Water accounting answers one question with real precision: who used how much, when, and where the rest of it went. Nearly every allocation decision in the American West rests on that answer, and getting it right is harder than people outside the work imagine.
It is also not the question a drying landscape is asking.
A ledger reconciles claims against a supply. It does not change the supply. That is not a complaint about accountants — it is what an account is. Improve the accounting and you get a truer number. You do not get a wetter watershed.
what water accounting actually is
Water accounting is the practice of measuring, attributing, and reconciling water over a defined area and period — how much was diverted, how much was actually consumed, how much returned to the channel, and whose use it was under which right.
Three jobs, in order:
- Measurement — diversion at the headgate, consumptive use in the field, return flow to the stream, losses in transit. A few of those are metered. Most are estimated.
- Attribution — whose use, under what priority date, against which allocation, contract, or storage account.
- Reconciliation — does the sum of uses match the supply the record says was available?
The whole professional difficulty sits in one distinction: diversion is not consumption. A ditch can divert far more than a crop consumes, and the remainder returns to the river for the next user downstream. Consumptive use — the share that leaves as evapotranspiration and never comes back — is the number that governs everything and the number no one can meter at every field — measured at a handful of flux towers and estimated everywhere else from satellite evapotranspiration. Most of what looks sophisticated in modern water accounting is an attempt to estimate that one quantity honestly.
The vocabulary changes by chair. Run a utility and the same discipline appears as source metering, treatment and distribution loss, non-revenue water, and the return-flow credits written into your discharge permit. Sit on a commission and it appears as consumptive-use estimates, transit loss, and reservoir storage accounts. Underwrite property and it appears as whether a named right will actually yield in a dry year. Same ledger, three dialects.
the ledger question and the factory question
Both questions are legitimate. One of them has a standing institution behind it. The other is funded where a partnership exists.
| the ledger question | the factory question | |
|---|---|---|
| what it asks | who used how much, when, and under what right | how much water does this landscape produce, and is that number rising or falling |
| unit | acre-feet, cubic metres, an account balance | condition — infiltration rate, soil carbon, canopy, floodplain connection, upland storage |
| time frame | a water year, a compact period, a rate case | decades |
| what a good answer buys | defensible allocation, enforcement, trust between parties who do not trust each other | more water entering the system in the first place |
| how it fails | estimates drift, return flows get double-counted, a conserved volume is recaptured by the next diverter | the ledger balances year after year on a shrinking total, and improving the total is nobody's job |
| who owns it | the state engineer, the commission, the utility's finance office | usually a voluntary partnership, if anyone |
That last row is the argument. Every acre-foot in a well-run ledger has a name attached to it. The production function — the land that turns snow and rain into a yield somebody can divert — has no account holder with a standing obligation — where it is funded at all, it is a utility partnership or a water fund, voluntary and patchy, which is exactly the pattern worth extending. Denver Water's Forests to Faucets is the familiar western example.
how they actually count
The clearest live example of water accounting done seriously is in the Upper Colorado River Basin, and it is worth reading closely because the people building it are precise about what it is.
The Upper Colorado River Commission approved a memorandum of understanding with the Bureau of Reclamation in October 2024, signed that December, to run provisional accounting on conservation and demonstration projects in 2025 and 2026. The stated understanding is that the Upper Division States — Colorado, New Mexico, Utah, Wyoming — "will seek credit for water that, as a result of these Qualifying Activities, flows to and is stored in Upper Colorado River Basin reservoirs." That word — credit — is the product name water finance has landed on, and what a water credit actually is takes the instrument apart.
How they count, in one pass:
- Conserved consumptive use on irrigated ground is estimated primarily with eeMETRIC version 2.3, a satellite evapotranspiration model released by Reclamation. The Commission adopted eeMETRIC in 2022 as a uniform method for agricultural consumptive use, which is the unglamorous prerequisite for four states agreeing on any number at all.
- Transit loss between the project ("Point A") and the reservoir ("Point B") is estimated by the states, accounting for proximity and intervening points of diversion. Utah has paired stream gages with a RiverWare model to do it.
- Legal protection of the volume happens outside the model. Participants filed fixed-time change applications with Utah's Division of Water Rights for each project.
It works. Eight Demand Management Pilot Program projects have been approved in Utah across the two cycles, and the state's own read of the Water Year 2025 accounting is that it shows water can be leased under an approved conservation program and moved downstream to Lake Powell. That is real infrastructure, built by people who were criticized for not having it: the earlier System Conservation Pilot Program paid for reduced use without tracking whether the saved water reached Powell, which is exactly the hole this closes.
Now the honest part, which comes from the MOU itself rather than from us. Section 6 states that provisional accounting of water flowing to a reservoir "will not influence or otherwise affect the operations of that reservoir" unless it is later converted into credit under a separate agreement. UCRC executive director Chuck Cullom put it plainly when the framework was proposed: "It is not an operational guide for Reclamation; it is a means for folks to understand how much water would be available in that account upon the implementation of a formal agreement or credit program."
