Low water
How much water are rivers and aquifers carrying right now, measured against what is normal at the same place at this time of year? And at what point does that say anything about the climate?
This page treats low water the same way as wildfire: four levels, strictly separated. What you see here is the current situation. It is compared against the years 1991 to 2020, yet it is still not a climate statement. That only emerges from the record across decades.
Four levels, and only two carry climate statements
Low water can also be viewed on four levels. They are kept strictly apart because they say different things.
1 · Live situation
not a climate statementCurrent readings for water level, discharge, groundwater level and spring flow. A snapshot, not a trend.
2 · Low-water classes
not a climate statementFour uniform classes from “no low water” to “extremely low”. An assessment of the situation, not a climate statement.
3 · Historical records
climate statementLow-water days since 1991, by half-year and river basin. Here, and only here, does the climate statement emerge.
4 · Projections
climate statementFuture discharge under scenarios, with a mandatory uncertainty band. A climate statement about the future.
All four levels below are backed by data. One limitation is deliberate: level 4 shows no projected low-water days. The DWD reference ensemble provides climate variables, not discharge. What is projected is therefore the condition, not the event, exactly as separated in level 3. For discharge itself, which also depends on river engineering and abstraction, there is no sound nationwide projection.
Level 1
Where measurements are being taken
Each dot is a monitoring station, coloured by its class today. This is the current situation and explicitly not a climate statement. That comes further down in level 3. The map is deliberately separate from the emitters map: causes and consequences do not belong in one layer.
Level 2
The current classification
NIWIS assigns each monitoring station to one of four classes every day. The comparison is always against the same station on the same calendar day during 1991 to 2020.
Level 3
The observed record: low-water days since 1991
A low-water day is a day on which discharge falls below MNQ, the mean low-flow discharge of the reference period. A transparent threshold count on official daily values, not an estimate of our own. The indicator follows WW-I-6 of the German adaptation strategy. Averaging across river basins is weighted by catchment area: the Rhine and Elbe together account for 83 per cent.
Five-year means
Low-water days per half-year, weighted mean. Summer 1 Apr to 30 Sep, winter 1 Oct to 31 Mar.
Why there is no trend line here
A regression across all 35 years would yield a tidy rise of about 0.6 days per year. Remove the three strongest years and 0.16 remains, essentially nothing. So the increase does not sit in a steady slope but in a cluster of extreme years since 2018. A line would hide that and imply an extrapolation the data does not support. Five-year blocks show what actually happened.
Keeping condition and event apart
As with wildfire: the event (low-water days) depends not only on climate but also on river engineering, abstraction and monitoring practice. The condition, by contrast, is pure weather. So both are worth looking at.
Summer precipitation in Germany
0 → 0 mm
First versus second half of the period. Practically unchanged.
Hot days per year
0 → 0
First versus second half of the period. Markedly more.
What this record cannot do
- 35 years is short for a climate statement. The period starts in 1991 because NIWIS reaches no further back, not because anything began there.
- The monitoring network is not identical across all years. NIWIS excludes stations with more than ten per cent missing values, but which stations that affects can change from year to year.
- Discharge is not a pure climate signal. River engineering, impoundment and abstraction for agriculture, industry and drinking water all play a part. A gauge measures what arrives, not what fell.
- These figures are not the official WW-I-6 indicator. NIWIS deliberately deviates: a shorter record, all available stations rather than a fixed set, and weighting by gauge catchments.
When the consequence hits the causes
Low water is the hazard where the circle closes. Thermal power stations need cooling water, from the very rivers whose gauges are coloured above. If water becomes too scarce or too warm, output has to be reduced, in individual cases down to shutdown. The same holds for inland shipping, which carries coal and oil, and for industrial sites that need process water. Anyone looking at the large energy-sector dots along the Rhine, Elbe and Danube in the atlas is also looking at plants that depend on what is running short here.
Deliberately qualitative only. A map linking power stations and gauges by distance would assert an exposure that proximity alone does not establish: it depends on the cooling method, water permits and the specific water body. As long as there is no sound data basis for that, this states a relationship, not a number.
To the emitters atlasWhat helps against it
Low water is the case where mitigation and adaptation differ most visibly. Cutting emissions addresses the cause, but slowly. Retaining water in the landscape, unsealing surfaces, sponge-city measures and peatland rewetting address the consequences, right where the water is missing. Both are needed, and neither replaces the other.
- ·Retaining water in the landscape: rain that infiltrates slowly instead of running off quickly refills the groundwater store that sustains summer flow.
- ·Unsealing and sponge-city measures: sealed surfaces divert rainfall into the sewer system, removing it from the water balance.
