§Research · drought & climate · August 2026
Contents
Short on time? Sections 03 and 15 carry the whole argument — the map of measures, and what it implies for the money we send abroad.
What drought is even measured by, and how to talk about it so the talk goes somewhere.
01 / The frame
When a species' environment starts to change, it has three options: change itself, change its surroundings, or leave. A fox, a beech tree, a human. And the same three moves describe what a country does.
What is it good for? It forces every proposal to state two things: which move it is, and at what scale — a garden, a field, a catchment, a country, a continent. That is exactly what public debate avoids. We argue about instruments, dams against wetlands, while each side quietly pictures a different scale. Which is why the argument never gets anywhere.
Two things belong with that. First, the options are not sequential and not exclusive. Societies do all three at once. And what decides the mix is not the severity of the drought — it is money, and how well the state works. Second, there is a fourth outcome the tidy biological trio hides: staying and failing. That is what happens to people who cannot afford to leave — and as section 15 shows, they are the majority.
02/ The four outcomes
That question does most of the work by itself. An example: expanding irrigation just so we can keep growing the same crops as before is changing the surroundings to dodge a decision about crops. And it locks in a permanent thirst at the very time there is less water.
Different crops, different working hours, different buildings. The cheapest option, and the one we talk about least — it needs no construction and no permit.
Hold water where it falls, or store it behind a wall. This is where almost the entire public argument happens, usually without saying at which scale.
The one everybody fears and nearly nobody analyses. Migration is expensive, so the most exposed people are frequently the least able to do it.
The option the animal comparison hides. For the poorest people it is the most common one: they cannot afford to leave.
03/ The map
Scale runs across, the move runs down. The “who decides” band is there on purpose: the further right, the less of it is up to you. Almost the entire public argument fits into two cells in the middle — reservoirs against landscape.
| Garden | Field | Catchment | State | World | |
|---|---|---|---|---|---|
| who decides | the owner | the farmer | municipality + catchment authority | government | nobody alone — only by agreement |
| Change yourself the cheapest move, and the least discussed | Different trees, mulch, no lawn small effect on the country ≈ 0; on your own water bill, real | Crop mix, sowing dates, farming practice big lever needs neither construction nor a permit | At this scale the move makes no sense. | The price of water big lever CZK 2–3 per m³ of groundwater, unchanged since 2001 — not even for inflation | What we import big lever ~4,500 l/day embedded in goods against 88 l from the tap |
| Change the landscape almost the whole public argument fits here | Soaking rainwater in small the effect stops at the property line | Organic matter in topsoil, no compaction big lever +1 % plant-available water across arable land is of the same order as the usable volume of the largest Czech reservoir | Field margins, wetlands, drainage audit big lever 1.1 m ha drained, most of the pipework past its service life | Reservoirs medium storage for abstraction; no effect on soil moisture above the dam | Adaptation finance big lever ~CZK 2 per person per year to the GCF, nothing ever to the Adaptation Fund |
| Leave what everyone fears and almost nobody quantifies | At this scale the move makes no sense. | Giving up the driest land unknown how many hectares this concerns, I found no documented figure | No measure here that I can evidence. | It happens, but nobody tracks it. | Migration unknown climate change amplifies the problem; a clean causal chain from drought to a person at the border does not survive the data |
| Stay and fail not a move — it is the result when the other three do not work | At this scale the move makes no sense. | Farming that runs on debt never reported as a consequence of drought | It happens, but nobody tracks it. | It happens, but nobody tracks it. | Those who cannot afford to move successful adaptation elsewhere may raise the numbers leaving for a while, not lower them |
How to read it: each cell is one measure for Czechia, placed by what it changes and how far it reaches. The figures are unpacked in sections 09 to 15. ⚠ The lever is my own ranking drawn from the sources, not a measured quantity — which is why it is kept separate from evidence, which is about the sources. Blank cells carry three different meanings, because “makes no sense”, “nobody tracks it” and “I have no evidence” are not the same thing. The last row sits below the line on purpose: nobody chooses to stay and fail.
Why more rain means less water — and why it does not worsen gradually but breaks.
04 / The physics
Warm air can carry more water vapour. For every degree of warming, roughly 7 % more water fits into it.
