Where South Africa's water goes

From every drop of rain to the tap and the field — the balance, the losses, and the biggest levers to save water and widen access. Every figure below is cited to a primary source (DWS, StatsSA, WHO/UNICEF JMP).

v5 · compiled 2026-07-18 · volumes in million m³ per year (Mm³/yr) unless stated · click a ⁿ to jump to its source · click any §8 mitigation tag for its definition · §12 carries the programme's early findings

450–465 mm

Mean annual rainfall — about half the world land average³

~9%

Share of rainfall that becomes river runoff (≈90% evaporates)⁹

89% → 64%

Households with water infrastructure vs. a reliable supply³

47.4%

Municipal water that is non-revenue (leaks, theft, unbilled), 2023⁵

1 · The cascade — rain to usable water

Two stages, drawn to different scales on purpose. The funnel (top) is true-to-scale: it shows how little rain survives to become dependable water. The Sankey (below) then zooms into that thin slice of usable water and splits it by user. Spine = year-2000 national accounts, the one dataset where every number ties out in a single document¹.

Precipitationgross rainfall over the country (MAP × land area)

~549 000 Mm³

Evapotranspirationlost to the atmosphere — soil, plants, open water

~500 000 ≈90%⁹

Mean annual runoffreaches rivers & recharges aquifers

49 040 ≈9%¹

Reliable yieldwhat dams, transfers & groundwater deliver dependably (2000)

13 227 ≈2.4%²

The headline: only ~1 drop in 11 of rainfall becomes river flow, and barely a quarter of that is dependably usable. South Africa already uses almost all of its reliable yield — which is why losses and efficiency matter more than hoping for more rain.

↓ Zoom: the 13 227 Mm³ of usable water, split by user (2000 sector accounts, Mm³/yr)¹

Agriculture drinks ~62% of usable water, so a small efficiency gain there frees more water than the entire urban+rural domestic supply combined. Surface water supplies ~9 500 (incl. ~3 000 moved by inter-basin transfers); groundwater ~2 000².

2 · How much each person uses at home

Domestic use, litres per person per day. South Africa is a high household user despite being water-scarce — and its figure includes distribution losses, so the true gap to thrifty peers is even wider than it looks. Bars are shaded by measurement basis; compare like with like.

South Africa (supply basis, incl. ~41% losses) metered household end-use target / estimate

Two honesty flags on this chart. (1) SA's 237 L is derived from total municipal supply, which carries ~41% non-revenue water⁴ — so it's closer to "water put into the system per person" than metered home use; the UK/Germany/NL/Israel bars are stricter metered figures. (2) The "world average 173 L" is used by DWS but its own footnote admits it "does not appear in any report… based on expert opinion"⁴ — treat it as an estimate, not a measured statistic. SA's own 2025 target is 175 L⁴.

3 · Inside the home — where the litres go

Split the household supply by fixture and one fact jumps out: the water that leaves a home dirty enough to matter (toilet/black water) is far smaller than the water that leaves it only lightly used (grey water — bath, shower, basin, laundry). That asymmetry is the entire case for greywater reuse and dual plumbing.

Typical middle-income SA household — % of total demand (REUM suburban model)¹¹

The reuse arithmetic. Recoverable grey water is >50% of indoor use¹⁴; toilet flushing is only ~28% of indoor use¹¹. Grey supply therefore exceeds toilet demand — a home's toilets can be flushed entirely on its own recycled greywater, with surplus left for the garden. In a middle-income area, 83% of potable water in becomes sewage out¹⁴ — most of it needlessly drinking-quality.

Indoor breakdown — international benchmark (USA, REU2016)¹²

Regroup those bars — the toilet is the biggest single fixture only because greywater is spread across shower, tap, washer & bath. Combined, greywater is ~2.5× the toilet:

Toilets are ~a quarter of indoor use almost everywhere: USA 24%¹², UK 22%¹³, SA ~28%¹¹ — a big single slice, but grey water (everything else clean) outweighs it ~2–2.5×, which is exactly why grey can flush every toilet.

Litres per use — and the efficient alternative¹¹

Use	Conventional	Efficient
Toilet flush	14.3 L	6 / 3 L dual-flush
Shower	~60 L	~30 L low-flow head
Bath	~80 L	— (shower instead)
Washing machine	~114 L/cycle	~60 L modern

Stack the retrofits and demand falls ~39%. REUM models dual-flush (−20% indoor), low-flow showerheads (−34% hot water) and xeriscaping (−39–50% of stand demand) combining to −38.7% total demand and −47% wastewater¹¹.

