What the precipitation deficit does not measure

At the end of July 2026, the national precipitation deficit in the Netherlands — the cumulative difference between evaporation and rainfall that serves as the country’s standard drought indicator — stood at nearly 250 millimetres. A news item translated the figure for its readers: twenty-five ten-litre buckets per square metre.¹ The image is vivid and it is wrong, not in the arithmetic but in what it suggests. No twenty-five buckets have disappeared from the system.

More interesting than the error is where it comes from. The same news item contains everything that actually matters for the governance of water, and it sits in the surrounding paragraphs. The Rhine is delivering little water, there is hardly any meltwater, the salt front is pushing further into the province of Zuid-Holland — the urbanised west around Rotterdam and The Hague — through the Nieuwe Waterweg and the Hollandsche IJssel, the open waterways that link the sea to the inland system; fifteen thousand litres of fresh water per second are being let in from the Amsterdam-Rhine Canal to flush the polders; and on the Twente Canal, in the east of the country, the locks are staying shut to hold water back. Inflow, stock, distribution and quality. Precisely the four quantities that do not appear in the summary figure.

This piece is about how an indicator that measures one thing can come to carry the conversation about something else, and why this particular figure was the one that got the job.

A demand quantity, not a balance

The precipitation deficit — the figure kept by the Royal Netherlands Meteorological Institute (KNMI) — is the daily sum of reference crop evapotranspiration minus the rain that fell, averaged over thirteen measuring stations, cumulative over the growing season from 1 April to 30 September and, since 2026, also tracked year-round.² That reference evapotranspiration is calculated by the Makkink method and models the evaporation of short-mown grassland on a soil amply supplied with moisture. The figure thus describes what the atmosphere would have liked to extract from a crop with unlimited water at its disposal.

That makes it a demand quantity, not an entry in a balance. As soon as the soil dries out, actual evaporation drops below the reference: the plants close their stomata and there is nothing left to evaporate. At precisely that moment the counter keeps running at full speed. The physically deeper the drought becomes, the further the indicator departs from the process it is supposed to describe. That is not a measurement error but a property of the construction. The buckets describe a demand that was largely never met.

A single calculation rule settles the difference with a balance. If the sum falls below zero, it is reset to zero. A wet spring is not carried into the summer. A balance books surplus and deficit to the same account; this counter knows only deficit and starts afresh after every wet spell. Whoever reads it as a measure of stock is reading a quantity that by construction cannot contain one.

What is sharply in view

The flow is well covered, and the precision with which that happens is the mirror image of the vagueness elsewhere. On 16 July the Rhine discharge at Lobith — the point where the river enters the Netherlands — stood at around seven hundred and fifty cubic metres per second, below the minimum of roughly seven hundred and eighty from the dry summer of 1976, the standing reference drought of the Dutch series, having still been around a thousand at the start of July.³ At the end of July the national government reported that the Rhine at Lobith had never been measured this low.⁴

The administrative scaling-up follows those figures step by step. On 1 July the National Coordination Committee for Water Distribution (LCW) — the body through which the national government and the regional water managers coordinate scarce water — moved to level one, impending water shortage, and on 16 July to level two, actual water shortage, at which point the Water Shortage Management Team, chaired by the director-general of Rijkswaterstaat (the Directorate-General for Public Works and Water Management), meets weekly.⁵ Since then the weirs on the Rhine branches have been operated differently, ships are locked through less often to limit salt intrusion, and deployment of the weir at Hagestein is on the table.

These figures are not model outputs but measurements, at fixed points, by parties held operationally to account for them every day: navigable depth, lock management, intake capacity, security of drinking water supply. Where an organisation has to act, measurement is sharp — up to and including the finding that a record has been broken.

What is not in view

The stock is another story. Groundwater is measured as hydraulic head in monitoring wells and summarised in an index. That is a level, not a volume. The translation into a stock runs through the National Hydrological Model, which was set up primarily for national applications and expressly not for detailed regional or local analysis.⁶ A statement about the available freshwater stock is thus a model outcome, and precisely at the scale on which the decisions fall, the model is not meant for it.

On the abstraction side the gap is larger and more concrete. Provinces are the competent authority for the large groundwater abstractions; water authorities (waterschappen) — the elected regional bodies that manage dikes, water levels and water treatment — for the smaller ones by farmers, private users and industry; and for a considerable share of those smaller abstractions no permit or notification requirement exists, so they are not registered.⁷ When the House of Representatives — the lower house of the Dutch parliament — asked in June 2020 for insight into the number and size of groundwater abstractions, the answer came in March 2021, and its core was that the requested insight cannot be given in full.⁸ The Court of Audit for the eastern provinces (Rekenkamer Oost-Nederland) moreover ran into a units problem in Gelderland and Overijssel that blocks comparison from the outset: water authorities set their registration thresholds in pumping capacity per hour, provinces in annual totals, and the two cannot be converted into each other, because capacity is not the same as quantity pumped.⁹ Whether that problem takes the same form outside those two provinces has not been established; the division of competences that causes it applies nationwide.

