Project
Follow the water. Protect its pathways.
Riverflow makes Switzerland’s water cycle observable—from mountain precipitation, snow and ice to groundwater, rivers, lakes, wetlands, use and return. The purpose is practical: detect change, test whether essential pathways are protected, and support evidence-led legal action when they are not.
Last revised 22.08.2026. Sources and comparisons were checked against their published records.
One country, four seas
Switzerland has no coast. It is nevertheless the headwater of four European sea basins: water that falls on the Alps leaves the country in four directions and reaches the North Sea by the Rhine, the Mediterranean by the Rhône, the Adriatic by the Ticino and the Po, and the Black Sea by the Inn and the Danube. The four basins are shared with sixteen other states. Nothing else on this map is as easy to state and as rarely drawn: the country is upstream of a continent.
The shares below are not quoted from anywhere. A one-kilometre grid is laid over the border polygon, every cell is assigned to the river system of the nearest mapped channel, and the cells are counted. The four systems that leave the country are identified by the reach their water leaves through, which is a field the river network already carries.
Read the row of marks under the band as the frontier itself. Three of the four basins have a Swiss gauge standing on the border; the Adriatic has none. Read one step further and only two of those three gauges are measuring a whole river.
What Switzerland is to the Rhine
The Rhine basin is the one place where the Swiss share of a European river can be derived here from primary figures at both ends: the long-term mean at the last Swiss gauge, and the long-term mean the international management plan states for the same river above the Dutch border. The result is disproportionate to the map.
A seventh of the catchment supplies nearly half the water. That ratio is what an Alpine headwater is, and it is the reason a Swiss drought is a Dutch problem. BAFU states the same 45 % for the Rhine, and further states that Switzerland supplies about 20 % of the Rhône's flow, 10 % of the Po's and 1 % of the Danube's (Wasser: internationale Zusammenarbeit, 19.05.2025).
The two figures agree, and the agreement is worth one sentence of caution. The numerator
above is modelled — HydroRIVERS DIS_AV_CMS, the natural mean at the
reach Basel stands on — while the denominator is a measured mean published by the
river's own commission. A derivation with one modelled term is a check, not a source, and
since this site marks that difference everywhere else, it marks it here.
The ice each river is holding
A glacier is a reservoir with no concession or operator. It releases most water when the rivers below are lowest. The question here is not only how much ice remains, but which river holds it. Each glacier is assigned to a basin from its centroid using the grid’s nearest-channel rule.
The Rhône basin holds more than half of what remains of Swiss ice on a fifth of the territory, which is why the Rhône is the river whose late summer changes first. The Inn has lost the largest share of all four, and the Inn is the Danube's Swiss headwater — the one basin where no commission Switzerland belongs to is watching.
Where the registers stop
Every layer on this map ends at the border, because every register behind it is federal or cantonal and federal registers end where the federation does. The water does not. That mismatch is the standing legal fact of this section, and it is worked in four different ways on four rivers:
- The Rhine. The Convention on the Protection of the Rhine of 12 April 1999 binds Switzerland alongside the riparian states and the European Union, and the IKSR coordinates under it; the KHR has coordinated the hydrology since 1970. Switzerland is not an EU member and the Water Framework Directive does not bind it, yet its territory lies inside the international Rhine river basin district, and the internationally coordinated management plan drawn under Art. 13 of that directive counts it among the district's nine states.
- The Rhône. The CIPEL, under the Franco-Swiss agreement of 16 November 1962, in force since 1963, covers Lake Geneva and the Rhône below it.
- The Po. The CIPAIS, under the Italo-Swiss agreement of 20 April 1972, covers the waters the two states share — the basin whose export this map cannot measure.
- The Danube. Nothing. Switzerland is not a party to the Danube Protection Convention and the ICPDR's regime does not reach the Inn's Swiss headwaters, which is the basin that has lost the most ice.
A litigator reading this page should take one thing from it. A Swiss figure on a Swiss layer is, on four rivers, also a fact about a river that belongs to somebody else, and the register that produced it was never asked that question. The rest of this site is about the age and the class of those figures. This section is about their reach.
What this is
The goal is to understand the whole water cycle and the law protecting its essential pathways: precipitation, snow and ice, soil, groundwater, rivers, lakes, wetlands, storage, use, return flows and water leaving Switzerland. The purpose is practical: make change visible, protect wetlands and support evidence-led legal action when protection is not enforced. This release is the surface-water foundation. It joins current river observations, reservoir history, glacier inventories, water-use registers and protected wetlands without pretending the whole cycle is live.