Income changes everything. The suburban profile above is ~43% outdoor (garden + pool). In a low-income/township household, outdoor use is <5% — demand is almost entirely indoor, dominated by toilet, washing and basin¹¹. So the "average" household is a fiction: greywater/garden savings target affluent suburbs; reliable supply is the issue for the poor.

4 · The access gap — infrastructure isn't the same as water

The defining feature of South African water is not a lack of pipes — it's pipes that run dry. The gap between having infrastructure and getting a reliable service is the real crisis, and it has widened since 1996³.

Households with water infrastructure89%

89%

Households with a reliable supply64%

64%

▲ 25-point gap — only 10.3 million of 16 million households can count on water³

Households with improved sanitation83%

83%

3.2 million households still without — incl. 2.1% bucket toilets, 1.6% none⁶

Urban vs rural — the safe-water ladder (JMP 2022)⁸

Basic water · urban

98%

Basic water · rural

75%

Safely-managed · urban

77%

Safely-managed · rural

52%

"Safely-managed" = on premises, available when needed, free of contamination. Half of rural South Africa is below that bar. 5.1% of rural people still drink straight from rivers, streams or pools⁸.

48%

of households hit water interruptions in the past year⁷

~627 000

households still fetch drinking water from rivers, wells & springs⁶

14.1 m

people without safe sanitation (2017 baseline)³

39%

of wastewater treatment works in a critical state¹⁰

Where the gap bites hardest — households with piped water (GHS 2023)⁶

Limpopo64.2%

Western Cape99.3%

Eastern Cape67.2%

Gauteng97.6%

Mpumalanga (sanitation 67%)low

National average~89%

Limpopo & Eastern Cape piped-water access has fallen over two decades — infrastructure ageing faster than it's replaced.

5 · Biggest levers — saving water & widening access

Ordered by the volume they can recover. The two largest sit upstream of the tap: water saved as a leak or a thirsty weed never has to be pumped, treated or paid for.

Fix the leaks (non-revenue water)

~2.1 bn m³/yr at stake

47.4% of municipal supply is non-revenue; real physical losses alone are 1.41 bn m³/yr⁵. Halving losses recovers more water than any new dam — and it's the cheapest source per m³. Losses rose from 37% (2014) to 47% (2023).

Curb domestic over-use

237 → 175 L/person/day

Hitting the national target would cut household demand by ~26%⁴ — pressure management, metering and tariffs on the 27% municipal slice, the fastest-growing use.

Irrigation efficiency

15–25% on 7 920 Mm³

Flood → drip/precision + scheduling on the single largest user (~62%)¹. A modest gain here dwarfs total domestic demand.

Clear thirsty alien plants

catchment yield recovered

Wattle, pine & eucalyptus in catchments intercept runoff before it reaches rivers. Clearing (Working for Water model) restores streamflow and creates jobs.

Reuse & recycle effluent

a drought-proof supply

Direct/indirect potable reuse (the Windhoek model) turns a discharge into rainfall-independent supply — but first the 39% of failing treatment works¹⁰ must be fixed.

Rainwater harvesting & groundwater

access at the edge

Groundwater potential (~7 500 Mm³) is largely untapped vs. ~2 000 used². Rooftop tanks + managed aquifers reach the rural households the grid can't — the fastest route to widening access, not just supply.

6 · The whole system on one page — where the water really goes

Before any talk of investment, sit with the picture itself. One diagram, rain to final fate, read left-to-right. What it shows is uncomfortable and clarifying: the overwhelming majority of our rain never reaches us; of the thin slice we do capture, agriculture takes nearly two-thirds at first use; and of the water piped to towns, a large share is lost or leaves as lightly-used greywater before it ever does a second job. Only when you see where the water actually goes does the question of what to do about it answer itself — which is why the investment reading (below the diagram) comes second, not first.

First, to scale — how little rain becomes usable water¹

Evapotranspiration — ~90% of rainfall (500 000 Mm³)

runoff

The blue sliver on the right (2.4%) is the 13 227 Mm³ of usable water — the whole Sankey below is just that sliver, magnified.

Availability Use Loss (recoverable) Reuse opportunity Quality & health (return flows)

First, the reflection — where the water really goes. Nature keeps ~90% of the rain (evaporation) before we touch it. Of the ~13 200 Mm³ we can dependably use, agriculture consumes ~62% at first use and never returns it. Only ~27% reaches towns as municipal supply — and of that, a large fraction is lost to non-revenue water⁵ or leaves the home as barely-used greywater¹⁴. So the system isn't short of water so much as it uses each drop once, carelessly, and lets the rest evaporate or leak. The scarcity is as much about how we hold water as how much falls.