- ·Peatland rewetting: works twice over, as a store of water in the landscape and as a carbon sink. One of the few levers that addresses cause and consequence at once.
- ·Forest conversion: mixed forests hold water better than pure conifer stands, and they burn less readily. The same lever acts on both hazards covered by these pages.
Conflicting aims
Adaptation that can make the problem worse
Not every measure taken against a consequence of climate change improves matters. Some shift the damage elsewhere; some reinforce the very cause they are aimed at. There is a term for this: maladaptation. For low water it can be shown with one case, and that case also explains why our rivers became this vulnerable in the first place.
How the rivers got into this state
The starting point is not climate change but the 19th century. Four steps, each documented individually by public authorities:
- 1Straightening and narrowing. A straightened course is shorter and therefore has a steeper gradient over the same drop in elevation. The regional authority of Baden-Württemberg, on Tulla's Rhine correction (1817 to 1879): erosion of the river bed followed from the shortened flow distance.
- 2No replenishment. Weirs and barrages hold back bed load. The Bavarian Environment Agency calls this a bed-load deficit, from which incision of the river bed follows.
- 3The floodplain loses its water. The same agency, verbatim: as the river bed incises, groundwater levels and the frequency of floodplain inundation fall with it.
- 4What is left. Only about one third of the former floodplain area can still be flooded at all today. Of the floodplains that remain, nine percent count as near-natural, 58 percent as heavily or very heavily altered.
That removes the retention which the section above names as the remedy. A deeply incised river in a disconnected floodplain drains water faster and stores less. Drought hits such a landscape harder.
The case: dredge deeper
When low water stops ships from sailing fully loaded, deepening the fairway is the obvious answer. This is where the chain closes on itself: deepening and narrowing accelerate the same erosion that destroyed the retention. The measure acts on the symptom and on the cause at once, only in the wrong direction.
Two rivers, two orders of priority
Notably, the federal government answers this question in opposite ways on two rivers. For the Elbe, the 2013 key-points paper states that no further development to improve navigation conditions will take place. The target fairway depth of at least 1.40 metres on 345 days a year applies expressly only in so far as it does not impede efforts against bed erosion. On the Middle Rhine, by contrast, deepening to a continuous 2.10 metres has been a statutory priority need since December 2016. There the fairway takes precedence; on the Elbe, the river bed does.
The case for inland shipping
The other side has strong arguments, and they belong here rather than in a footnote. When the Rhine ran dry in 2018 the effect was measurable across the economy: according to the Kiel Institute for the World Economy, low water depressed industrial output by around 1.5 percent at its peak and cost roughly 0.4 percent of annual economic output. And inland vessels emit far less than lorries: 32 against 118 grams of CO₂ equivalent per tonne-kilometre (German Environment Agency, reference year 2024, including upstream energy supply).
That figure has a third column, though. Rail freight comes in at 14 grams, less than half the inland vessel. “Climate-friendly mode of transport” holds against the lorry and does not hold against rail. Anyone wanting to shift freight therefore has more than one option, and the question of expanding the waterway is not the same as the question of shifting freight off the road.
Where this figure belongs
The 0.4 per cent of economic output above is a figure for one event, from a calculation of its own by the Kiel Institute for the World Economy. What such damages look like taken together is on a page of its own: the study commissioned by the federal economics ministry puts the drought and heat summers of 2018 and 2019 together at around 35 billion euros. Two different calculations, two different boundaries. They are not added up.
What climate change has cost so farWhat is discussed instead
- ·Bed stabilisation and bed-load feeding address the incision rather than deepening it. On the Elbe this has been the primary strategy for years.
- ·Removal of structures, dyke relocation and reconnecting floodplains tackle the chain at its beginning. The Integrated Rhine Programme, for instance, is creating retention areas in former floodplains at 13 sites on the Baden-Württemberg bank, under a treaty with France.
What is established, and what is not
- The effect on flooding is established and quantified. The Federal Environment Agency reports a restoration on the river Nebel that enlarged flood retention capacity by around 16 percent; a technical alternative would have cost some 174 million euros. The Integrated Rhine Programme, too, is expressly a flood protection programme with ecological co-benefits.
- The effect on low water is not. The Federal Environment Agency puts it cautiously: large-scale restoration “can contribute” to preventing low water. We are not aware of a study quantifying that effect. We report it as it stands rather than more sharply, although the sharper version would suit this section better.
- Restoration does not create water. It improves retention and timing. It does not remedy a rainfall deficit, any more than deepening does.
- The bottleneck is land. The former floodplain now belongs to farming, settlements and infrastructure. That, not a lack of knowledge, is where such projects fail. And the conflict with navigation does not dissolve simply because the ecological side sounds more appealing.