Here the measurement architecture follows the division of competences, not the water system. Whoever wants a judgement on the flow measures for himself. Whoever wants a judgement on the stock must piece together twelve provinces, twenty-one water authorities and a model, each registering in different units and at different moments.

Why precisely this number carries the conversation

Seen this way, the precipitation deficit is not an unfortunate choice but a predictable one. It is the only drought quantity that is available daily, applies nationwide, fits into a single number and has a reference series behind it reaching back to 1906. The same chart shows the median, the line exceeded in five per cent of years, and the course of 1976.¹⁰ Hence the reporting arrives effortlessly at the five per cent driest years and at 2018, the previous severe drought year: those comparisons come ready-made with the instrument. Stock, supply capacity, storage capacity and abstraction rights are regional, slow, spread across managers and impossible to capture in a single series.

The number that is easiest to count becomes the number the conversation is about. The indicator is not chosen because it describes the problem, but because it exists. Its persuasive power comes precisely from the property that makes it unsuitable: it has the shape of a water balance — in millimetres, cumulative, with a percentile position — while it contains only one entry, and that entry is atmospheric. A substitute that copies the form of the original without its function is harder to recognise than a substitute that looks different.

This year’s innovation shows how persistent that is. Since 2026 KNMI has monitored the precipitation deficit year-round as well, in line with advice from the Expertise Network on Freshwater and Drought, and it justifies the step with the winter of 2018 on 2019, when groundwater levels were insufficiently replenished after the dry summer and the 2019 growing season therefore started at a disadvantage.¹¹ The reasoning is purely hydrological and is entirely about the stock. The answer is an extension of the time axis of the same atmospheric quantity. Whoever wants to track stock phenomena gets a longer measurement of demand.

That a national average over thirteen stations sets the picture while every decision that follows from it is regional is the direct consequence. The same 250 millimetres means one thing in the areas that can be fed with river water and something else on the elevated sandy soils of the east and south, where no water can be brought in at all. The only distinction that means anything operationally — between areas that can be supplied and areas that drain freely — disappears into the average that carries the conversation.

What the figures relieve you of

A precipitation deficit has no addressee. It is weather, and no one is responsible for weather. Water management, by contrast, is an outcome of decisions: level management, drainage, permitted abstractions, spatial planning, and decades in which water was discharged as quickly as possible because discharging solved yesterday’s problem.

KNMI’s climate bulletin of 23 July shows how precisely that separation is drawn and how easily it then blurs. The bulletin itself distinguishes three kinds of drought — meteorological when little rain falls, agricultural when crops grow poorly and the soil is dry, and hydrological when river discharges and groundwater levels are low — and establishes that this summer all three are occurring.¹² The same institute supplies the single number in which the last two do not appear. Both registers stand on the same website, and precisely that makes the finding tenable: the knowing and the summarising live in the same organisation and do not correct each other.

More telling still is what the underlying study computes with. World Weather Attribution assesses the drought by the soil moisture in the top metre, compared with all years since 1950, and concludes that the soil moisture deficit in the eastern region to which the Netherlands belongs has become roughly eleven times more likely through human-caused warming. It comes with the finding that closes the mechanism: average precipitation in the growing season has hardly changed; the drought has become more likely mainly through higher temperatures and the evaporation that accompanies them.¹³ The moment science genuinely wants to make a statement about how dry it is, it switches to a stock quantity. In the same bulletin the precipitation deficit appears only in the scenario chart.

Something uncomfortable for the indicator follows from this. Climate change loads almost exclusively the evaporation term, and that is the term that drives the deficit’s counter. The number will therefore keep setting records for a reason that says nothing about the capacity of the water system. Its most alarming signal is at the same time its least actionable.

The finding that without climate change it would not have been this dry is scientifically grounded and at the same time the perfect cover, because it places the cause at an address no Dutch water manager is in charge of. No one needs to intend it that way, and there is no need to read it that way either. It is enough to note that the cover is available free of charge and that no party has any incentive to give it up. KNMI itself, for that matter, does not use it: the bulletin ends with adaptation — laying in freshwater stocks while also providing storage for extreme rainfall. That closing sentence points straight back to the decision layer this piece is about, and it rarely reaches the public debate.