…and only then, the investment reading it forces. Two right-hand endpoints stand out because the diagram puts them there, not because we went looking: ~1 400 Mm³/yr of non-revenue water⁵ and ~660 Mm³/yr of recyclable greywater¹⁴ — both recoverable without a new drop of rain. Mapped to the WRC's four RDI themes: Availability (groundwater 7 500 potential vs 2 000 used), Use (irrigation precision; 237→175 L), Loss (halve NRW — metering, pressure, leak AI), Quality & health (fix the 39% failing works, unlock reuse).

How to read the volumes: the yield→sector spine is measured 2000 national accounts¹. The municipal→loss/delivered and delivered→end-use splits are proportional overlays from cited ratio studies (NRW 41–47%⁵; suburban end-use¹¹) applied to that spine — illustrative of structure and relative size, not same-year measured sub-volumes. "Municipal delivered" includes commercial users, shown at household proportions.

7 · The chain — rate-limiting steps, and why saved water is not yet delivered water

Sections 1–6 are a volume account: where the water goes. This section adds the second axis: whether each stage of the delivery chain can pass the flow at all. Water reaches a tap through a series of processes — capture, conveyance, treatment, reticulation — and, as in any series process, the throughput of the whole chain is set by its rate-limiting step, not by the total feedstock. A saving becomes delivered water only if it enters the chain downstream of the binding constraint, or if that constraint itself is relieved. Two problem types follow — source-loss problems (volumes) and rate-limitation problems (capacities) — a latitude and a longitude. Each needs its own data, and its own mitigations.

Loss axis — volume that disappears at this stage (the Sankey dimension) Rate axis — the throughput ceiling at this stage (the capacity dimension)

The rule of the chain: delivered water ≈ the minimum, over all stages, of (flow in × (1 − loss)) capped by stage capacity. This is why the country can be simultaneously water-scarce and dams-full-taps-dry: Gauteng in 2023 held a near-full Vaal system while Johannesburg's reservoirs ran empty (a pumping and reticulation ceiling); Giyani and Hammanskraal are the same signature. The 89%-infrastructure-vs-64%-reliability gap³ is the national fingerprint of a system whose binding constraint sits mid-chain — at works, pumps and pipes — and a mid-chain constraint is untouched by upstream supply gains.

Feedstock recoveries · upstream

must pass every ceiling

Alien clearing, evaporation suppression, catchment restoration add raw water at the top of the chain. Each recovered m³ must still be captured, conveyed, treated and reticulated before it is access — four gates, each with a ceiling. Real value where raw supply is the binding constraint; stranded where it is not.

Debottlenecking · mid-chain

adds no water — raises the ceiling

Works refurbishment, pump O&M, conveyance upgrades act on the rate axis. Two abstract moves, as in any flow system: reduce friction — recover the capacity already designed and paid for (maintenance, pressure control, refurbishment) — or increase cross-section — new treatment modules, parallel mains, added storage. Friction first: it is the cheaper of the two and most of the lost ceiling is friction.

Post-treatment recovery · downstream

finished inventory

A m³ recovered from reticulation leakage or a running cistern has already paid every toll — captured, conveyed, treated, pumped. That is the deep reason non-revenue water is the cheapest m³ in the country. Demand substitution (greywater, xeriscaping) acts at the same point: it frees finished water and creates headroom at every stage above it at once.

Reading the mitigation ledger (§8) against the chain

The data corollary: the loss axis is served by flow accounts — and §9–10 show even those have gone dark. The rate axis needs a capacity register: design versus actual throughput per works, pump station, main and reservoir zone. No such national register is published. Locating the binding constraint per supply system is therefore impossible from the public record — which makes the capacity register the rate-axis twin of the sector account, and both belong on the §10 instrument panel.

8 · The mitigation ledger — every loss, and every counter to it

Each loss endpoint in the Sankey above, matched to the full set of known counters. Rows run in the order the water flows — catchment to tap. The green tag in each row is the first move: the intervention with the best evidence of recovering the most water per rand. Click any tag for the full definition, the South African evidence, and its sources. Volumes are the Sankey's; two rows (conveyance, irrigation) are losses the country does not currently measure.

The cross-cutting row that makes every other row possible: metering and validated water accounts at every tier — household, farm gate, municipality, national — plus scarcity-reflecting tariffs and the institutional capability to act on the numbers. Half the tags above are unmanageable while two-thirds of national use goes unmeasured (section 9). Measurement is not overhead on the mitigation programme; it is the mitigation programme's first line item.