But that much water does not actually get in. Evaporating it costs energy, and there is less of that around than would be needed. The amount of water genuinely moving through the cycle grows only about 4.3 % per degree. So the air can hold more but is not carrying more — and that gap is its thirst. The technical name is vapour pressure deficit; in drought reporting you will more often meet the broader atmospheric evaporative demand, which wind, sunshine and humidity drive as well as heat. The bigger that thirst, the harder the air pulls water out of every leaf, every puddle and every patch of soil. Over Europe it has been measurably rising since the 1980s, at roughly twice the rate of model runs in which humans are absent.
You know it from a washing line: clothes dry faster when it is warm, and fastest when the air is dry as well.
This is the whole article in one line: climate change is not mainly taking away the water that arrives in Czechia. It is speeding up the rate at which the air takes it back.
Soil and air talk to each other, and they can be in two different states. In the wet state Czechia has always been in, there is plenty of water in the soil and only the available sunlight decides how much evaporates. A dry spell runs its course and ends.
In the transitional state now moving up from the south it is the other way round: water is scarce, so water is what decides. That closes a loop. Dry soil has nothing to cool itself with, so more sunlight goes into heating the air. Hotter, drier air pulls harder on the soil. The soil dries further. Round and round. The summer of 2003 was the first big demonstration: the heat was extreme partly because the soil had already dried out.
Three consequences follow, and all three show up in Czech data:
One more thing: European heatwaves are growing three to four times faster than elsewhere at our latitude, driven by circulation changes that climate models under-simulate. So the projections are more likely cautious than overblown. Plan against the worse case, not the average.
What we remember from our own dry years, and where that memory misleads us.
05/ Why it breaks
Soil and air talk to each other, and they can sit in two states. The difference is not the strength of the drought — it is whether the last step feeds the first.
where Czechia has always been
There is enough water in the soil, so evaporation is set by solar energy alone.
The dry spell runs its course and ends.
moving up towards us from the south
Water is scarce, so water sets the evaporation — and that closes the loop.
and round it goes — the drought drives itself
Three things follow, and all three are visible in Czech data: drought accelerates itself; heat and drought stop being two separate threats; and annual rainfall stops being a measure of drought. The summer of 2003 was the first big demonstration — the heat was that extreme partly because the soil had already dried out.
06/ How much already
Individual years jump around, which is exactly why drought can be argued about for ever one year at a time. The line is what decides.
07/ The years you remember
| Year | what it was | why |
|---|---|---|
| 1947 April to November | drought, failed harvest | it barely rained |
| 1997 and 2002 Moravia, then Bohemia | floods | water all at once |
| 2003 all of Europe | heatwave | heat plus dry soil |
| 2015 Písek rationed | drought | little snow, heat |
| 2018 bark beetle takes off | drought, calamity | evaporation, not rain |
1947 is still the Czech benchmark: from April to November it effectively did not rain, ponds and wells dried up, the harvest failed, and specialists still cite it as the worst drought in the rivers in the whole measured record. Detailed records from it barely exist, so nobody can compare it precisely with today. 2018 was, per the Czech Hydrometeorological Institute, the second-driest year since records for this began in 1961 — behind 2003.
08 / Memory
"When I was a child there was snow every winter." This one is true, and it matters more than it seems. Comparing 1991–2020 with 1961–1990, roughly 40 cm less snow falls here per year. Snow is not scenery — it is a reservoir with a slow outlet. It melts over weeks in spring and feeds soil and groundwater the entire time. Rain falling instead of it is gone in days. So what is disappearing is precisely the kind of storage the landscape needs most.
"We've had droughts before, it'll pass." Half true. Individual dry years always happened, and 1947 was probably worse than anything since. What is new is different: drought stopped being erased. The top few centimetres of soil recover within days of good rain. The deeper layers and the groundwater refill mainly in the cold half of the year, when almost nothing evaporates. When that half of the year is warmer and snowless, the missing water carries over from year to year and adds up. Which is why across 2014–2018 more water was missing from Czech soil than in any stretch in 253 years.
"Once it rains properly, the drought is over." Not any more. When it rains the grass goes green and everyone relaxes — but that is only the top layer. The store that wells and rivers draw on in August is refilled somewhere else, at another time of year.
Here is the break that makes "we've seen this before" insufficient. 1947 was a drought caused by missing rain. Today's is not. In 2018 roughly the normal amount of rain fell — and there was still a record amount of water missing from the soil. Those are two different diseases with the same symptom.