What this asks of the reporting

Two things can be done without changing the law, and both concern how the figure is presented. Do not report the precipitation deficit without the state of the stock beside it, even if that state is available only as a level. And split every national drought picture as standard into areas that can be supplied with river water and areas that drain freely, because that is the only division that leads to different actions.

What happens once this indicator meets administrative reality — how scarcity is distributed with an instrument covering only one of the two stocks, and why the weighing behind such a decision is not published — is worked out in Distribution Without a Balance on Statecraft.

---

Footnotes

1. Job van der Plicht, ‘Ingrijpende maatregelen naderen door wekenlange droogte: zitten op het randje’ (“Far-reaching measures approach after weeks of drought: right on the edge”), NU.nl (a Dutch news site), 30 July 2026. The figure concerns the national average precipitation deficit as of that date. ↩

2. KNMI, Achtergrondinformatie neerslagtekort (background information on the precipitation deficit), knmi.nl, consulted 2 August 2026. The sum is reset to zero as soon as it falls below zero, so the value cannot become negative. The thirteen stations are spread across the Netherlands, from De Bilt to Westdorpe and Winterswijk. ↩

3. KNMI climate bulletin, Sjoukje Philip and Sarah Kew, ‘Zonder klimaatverandering was het nu niet zo droog geweest’ (“Without climate change it would not have been this dry”), 23 July 2026. Compare Unie van Waterschappen (Dutch Water Authorities), ‘Droogte en lage rivierafvoeren vragen om verdergaande maatregelen’ (“Drought and low river discharges call for further measures”), July 2026, which reports the national precipitation deficit of 200 mm around mid-July, comparable to 2018, and groundwater levels predominantly low to very low for the time of year. ↩

4. Ministry of Infrastructure and Water Management, ‘Droogte blijft toenemen: Rijn op laagste punt ooit’ (“Drought keeps increasing: Rhine at its lowest point ever”), rijksoverheid.nl, 30 July 2026. ↩

5. Rijkswaterstaat, ‘Droogte houdt aan: de Landelijke Coördinatiecommissie Waterverdeling schaalt op naar dreigend watertekort’ (“Drought persists: the National Coordination Committee for Water Distribution scales up to impending water shortage”), 1 July 2026, and ‘Extra maatregelen door aanhoudende droogte, opschaling naar feitelijk watertekort’ (“Additional measures due to persistent drought, scaling up to actual water shortage”), 16 July 2026. The current status and the regional measures in force are published in the Drought Monitor of the Netherlands Water Management Centre, waterberichtgeving.rws.nl. ↩

6. Netherlands Hydrological Instrumentarium, documentation of the National Hydrological Model, nhi.nu; see also STOWA, Deltafact Inzetbaarheid hydrologische modellen voor waterkwantiteitsvraagstukken (usability of hydrological models for water-quantity questions). ↩

7. Report Melding- en vergunningplicht (kleine) grondwateronttrekkingen (notification and permit requirements for (small) groundwater abstractions), presented to the House of Representatives in 2024. Small abstractions are defined there as abstractions of up to 150,000 m³ per year. ↩

8. Interprovinciaal Overleg (the association of the twelve provinces) and Unie van Waterschappen, Overzicht Grondwateronttrekkingen provincies en waterschappen (overview of groundwater abstractions by provinces and water authorities), presented to the House of Representatives by the Minister of Infrastructure and Water Management on 29 March 2021, following a request from the committee debate of June 2020. The report estimates the number of groundwater abstractions for irrigation at roughly 15,500. ↩

9. Rekenkamer Oost-Nederland, Als elke druppel telt (When Every Drop Counts), presented to the provincial parliaments of Gelderland and Overijssel, 28 September 2022. ↩

10. KNMI, Achtergrondinformatie neerslagtekort. The calculation of the national average was updated in 2026; record values, the median and the line of the five per cent driest years are based on the 1906–2025 series. ↩

11. KNMI, ‘Vernieuwde berekening neerslagtekort: nu hele jaar droogtemonitoring’ (“Revised calculation of the precipitation deficit: drought monitoring now year-round”), January 2026; Expertise Network on Freshwater and Drought, Voorbij de seizoenen: jaarrond voorbereid op droogte (Beyond the Seasons: Prepared for Drought Year-Round). ↩

12. KNMI climate bulletin, 23 July 2026. The distinction between meteorological, agricultural and hydrological drought is elaborated there. ↩

13. World Weather Attribution, Increasingly hot Europe faces more severe droughts and growing challenges for water and land management, July 2026, led by KNMI researchers. For the western region the study arrives at a roughly five times greater probability at 1.4 degrees of warming since the start of the last century. ↩