9 · The trend — the system is moving the wrong way

First the sector ledger across two decades, then three time series — one message: the pressure is rising and our grip is loosening, and none of it is about less rain. It's more people, more loss, and a demand we haven't bent. The encouraging part is in the middle panel below: the 2030 shortfall is a choice, not a forecast.

Sector demand, three snapshots — measured 2000¹, derived 2015³, and the data gap at 2023⁵

The account went dark. The last full national picture of who uses the water is the Master Plan's 2015 estimate — and even that is a percentage split, not measured sector volumes. Since then only the municipal slice is nationally measured (No Drop 2023: 4 400 Mm³ system input — up ~27% on 2000's urban+rural supply, while reliability fell). For irrigation — ~62% of all use — there is no published post-2015 national measurement at all. For a water-scarce country, not knowing where two-thirds of its water goes is itself a finding — and an RDI agenda item (metering, accounts, monitoring) as concrete as any dam.

Reading the numbers: 2000 = NWRS-1 measured sector accounts¹ (urban + rural combined here as "municipal" for comparability). 2015 = the Master Plan's published shares (61% agriculture · 27% municipal · 2.5% mining + 3.5% industry · 2% power · 3% forestry) × its 15 294 Mm³ total³ — derived, rounded. 2023 municipal = No Drop system input⁵ — a supply-side basis that includes losses. Totals: 12 871 → 15 294 → not published.

Non-revenue water — climbing

Losses rising ~1 point a year: 36.8% (2012)¹⁵ → 41% (2019)³ → 47.4% (2023)⁵. Good practice is ~15% (dashed). Nearly half the water put into municipal systems is now never sold.

2030 — the deficit is a choice

Demand rose 12 871 → 15 294 Mm³ (2000→2015)¹ with population. By 2030, business-as-usual opens a ~17% deficit (2.7–3.8 bn m³); but cutting losses to 15% and use to 175 L flips it to a small surplus³. Same rain, opposite outcomes.

More people, less reliable

+38% more people since 2001 (44.8 → 62.0 m)⁷, yet only 64% have a reliable supply — below the 1996 level³. Infrastructure grew; reliability didn't. National access edged up (84→87%), but it fell in 5 provinces⁶.

The thread through all three: South Africa's deepening water stress is manufactured downstream — by loss, by single-use, by demand outrunning maintenance — far more than it is dictated by the sky. That is the hopeful reading, because downstream is where we have agency.

10 · Plan–Do–Study–Act — the instrument panel for running this plan

A ledger of mitigations (§8) is not yet a plan, and a trend chart (§9) is not yet management. What closes the gap is the improvement loop: Plan a numbered target on one lever, Do it at a scale one team owns, Study the instruments against the baseline, Act by standardising what worked — and turn the wheel again. Everything the loop needs is already on this page: the levers, the baselines, and the national instruments that read them.

Plan — pick the lever and the number

Each ledger row has a starred first move; each metric below has a baseline and a target. A plan names the zone, the owner, the number and the date — "NRW in this DMA from 47% to 40% by June", not "reduce losses".

Do — implement at owned scale

Pilot where one team owns the result: a district metered area, an irrigation scheme, a treatment works. The ledger tags are the work orders; the starred tag goes first.

Study — read the instruments

Quarterly readout against baseline. Nationally the gauges are No Drop, Green Drop, the GHS and JMP; locally, minimum night flow, farm-gate meters and works compliance. Section 9 is what the national instruments say today.

Act — standardise or steer

What worked becomes the standard — a bylaw, a budget line, an O&M schedule, the next DMA. What didn't gets a revised plan, not a quiet burial. Then the wheel turns again. The evidence that feeds this quadrant comes from the two research fans in §11.

The Study panel — nine metrics that tell us where we are (baselines as cited elsewhere on this page)

Metric	Baseline	Target	Instrument · cadence

The ten-year Study result is already scheduled: the 2030 fork in section 9 — a −17% deficit on business-as-usual, or a small surplus if losses fall to 15% and use to 175 L³. This panel is the quarterly version of that annual reckoning: it exists so the fork is steered, not discovered.

The Do column — where implementation of the ledger stands today (2026)

Live — operating now Partial — proven, not scaled Stalled — started, then stopped Not started

Read the board honestly and the pattern repeats the page's thesis: the measurement instruments (No Drop, Green Drop) have been rebuilt; the site-scale proofs (Sebokeng, eMalahleni, Durban recycling, FruitLook) all work; what is missing is the middle — scaling proven pilots into standard practice. That middle is exactly what a running PDSA cycle is for.