Two practical consequences. First, the cure that ended 1947 no longer works reliably. Back then a wetter year came along and that was that. Now water is lost even in years with normal rainfall. Second, and more serious: the rules changed, not just the weather. Statistics of the "a drought this bad comes once in twenty years" kind describe a world running in that wet state. We are moving into a different one. So numbers about how often things used to happen are more likely to understate future risk than to overstate it.
Experience is still worth something — it tells you what a drought looks like. It does not tell you how often one will now arrive, how long it will last, or what will end it.
09/ Four numbers
All four are well established, and all four point somewhere the usual debate does not go.
Roughly twice as fast as the world as a whole. Rainfall over the same period has not fallen, and is not expected to. Water is disappearing because more of it evaporates.
Across 2014–2018, more water was missing from the soil than in any stretch since 1766 — while 2018 rainfall sat inside the normal range. Annual rainfall totals are not a drought indicator here.
We use about 88 litres a day from the tap. Another 4 500 or so are used elsewhere to make our food and things, mostly abroad. Against that, shorter showers barely register.
Anyone pumping groundwater pays the state 2–3 Kč per cubic metre. That figure has not moved since 2001, not even for inflation. And it is lower than for river water — so the scarcer water is the cheaper one.
Czechia as the roof of Europe: nothing flows in, so residence time is what decides.
10/ Where the water is actually used
Both figures are daily use per person. The difference is not in how much water we see — it is in where it gets used.
51×how much more water we use indirectly than directly
From the tap at home 88 l / day
what we see and talk about — showering, laundry, flushing
Embedded in what we buy ~4,500 l / day
virtual water in food and goods; mostly used abroad, often where there is less of it than here
1 drop = 88 litres, i.e. an entire day's use from the tap
This is why savings at the tap are negligible against the water embedded in goods. It does not mean saving at home is pointless — it means what we buy is what decides, and that we export that consumption elsewhere. Zdroj: ČSÚ, Water management in Czechia (88 l); virtual water per research §4
11 / Where the water goes missing
Czechia is the roof of Europe: no significant river flows in. Every available drop fell here and has not yet left, which makes residence time the master variable — and residence time is set by the landscape, not by the sky.
Over the 20th century that residence time was engineered decisively downwards. Not maliciously: drainage, consolidation and channel straightening were the right answer to the problem of the day. The problem changed and the infrastructure did not. More than 1.1 million hectares — about a quarter of Czech farmland — is drained, through an estimated million kilometres of buried pipes. Most are at the end of their design life, and many have no clearly accountable owner. Over half of arable land is threatened by water erosion, with no systematic protection on most of it.
This matters more than it sounds, because topsoil is the largest and cheapest water reservoir the country owns. Its capacity is a function of organic matter, structure and depth — all three of which erosion and compaction are steadily removing. Add one percentage point to the water plants can reach across Czech arable land, and you have roughly what the country's largest dam usefully holds. And it costs no valley.
The bark beetle calamity was enormous — salvage logging was still 78.8 % of total harvest in 2022. You would think that losing the forest meant losing the water. The data do not bear that out. In a Šumava catchment, a mosaic of management regimes from non-intervention to clear-felling produced no statistically significant difference in annual runoff or evapotranspiration. And because conifers intercept and transpire more than broadleaves, the species conversion now being forced on Czech forests is, in pure water-yield terms, favourable.
A lesson that holds elsewhere too: with water, people often get wrong even whether something helps or harms — never mind by how much. And the effect is almost always smaller than the talk about it. Measure, don't narrate.
12/ The false choice
Each one holds water in a different place, for a different length of time, and each fails differently. Which is why it makes sense to have both, not to choose.
| Property | Reservoir | Soil & landscape storage |
|---|---|---|
| Helps against | too little in the river | dry soil |
| Buffers for | weeks to a few years | days to weeks |
| Spatial reach | one point on a river | the whole land surface |
| Dry-year losses | evaporation off the surface | water used by plants — the point |
| Can it be undone | no — a valley, decades | mostly yes |
| Politically | one project, one ribbon | a million parcels |
A dam does not make water. It holds it back and releases it later. It can only work with water that would have flowed past the dam site anyway — so on the moisture in fields above it, the effect is exactly zero. Czech studies reckon climate change will take typically under 30 % off the reliable yield by 2050. Put simply: a dam secures water for people, the landscape secures water for the whole country.
