A digital twin of the Jungfrau–Aletsch glaciers: putting meltwater to best use · Team Swisstainability
Screening result from public data, not a construction recommendation. Sites, volumes and glacier areas come from the team's terrain analysis. Protected areas are checked against federal inventories. Natural hazards are read from federal index maps only; geology and access are not assessed, so no site is cleared.
Terrain, 299 glaciers and the paths their meltwater takes, 15 basin sites with their volumes, federal protection inventories, hazard index maps, hydropower plants and dams.
Runoff period, emissions path, dam height, melt rate and glacier size. Change one in the row above and the ranking, the storage, the hydropower stress test and the business case follow.
Score out of 100 and usable volume behind the chosen dam. Pick a site to open its evidence on the Sites sheet.
Base maps: team QGIS exports, © swisstopo. Satellite: Sentinel-2 mosaic of , swissEO S2-SR, © swisstopo, contains modified Copernicus Sentinel data. Glacier outlines: SGI 2016 (GLAMOS). Red line: study area (781 km²). Numbers on the map are the current rank. A dark dot marks a site under strict protection. Hazard layers: SilvaProtect-CH and the map of potential permafrost distribution (FOEN), index maps valid to 1:50,000, in the colours of map.geo.admin.ch. Plants and dams: SFOE. Meltwater flow paths are not in the evidence pack yet.
Rivers: Federal Office for the Environment, through the federal linked-data service (lindas.admin.ch). Air temperature and precipitation: MeteoSwiss automatic stations (data.geo.admin.ch). The gauges lie below reservoirs and diversions, so they show what each river carries now, not the inflow at a basin site.
A prototype of trigger conditions, not a warning service. A day is marked Watch when one condition is met and Elevated when thaw and heavy precipitation fall together. The two thresholds are the team's first setting. The card further down runs them over every day since March 2021 and four documented events; that is a test, not a calibration. Forecast: MeteoSwiss ICON-CH2 through Open-Meteo, at the 15 site locations, refreshed hourly. Official warnings: natural-hazards.ch.
A test, not a calibration. Every day is judged on the forecast for that day, kept in the forecast archive of Open-Meteo, and worked out as in the outlook above: the highest zero-degree level and the mean precipitation of the 15 site locations. The archive serves the best model it has for each date, so the model changes over the years; the live outlook uses MeteoSwiss ICON-CH2. A thunderstorm cell that misses the 15 points is not seen, and a multi-day event is judged on its strongest day. Four events cannot set a threshold: they show that the rule reacts on days that mattered, and the yearly counts show how often it reacts on days that did not.
The Birch glacier above Blatten in the Lötschental collapsed on 28 May 2025. Ice and rock buried most of the evacuated village and dammed the Lonza. The federal gauge at Blatten has sent no reading since that day.
Three of the 15 ranked sites lie in that valley. Blatten is excluded as a settlement area, and the terrain analysis notes that the ground there has changed fundamentally. Fafleralp and the Jegigletscher forefield lie upstream of it.
Avalanches are left out because the federal index map only models avalanches that start in forest, which says nothing above the tree line. The hazard index maps date from 2005 and 2006.
The six exits the terrain analysis lists, km² of glacier together. Its flow balance covers the whole terrain model, which reaches beyond the study area: 324.7 km² of glacier in the model, 324.4 km² arriving at its edge, so the routing closes. The report does not break down the remainder.
Two thirds of the area lies above 2,500 m and mean slope is 30° or more in every band, so flat ground is rare. All 15 ranked sites sit between 1,524 and 2,332 m, below most of the ice.
14 glaciers larger than 5 km² make up 75% of the glacier area. For the Grosser Aletschgletscher the analysis found no suitable site inside the area: its tongue ends in the steep Massa gorge. Thinning: how much the surface sank per year from 2000 to 2019, measured from space (Satellite sheet).
Sentinel-2 true-colour scenes: swissEO S2-SR, © swisstopo, contains modified Copernicus Sentinel data .
Sentinel-2 passes over the area every two to three days, and swisstopo publishes each mosaic 10 to 24 hours after the overpass. Your browser reads the picture and the cloud mask straight from the cloud-optimised files of the federal data catalogue (data.geo.admin.ch). Cloud share: cloud and cloud shadow inside the study area. Click a pass to see it.
A first-order stress test, not a plant model. The plants are matched to the nine catchments by name and location on the map; the match is this dashboard's and has not been checked with the operators. A run-of-river plant follows the monthly runoff up to its turbine capacity. A storage plant is shown by the energy of the water that reaches it, because its reservoir decides when that water is turbined. Every plant is scaled so that the reference runoff of 1981–2010 gives its expected production in the SFOE statistics. Runoff: Hydro-CH2018 at the gauge of each catchment, ensemble median; the range takes the lowest and the highest model month by month. Plants further down a river also take water from below the gauge, and pumping is left out.
Energy held in the Swiss storage lakes as a share of what they can hold, week by week. The lakes are drawn down over the winter and refilled by snow and ice melt; how early and how fast they refill differs from year to year. Source: SFOE, filling of the storage lakes, read when this page was built.
Net import is imports minus exports, summed over the winter half-year. Source: SFOE energy dashboard, cross-border exchange (Swissgrid figures). The winter energy of the basins is the illustrative Trift scaling of the Business case sheet, for the new sites still in play with the settings above. Nuclear outages and prices move the winter balance as much as the weather does; the grid itself, its lines and its market, is not modelled.
Order-of-magnitude estimate, scaled in a straight line from the KWO Trift project (CHF 387 m, 85 Mio m³, 215 GWh of winter energy) and assuming a similar chain of power stations downstream. Investment follows the basin volume, energy follows the water the glaciers can supply. Head, the power chain and costs need a feasibility study.
In this scenario every glacier shrinks by the same share. Real retreat removes low-lying ice first, so a basin below a small, low glacier loses its supply sooner than shown. Move the Glacier size control at the top to apply a size to the whole dashboard.
Plants of 300 kW or more from the SFOE hydropower statistics () and dams under federal supervision. Distances are straight lines; whether water from the site can reach a plant is not checked.
Swiss reservoirs fill with snow and ice melt from spring to early autumn and are drawn down over the winter, when the country imports electricity. Source: SFOE, filling level of the storage lakes, read when this page was built. The grid itself is not modelled.
A site cannot reach high confidence until the open layers are checked by a specialist.
Method limits. Surface runoff only (D8 routing on the 10 m swissALTI3D mean); subglacial flow and karst are missing. Meltwater is glacier area × the melt rate, with no rain or snowmelt from outside the glaciers. Glacier outlines (2013–2018) are older than the terrain model (2024/25), so forefields that have become ice-free since were not tested. Volumes tend to be slightly too small and dam length is indicative.
Sources and attribution.