11 · Adaptive management — the two research fans that keep the wheel honest

The formal name for §10 run properly is adaptive management: treat every mitigation as a hypothesis, monitor what is implemented, evaluate what is competing, and let the evidence — not habit, not politics — set next year's slate. Two research fans, different in kind, converge once a year on a portfolio decision. Fanning out is the research programme; fanning back in is the deliverable.

Ray labels are worked examples drawn from the §8 ledger, not an exhaustive slate — each ledger row seeds questions for both fans. The two fans are deliberately different in kind: the evaluation fan asks which mitigation earns its place (comparative, counterfactual); the monitoring fan asks is the implemented one still performing (longitudinal, instrumented). Confusing them produces research that never decides and dashboards that never learn.

The evaluation fan · feeds Act

comparative · episodic

Asks counterfactual questions across the whole ledger: what did each mitigation recover, per rand, against what would have happened anyway? Method: paired trials and before–after–control designs, with rainfall-corrected baselines so a wet year can't flatter an intervention. Output: effectiveness rankings and implementation playbooks — the artefact that turns one municipality's proof into the next one's default.

The monitoring fan · feeds Study

instrumented · continuous

Watches every implemented mitigation against its expected trajectory. Method: instruments, not studies — minimum night flows, farm-gate telemetry, audit scores, ladder surveys — reviewed quarterly on the §10 panel. Output: early warning of decay (a pressure-managed zone drifting back, a cleared catchment re-invading) and confirmation that scaled measures still perform.

The fan-in · the deliverable

scale · hold · kill

One annual portfolio decision, against declared criteria: R/m³ actually recovered as the common currency, equity of access and implementability alongside. Every mitigation leaves the room with a verdict, and the verdicts become next year's implementation slate. The licence to kill is non-negotiable — a fan that can only recommend "continue and expand" is decoration. An honest one would have ended War on Leaks years earlier and redirected the money.

Why the name matters. Adaptive management is an established, citable method from natural-resource governance — monitoring and evaluation feeding a periodic portfolio decision under uncertainty — so this lands as method applied to the national water ledger, not invention. It is also fundable as exactly that: this page already argues that not knowing where two-thirds of the water goes is an RDI agenda item (§9), and a two-fan research programme wrapped around the mitigation ledger is precisely the shape of programme water-sector RDI funders exist to commission — the measurement instruments (No Drop, Green Drop, GHS, JMP) become its data spine, and the §10 status board becomes its first-year work plan.

12 · What the evidence already shows — early findings

The two fans (§11) are not only a proposal; they have already turned. Three results are in hand, each drawn from the public record, and together they answer the question the whole page circles: why the losses persist, and where the fix actually sits. Each is written up as a full paper; the panels below are the finding in one view.

A · Governance predicts water-loss performance — and it survives the obvious challenge

clean / good auditqualifiedadverse / disclaimer

The finding. Across 100 water services authorities, a municipality's audit outcome — a governance measure the Auditor-General publishes entirely outside the water sector — predicts its No Drop water-loss score strongly (Spearman ρ = 0.56). The association survives control for municipal size and poverty: each step up the audit ladder is worth ~11 score points independent of how large or poor the municipality is (β = 10.7, p < 10⁻⁵), and the model separates below-adequacy WSAs at AUC 0.85. A municipality with a disclaimer audit and high water losses does not first need a leak-detection programme; it needs the administrative function that would sustain one.

B · Three long-record mitigations, judged by the scale / hold / kill rule — on the evidence available at the time

The recurring failure was the decision, not the intervention. In every case the evidence sufficient to act was public years before the budget responded: a proven no-capital model left unreplicated, a review-recommended restructuring not adopted, a departmental admission of failure not acted upon. War on Leaks alone spent about R2.5 bn after its own department admitted, in 2018, that it had reduced no losses.

C · The coverage matrix — what the country still measures, and what has gone dark

Read it as the programme's data spine. Measurement persists at the hydrological edge (rain, streamflow, dams) and the household edge (annual surveys), and the middle audits were revived (No Drop 2023, Green Drop 2022). But the national sector account has had no measured point since 2000, and irrigation — about 62% of use — is the darkest large flow. The rate axis (the capacity register, §7) is darker still: never assembled. Both axes must be lit before the fans run on more than the public scraps.

13 · The scoreboard — Blue, Green and No Drop

The Department of Water and Sanitation runs three incentive-based audits — the country's official scoreboard for exactly the losses this page maps. Each gives every municipal system a weighted score out of 100; certification sits at the top of the scale, and a cumulative risk rating flags the bottom. The audits lapsed after 2014 and were revived in 2021–2023, which is why the trend lines in §9 have a decade-wide gap in their middle. They are the Study instruments of the §10 wheel: independently audited, nationally comparable, on a published cycle.