Three countries hit this before us. Their failures travel better than their successes.
13 / What the world already learned
Three countries hit scarcity before us, chose different instruments, and are far enough along that outcomes are visible. The failure modes transfer better than the successes.
Spain built the Tajo–Segura transfer. Wells in the receiving area went from 7 829 to 20 350, and most of them are registered nowhere. Forty of sixty-three groundwater stores were seriously damaged. In the worst places 40 % of irrigated land is illegal. And the Mar Menor lagoon suffocated on the fertiliser that ran off those fields. The infrastructure worked exactly as designed and made the underlying problem worse. New water arrived, irrigated land spread far past anything the project intended, and people took the extra water from underground.
This is called the reservoir effect, and here it played out across a whole country. More water tempts people into using more, and that eats the whole benefit. And the more people depend on that store, the greater the damage when it finally runs out.
14/ Three natural experiments
| Country | instrument | demand capped? | outcome | cost fell on |
|---|---|---|---|---|
| Australia Millennium Drought, 2001–09 | rules | yes — capped | use under control | farming regions |
| Spain Segura basin + Tajo transfer | construction | no | shortage continued | groundwater, wetlands |
| Israel desalination + reuse | technology | partly — by price | enough water | energy, coast |
Australia capped how much could be taken, and paid for it visibly: dairy farms in the Murray region fell from 4 500 to about 850. Spain set no cap and paid differently — quietly, and irreversibly, in wrecked groundwater and wetlands. Israel partly replaced a cap with price and central control, and moved the cost into its energy system; desalination is so energy-hungry that the IPCC cites it as an example of a fix that worsens the problem later. So what decides the outcome is the cap on use, not the choice of technology.
15 / The uncomfortable part
Water does not stop at a border and neither does the argument. Two-thirds of humanity already lives under severe water scarcity at least one month a year. The Mediterranean — unlike Czechia — is genuinely losing rainfall, and that is known with very high confidence. Southern Europe is not a preview of our weather. It is a preview of our neighbourhood.
The migration argument has to be made carefully or it backfires. Syria is the example people reach for: drought supposedly triggered the civil war. That explanation did not hold up. A 2017 reassessment found no convincing evidence that climate change caused that drought. It also showed the famous figure for people driven from their homes was ten times the UN's estimate at the time. The argument is still unsettled today. Anyone arguing a clean causal chain from a drought to a person at a border will lose that argument on the evidence.
The defensible version is narrower and stronger: climate change amplifies problems. It takes water and harvests away from states that are fragile already. And one more thing belongs with it, which sounds surprising. When adaptation elsewhere succeeds, the number of people moving may go up for a while. Moving costs money, and the poorest do not have it yet. If your support for adaptation finance depends on a headcount falling, the numbers may not cooperate. Make the case on stability and avoided suffering instead.
What is not in dispute is the price. The World Bank's own modelling says decisive mitigation plus green, inclusive development would cut projected climate migration by as much as 80 %. Adaptation investment returns two to ten times its cost. International public adaptation finance currently runs about twelve to fourteen times below assessed need — and is falling.
For a small rich country the sums are blunt. Czechia emits about 9 tonnes of carbon dioxide per person a year. The EU average is around 7. For its own climate measures it is receiving roughly 450 billion CZK from EU funds. And to the Green Climate Fund, which pays for adaptation in poorer countries, it sends about 2 CZK per inhabitant per year. To the Adaptation Fund it has never sent anything. There is no way to do these sums where our ponds and field margins matter more to the world than our emissions and the money we send. Both are worth doing. Only one of them is being done, and it is the one that only helps us.
What this is, how deeply it was read, and what I could not track down.
16 / Method, and what I got wrong
This is a deep research pass — roughly seventy-eight sources across atmospheric physics, Czech hydrology, soil science, forest hydrology, water economics, migration studies and three comparative national cases. Each source is graded by how far it was actually read: full text, abstract, or search summary only. No claim that carries an argument rests on a source read at summary level.
Roughly a third of the material was read at abstract level, and that is stated rather than hidden. Two key Czech documents were downloaded and read in full: the national water-modelling database behind the runoff projections, and a factsheet on climate finance. Both carry numbers the conclusions depend on.
Publishing the gaps is part of the job. A text that lists only what it found is closer to advertising. One that admits what it could not pin down can be built on.