The organising idea is not the drawing. It is this: every figure states its class of evidence and its age. A reading from a gauge and an inference from a register two decades old are both on the screen, and they are never allowed to look alike. The bar at the top of the map is the drawing's own composition by class of evidence, counted live, and it says that roughly two per cent of what you see is measured.
Who it is for
For anyone who must know a number’s origin before using it: scientists, lawyers, journalists, public authorities and communities living with the water. The repository is public, and every useful correction should be able to improve the shared record.
It is also, deliberately, an argument aimed at the people who publish the underlying data: that a data state belongs on the face of a layer, not three clicks away in metadata.
Build this with us
Riverflow is a public, collaborative project. Its shared goal is to make Switzerland’s whole water cycle observable—from precipitation, snow and ice through soil, groundwater, rivers, lakes, wetlands, storage, use and return—so change can be seen early, protection can be checked and legal action can rest on evidence.
We invite contributions from hydrology, glaciology, limnology, ecology, water quality, geospatial engineering, public administration and Swiss environmental law. A useful contribution can be a corrected value, a primary source, a missing cantonal dataset, a method review, a translation or a clearer, more accessible interface.
What is new here
Most source data already exists. The gauging network, the residual-flow basis, the four registers of water use and the glacier inventories all exist as official federal layers, and the official products are authoritative. Two things are added.
- Propagation onto the network. The federal source publishes station points. This map draws 8'711 unique reaches and states which of three evidence classes each one has.
- One canvas, and one reading. Flow and use and law are never on a single surface federally; they are separate toggles in separate products.
- The overlaps, stated. 415 reaches here run through an alluvial zone of national importance, where the Auenverordnung's duty about the water regime and the Art. 31 GSchG floor land on the same water. Hovering the reach says so. No federal product joins those two layers, because they are two products.
This review found no Europe-wide measured equivalent. Copernicus EFAS and GloFAS provide modelled reach-level discharge; national portals stop at their borders; EStreams is a historical catalogue. No single measured European source covers the same ground.
What it is not
- Not a forecast. Measured and modelled water, as at the moment stated.
- Not a finding of breach. A datum here is a fact about a river, a reservoir or a glacier. The step to a finding is the lawyer's, and it needs the concession, the licence and the site.
- Not legal advice, and not a substitute for reading the statute. The legal page quotes the texts and cites the decisions so that both can be checked.
- Not a complete picture of Swiss water. The limits section lists what is missing and why.
How to cite it
Cite the underlying federal source for any figure you rely on; the table on the sources page gives the holder, the data state and a link for each one. Where the map's own treatment is what you are referring to — the propagation, the evidence classes, the vintage audit — cite the site and the date you read it, because the live layers change by the minute and the baked ones change weekly.
https://opengovclimate.ch/riverflow/
Read 21.08.2026.
A permanent link to a particular river and layer is the address bar: the map writes the layer and the view into the URL, so copying it copies the view.
Prior art
Checked on 19 August 2026, by fetching rather than from memory. The findings that shaped the project:
- The BAFU live map shows current discharge and level classified against the 1991–2020 norm period, and the same layer is public GeoJSON with a 300-second refresh. That is the official day-of-year statistic this map still lacks: it uses a modelled long-term mean instead, and marks it not quotable.
- A discrepancy, unresolved. The federal map page describes that classification as daily means against 1991–2020; the WMS legend for the same layer describes monthly percentiles of long-term hourly means pooled from the start of record. Two official descriptions of one layer that do not agree. Resolve it before quoting any class in a legal document.
- All four water-use registers already exist as layers on map.geo.admin.ch. The layer stack here is not new data.
The full note, with the identifiers and the counts, is in
docs/PRIOR-ART.md in the repository.
Colophon
Built by Jonas Hertner. The source is on GitHub, including every build script and the reasoning behind the numbers that had to be tuned.
The static publication uses Bash, Node and pinned mapshaper for its data build. Two canvases: the base map redraws when the view or the data changes, and the current runs on its own canvas at sixty frames a second over the top, so 3'200 particles never force a redraw of 8'711 reaches.
Type is set in three registers, one per kind of statement. Prose is Archivo. Every measured figure is IBM Plex Mono with tabular figures, so that a column of readings lines up and a number never has to be read as a word. Anything the law requires is Source Serif 4 in a warm earth against the cool ground — bone on the night surface, sepia on the day one — because a minimum residual flow is not a measurement, and must never be able to be mistaken for one at a glance. That distinction is the whole design, and it is the one thing that does not change when the surface does.
The mark at the top of every page is the three classes of evidence at their real proportions. The sliver on the left is everything this country measures.