🔵 Blue Drop — drinking water

26 of 958 systems certified (2023)

Audits every water supply system (958 systems, 144 authorities in 2023) on weighted criteria: water-safety planning, treatment-process management, drinking-water-quality compliance, management commitment. Certification needs ≥95%; a score below 31% marks a system in a critical state. 2023: 26 systems certified, down from 44 in 2014; 277 systems critical, up from 174²⁴. The 2025 progress assessment shows the risk edge improving slightly: 61.9% of systems low-risk, 7.9% critical-risk²⁵.

🟢 Green Drop — wastewater

14 of 848 systems certified (2025)

Audits every wastewater system. Effluent quality itself carries only 30% of the score; the other 70% is capacity, environmental, financial and technical management — the audit is of the institution, not just the water¹⁰. Certification ≥90%; below 31% = dysfunctional. The 2025 report (audit year 2023/24): 396 of 848 systems (47%) critical, up from 39%; 66 systems (8%) excellent or good, down from 14%; 14 certified — five in the Western Cape, one in Mpumalanga, eight in Gauteng, and six provinces with none²⁵.

⚪ No Drop — water losses

4 authorities ≥90% (2023)

Audits water-use efficiency: the water balance itself (IWA format), metering coverage, non-revenue water, and the conservation/demand-management programme. Certification ≥90% on the same five-band scale. 2023: only four of 144 water services authorities achieved it; national non-revenue water 47.4% (2.1 bn m³/yr)⁵, barely moved at 47.3% in the 2025 progress assessment²⁵. The No Drop score is the audit mirror of this page's §8 ledger — and it is the score the §12A governance finding predicts.

The five performance bands — one scale across all three programmes

<31 critical

31–50 poor

50–80 average

80–90 good

≥90 excellent

Certification sits in the excellent band — ≥95% for Blue Drop, ≥90% for Green and No Drop. The width of each segment above is its share of the 0–100 scale; the national picture is that nearly half of wastewater systems sit in the leftmost band while the certified sliver holds fourteen. A municipality can read its own three scores as a triage order: Blue Drop protects people this month, Green Drop protects the rivers this year, No Drop funds both by stopping the 47% that vanishes unbilled.

14 · Water credits — paying for verified saved water

Every counter in the §8 ledger saves cubic metres, and §7 showed why saved water is not yet delivered water: someone must finance the intervention long before the tariff income moves. A water credit is the instrument built for that gap — one independently verified cubic metre of water saved, reused, treated or replenished in a named catchment in a named year, sold to a buyer who wants the benefit on their books. It is the water twin of the carbon credit, standardised under the World Resources Institute's Volumetric Water Benefit Accounting (VWBA)²⁶.

How a credit is minted

Five steps, none optional: baseline (the metered water balance before the intervention — the same IWA balance No Drop audits); intervention (a ledger item: pressure management, leak repair, reuse, alien clearing, harvesting); measured delta (metered, not modelled, wherever possible); third-party verification of the volume; registry issue carrying catchment, vintage and method, retired on sale so each cubic metre is claimed once.

Who buys, who sells

Buyers are water-stressed corporates with public replenishment targets — breweries, soft-drink bottlers, data centres, mines — who must show volumetric benefit in the catchments they draw on. Sellers are whoever executes the intervention: a municipality cutting non-revenue water, a water fund clearing thirsty alien plants, a provider whose reuse unit delivers audited recycled water. South African precedent exists on both sides: catchment water funds already finance alien clearing on this logic, and credit platforms are issuing water credits for projects in South Africa today²⁷.

Why the audits make the market

A credit is only as good as its verification, and the verification rail already exists on this page: the No Drop water balance is the baseline document, the §10 instrument panel is the metering, and the §12A governance finding says which sellers a buyer can trust. The largest creditable pool in the country is the non-revenue water itself — 2.1 bn m³/yr⁵ — which no municipality can currently afford to chase at the speed the deficit demands. Credits let the beneficiary of the saved water pay for the chase.

The honest caveat. South Africa has no statutory water-credit market; today's credits are voluntary corporate-stewardship instruments, and a municipal seller would need the metering and audit trail this page keeps insisting on. That is the point: the same instrument panel that runs the §10 wheel is the minting machinery for credits — build it once, and it both steers the plan and pays for it.

Sources & data notes

Every headline figure traces to a primary document below. Two figures are flagged as derived / estimate and should be cited with their caveat.

