends on how the glacier has been behaving for the past several decades. To quantify this, Huss built a modeling framework using a modified version of the Herron-Langway firn densification model, a tool that tracks how each annual layer of snow compacts into ice over time. He set ice density at 900 kilograms per cubic meter, initial firn density at 490 kilograms per cubic meter based on 19 compiled firn profiles, and pore close-off at 830 kilograms per cubic meter, after which ice density is approached at 10 kilograms per cubic meter per year. The model was tuned specifically for temperate mountain glaciers. He then ran four experiments on a simplified synthetic glacier, a constant-width slab spanning from 300 to 2000 meters elevation, with no ice flow dynamics, preceded by a 50-year spin-up to reach equilibrium. Experiment one involved an abrupt equilibrium-line shift of 100 meters held for 50 years. Experiment two featured a linear trend of 5 meters per year. Experiment three looked at a change in the spatial gradient of mass balance rather than the average. Experiment four repeated the step change from experiment one, but with 200 realizations of interannual variability layered on top. Then he applied the same model to two real Swiss glaciers: Griesgletscher and Silvrettagletscher, using observed annual mass balance maps spanning roughly five decades and geometry updated from multiple digital elevation models. Over the period from 1987 to 2007, Griesgletscher was losing 1.15 meters of water equivalent per year, while Silvrettagletscher was losing 0.66. These glaciers are under serious stress, and their firn zones show it. The headline result is this: for observation periods of roughly five to fifty years, under a wide range of climate scenarios, the conversion factor clusters around 850 plus or minus 60 kilograms per cubic meter. That's the recommended value. It sits below ice density because firn processes are genuinely present. Removing low-density firn layers makes the volume change larger relative to the accompanying mass change, pulling the effective density down. Over very long timescales, the factor drifts back toward ice density as firn re-equilibrates, and the model confirms multi-decadal values typically in the range of 882 to 939 kilograms per cubic meter. For the two Swiss glaciers, five-decade values were 878 and 866 kilograms per cubic meter, which are close to, but measurably below, ice density. Now, the failure modes, because this is where the finding gets genuinely strange. For short observation windows of one to three years, the conversion factor becomes unreliable in ways that are not small and not symmetric. Annual evaluations for Griesgletscher and Silvrettagletscher produced f values ranging from negative 500 to positive 6500 kilograms per cubic meter when annual balances were close to zero — near the range of negative 0.2 to positive 0.2 meters water equivalent per year. In the synthetic experiments, two-year values under changing mass-balance gradients reached 1720 kilograms per cubic meter, while five-year values reached 1405. These numbers exceed ice density, which seems physically impossible. But it isn't a modeling artifact. The explanation is geometric. When the mass-balance gradient changes, the accumulation and ablation zones respond differently. Volume can increase in one area while mass decreases in another, or compaction can alter volume with almost no mass change. In the extreme, mass change and volume change can decouple entirely. In principle, the factor can take any value from negative infinity to positive infinity when one of them is near zero. Huss is explicit: the range reported for short intervals, from zero to beyond 2000 kilograms per cubic meter, is real, not a curiosity. Sensitivity tests show that glacier geometry and climate regime shift the recommended value only modestly, on the order of negative 18 to positive 20 kilograms per cubic meter. So, 850 plus or minus 60 is genuinely portable across many mountain glacier settings. What isn't portable is applying it blindly to short surveys or to periods when a glacier's mass balance is hovering near zero. The practical upshot is pointed. Geodetic surveys are increasingly important because they can cover glaciers that nobody visits, using satellite data at continental scale. The temptation is to apply a single density conversion uniformly. Huss shows when that's reasonable: long surveys, stable mass-balance gradients, a firn area present, and volume change clearly different from zero — and when it isn't. For periods of three years or less, or for glaciers near balance, the uncertainty from the density assumption alone can exceed 0.05 meters water equivalent per year. That's large enough to obscure real signals. The number 850 plus or minus 60 kilograms per cubic meter is, in the end, a hard-won practical answer. It emerges not from a simple physical argument but from careful modeling of firn dynamics across dozens of scenarios, validated against two glaciers with exceptional observational records. Knowing the number is useful. Knowing its limits — the short timescales, the near-zero balances, the shifting gradients — is what separates a reliable mass balance estimate from a confident-sounding mistake. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.