DWS, National Water Resource Strategy, 1st ed. (NWRS-1), 2004 — Ch.2 Table 2.3: 2000 sector volumes (Irrigation 7 920, Urban 2 897, Rural 574, Mining & bulk industry 755, Power 297, Afforestation 428; total 12 871) and MAR 49 040 Mm³. dwaf.gov.za NWRS Ch.2 (Wayback)
DWS, National Water Resource Strategy, 2nd ed. (NWRS-2), 2013 — reliable yield 13 227 Mm³ (surface ~9 500 incl. ~3 000 transfers; groundwater ~2 000, potential ~7 500); MAR 49 000. dws.gov.za NWRS-2
DWS, National Water & Sanitation Master Plan, Vol.1 "Call to Action", 2019 — MAP 465 mm; 89% infrastructure vs 64% reliable supply; 10.3 m reliable-supply households; >3 m without basic water; 14.1 m without safe sanitation. gov.za NWSMP
DWS, National Water & Sanitation Master Plan, Vol.2 "Plan to Action", 2018/19 — domestic use 237 L/c/day, target 175 L/c/day by 2025; "world average 173 L/c/day" (the document's own footnote: "does not appear in any report… based on expert opinion"). gov.za NWSMP Vol.2 (PDF)
DWS, No Drop Report 2023 (Nov 2023) — system input 4.4 bn m³/yr; non-revenue water 47.4% (2.1 bn m³); total losses 40.8% (1.8 bn m³); real/physical losses 1.41 bn m³. ws.dws.gov.za No Drop 2023
StatsSA, General Household Survey 2023 (P0318) — piped-water & sanitation access, ~627 000 hh on rivers/wells/springs, provincial breakdown, 35.8% dysfunctional-service. statssa.gov.za GHS 2023
StatsSA, Census 2022 (P0301.4) — 8.7% of hh with no piped water at all; 48.4% experienced interruptions; provincial no-access map. census.statssa.gov.za
WHO/UNICEF Joint Monitoring Programme (JMP), South Africa file, 2022 — urban/rural basic & safely-managed water & sanitation ladder; 5.1% rural surface-water reliance. washdata.org/zaf
StatsSA, Natural Resource Accounts: Water Accounts, Report 04-05-01, 2004 (citing DWAF) — "90% of precipitation is used in evapotranspiration, the remaining 10% is available as run-off." statssa.gov.za Water Accounts
DWS, Green Drop Report 2022 — 334 of 850 (39%) wastewater treatment works in a critical state. DWS presentation to Parliament, 7 Jun 2022 (PMG)
Jacobs HE & Haarhoff J, Residential End-Use Model (REUM), Water SA 30(3), 2004 — SA suburban end-use split (toilet 16% total / 28% indoor, garden ~43%), per-use litres (flush 14.3 L; dual-flush 6/3 L; bath 80 L; washer 114 L), demand-management savings (combined −38.7% demand, −47% wastewater), low-income outdoor <5%. DOI 10.4314/wsa.v30i3.5078. doi.org/10.4314/wsa.v30i3.5078
Water Research Foundation, Residential End Uses of Water v2 (REU2016), Project 4309 (DeOreo, Mayer et al.) — USA indoor breakdown: toilet 24%, faucet 20%, shower 20%, clothes washer 16%, leaks 13%, bath 3%, dishwasher 2%. waterrf.org 4309
Energy Saving Trust / Water UK, At Home with Water, 2013 (100 000+ households) — UK: showers 25%, toilets 22%; 142 L/person/day. energysavingtrust.org.uk At Home with Water
SA greywater fraction — Rodda et al., Water SA 37(5), 2011; Ilemobade et al., Water SA 39(3), 2013 — greywater >50% of indoor use; 83% of potable becomes sewage in middle-income areas; kitchen = "dark grey" (Class III), often excluded from reuse. Rodda 2011 (SciELO) · Ilemobade 2013 (SciELO)
WRC, The state of non-revenue water in South Africa, Report TT 522/12, 2012 (McKenzie, Siqalaba & Wegelin) — national NRW 36.8% (~1 580 Mm³/yr), the 2012 baseline in the trend series. wrc.org.za TT 522/12. The 2030 "17% deficit / 2.7–3.8 bn m³" projection ³ originates in McKinsey, Confronting South Africa's Water Challenge, 2010, carried into the Master Plan.
Le Maitre DC, Versfeld DB & Chapman RA, "The impact of invading alien plants on surface water resources in South Africa: a preliminary assessment", Water SA 26(3), 2000 — invasive-alien-plant incremental water use ≈3 300 Mm³/yr (~7% of MAR). wrc.org.za Water SA 26(3) PDF · programme: DFFE Working for Water (est. 1995; >1 m ha cleared; ~20 000 jobs/yr).
Wegelin WA & McKenzie RS, Leakage reduction through pressure management: concepts and case studies, WRC Report TT 186/02, 2002 — pressure-leakage theory and SA case studies. wrc.org.za TT 186/02 · Sebokeng/Evaton advanced pressure-management PPP (2005; projected ~R20 m/yr savings, ~R30 m achieved in first full year): WRP case article.
du Pisani PL, "Direct reclamation of potable water at Windhoek's Goreangab reclamation plant", Desalination 188:79–88, 2006 — direct potable reuse since 1968, world's first; multi-barrier train; up to ~25% of city supply. doi.org/10.1016/j.desal.2005.04.104
Veolia / eThekwini Water Services, Durban Water Recycling — 47.5 ML/day tertiary plant (commissioned 2001) selling near-potable recycled effluent to Mondi and Sapref; frees potable water for ~300 000 people. veolia.co.za Durban Water Recycling
eMalahleni Water Reclamation Plant (Anglo American & BHP Billiton JV, commissioned 2007) — treats up to 50 ML/day of acid mine drainage to potable standard; ~12% of the local municipal drinking supply. UNFCCC case summary (PDF)
DWAF, Artificial Recharge Strategy for South Africa, June 2007 — national managed-aquifer-recharge strategy; Atlantis (Western Cape) recharging since the 1970s. dws.gov.za AR Strategy (PDF) · Four decades of water recycling in Atlantis, Water SA 42(4), 2016
Eskom, Cooling techniques at Eskom power stations, fact sheet CO-0005 — direct dry cooling at Matimba, Majuba, Medupi & Kusile: ~2 L/kWh → ~0.14 L/kWh (≈1.75% thermal-efficiency penalty); Medupi = world's largest dry-cooled station. eskom.co.za CO-0005 (PDF)
FruitLook (Western Cape Department of Agriculture & eLEAF) — free satellite irrigation-scheduling service since 2010/11; weekly ET, biomass & nitrogen at 20 m resolution; ~35 000 ha actively monitored. fruitlook.co.za · Elsenburg evaluation study (PDF)
DWS, Blue Drop Report 2023 (full audit, released 5 Dec 2023) — 958 water supply systems across 144 water services authorities; 26 systems ≥95% (Blue Drop Certification), down from 44 in 2014; 277 systems in a critical state, up from 174. ws.dws.gov.za BDN 2023 (PDF) · gov.za statement
DWS, Green Drop Report 2025 + Blue Drop & No Drop progress assessments (released 1 Apr 2026; audit year Jul 2023–Jun 2024) — 848 wastewater systems: 396 (47%) critical, 66 (8%) excellent/good, 14 Green Drop certifications (WC 5, MP 1, GP 8; six provinces none); Blue Drop risk profile 61.9% low-risk / 7.9% critical-risk; non-revenue water 47.3%. Engineering News, 1 Apr 2026
World Resources Institute / LimnoTech, Volumetric Water Benefit Accounting (VWBA) — the standard method for quantifying water-stewardship benefits in m³/yr; the accounting basis of voluntary water credits. wri.org VWBA
Green Finance Institute, Act4Water Water+ Credits (case study) — voluntary water-credit initiatives generating certified water credits ("CAPs") for projects including South Africa. greenfinanceinstitute.com

Comparison-country domestic use (chart 2), mixed metered-household sources: USA ~300–378 L (USGS Water Science School; US EPA WaterSense); Australia 191 L (2007); UK 137 L (Discover Water / Water UK); Germany 121 L (BDEW); Netherlands 129 L (Vewin); India ~126 L urban; Israel 137 L domestic; Namibia/Windhoek 163 L domestic. Bases differ (household-metered vs total-municipal); Spain's often-quoted 280 L was dropped as basis-ambiguous.

Derived node: gross precipitation volume (~549 000 Mm³) is MAP 450 mm × land area 1 221 037 km² — no DWS document publishes a single gross-rainfall figure. At 465 mm it is ~568 000 Mm³. The 2000 NWRS-1 spine (MAR 49 040 → yield 13 227 → use 12 871) is used because all three tie out in one primary document; 2015 figures are higher (use ~15 294, total available ~17 559) but group sectors differently. 2015 sector volumes (section 9 slopegraph) are derived: the Master Plan's published shares × 15 294 (agriculture 61% → 9 329; municipal 27% → 4 129; mining 2.5% + industry 3.5% → 917; power 2% → 306; afforestation 3% → 459); 2023 municipal (4 400) is the No Drop system-input volume — a supply-side basis including losses, not metered end-use.