Nobody in that room is claiming the accounting makes water. The confusion happens two rooms over — in a board deck, a rate filing, or a press summary — where "we now account for conserved water" gets read as "we now have more water." Those are different sentences.
the international version of the same ledger
There is a standardized version of all this. SEEA-Water, the United Nations water accounting framework, runs three account types: physical flow accounts (abstraction, use within the economy, return flows), physical asset accounts (the stock of water and its depletion over the period), and economic accounts (costs, water products, financing). The physical supply-and-use table is double-entry in cubic metres — each flow recorded once by origin and once by destination — which is precisely why the accounting is trustworthy and precisely why it cannot manufacture volume. Note which of the three is closest to the factory question: the asset account, the one tracking stock and depletion over the period. It is also the one that requires estimating quantities nobody can meter completely — aquifer volume, soil moisture, snowpack — which makes it the hardest of the three to keep current. If you want that thread pulled properly, the accounting of natural capital has its own page.
three things a ledger structurally cannot do
You might be reading this as "measurement is a distraction." It is not the claim. Bad accounting is how paid conservation leaks, how compacts get argued in the dark, and how a utility discovers its non-revenue water only after the bond issue. Count what you must count. Just be clear on what counting is unable to reach.
1. It cannot create supply. Every improvement in a water ledger improves the description of a fixed quantity. Sharper estimates redraw the split; they do not enlarge the total.
2. It cannot see condition. An acre-foot from a wet meadow with a connected floodplain and an acre-foot flashing off an incised, compacted catchment enter the ledger as the same unit. They are not the same water. One arrives late, cool, and filtered; the other arrives all at once, warm, and full of sediment. The ledger has no column for that, and the difference is most of what determines next year's yield.
3. It cannot make anyone responsible for the total. This is the structural gap. Accounting assigns ownership of shares. Nothing in it assigns ownership of the source. So the source is improved where a partnership happens to exist, in the years a grant cycle allows — which is why the total keeps drifting down while every individual account stays in good order.
Water accounting is a description of allocation. Restoration is a change in production. The first can protect the second. It cannot substitute for it.
what an underwriter needs that the ledger does not hold
For insurers this distinction is not philosophical, it is a pricing problem. A water ledger tells you who holds a right and how it is administered. It does not tell you whether the water will show up, or how fast. Exposure lives in condition: a burned, hydrophobic watershed delivers a debris flow and a treatment shutdown rather than a runoff hydrograph; a drained meadow system converts a good snow year into a spring flood and an August shortage. Those are loss events sitting in the same catchment the ledger describes as balanced.
Which is the reason nature-based approaches keep drawing underwriter attention — and also the reason we are careful with the term, since a parametric nature-based policy is still insurance: it pays after. The condition of the catchment is upstream of both the ledger and the loss.
the second ledger nobody keeps
The objection writes itself: the fix for a ledger is not another ledger. Agreed — unless the second one books a different quantity and puts money behind someone obliged to move it.
There is a second set of books for the same watershed. It records stocks — the ecosystems present and their condition — and the flows those stocks produce, water abundance among them. That is the natural capital view, and it answers the factory question directly: what is here, what shape is it in, what is it yielding, and is that trending up or down.
Ensurance is our attempt to attach money to that second ledger. An agent is an account standing for a named place. A certificate — specific ensurance — is issued one-to-one against that agent and funds the place's present condition, with proceeds routed to the people doing the work. It is not a water credit and not a tradable acre-foot; it is not fungible with another watershed and carries no neutrality claim. Post three in this series exists because that collision needs to stay honest.
Two limits we will state ourselves. The valuation underneath — the condition and service-flow numbers that let a finance committee see a landscape at all — is a bridge for capital, never the worth of the place. And our own stage is early: the agents, coins, and certificates are live, the volumes are small. Anyone describing a nature finance product as finished is describing a roadmap.
count what you must, fund what makes the water
The two questions want different budgets, and they do not compete.
If you sit on a commission or a state agency: keep building the accounting. Then ask what line item in your budget is responsible for the basin's yield rather than its division — and where that money would go if it existed. Usually the answer is a specific set of headwaters, meadows, and floodplains that no account holder is obliged to maintain.
If you run a utility: you already price source protection against capital projects, and the ratio usually favors the watershed. Source protection is the cheapest water you will never build. Accounting tells you how much you are losing; condition tells you what you can still get back.
If you underwrite: put catchment condition in the exposure model, not just the water right. The ledger describes the claim; the land determines the event.
The engine itself is not this post's job. The water cycle is broken where you live covers how the land-side of the cycle actually works, and restore one thing, fix ten covers which repairs pay off across drought, fire, and species at once. Read either one and the ledger stops looking like the plan.
frequently asked questions
what is water accounting?
Water accounting is the measurement, attribution, and reconciliation of water over a defined area and period: how much was diverted, how much was consumed, how much returned to the stream, and whose use it was under which right. Its hardest quantity is consumptive use, which cannot be metered at scale and is usually estimated — in the Upper Colorado Basin, with satellite evapotranspiration models such as eeMETRIC.
does water accounting increase supply?
No. Water accounting describes how an existing supply is divided; it does not change how much water the landscape produces. A supply-and-use account balances by construction, so better accounting yields a truer split of the same total. Supply changes only when the land's ability to capture, store, and release water changes — which is restoration, not bookkeeping.
what is the difference between water accounting and watershed restoration?
Water accounting is a description of allocation, measured in acre-feet over a water year. Watershed restoration is a change in production, measured in condition over decades — infiltration, soil carbon, floodplain connection, upland storage. Accounting tells you who got the water; restoration determines how much there is to get. Programs that track conserved water, such as provisional accounting in the Upper Colorado Basin, protect a volume from being recaptured; they do not create one.
the series
Six posts on credit and ledger words for water — what each instrument actually is, and who funds the landscape that produces the volume being counted.
- what a water credit actually is — the definition and the allocation trap
- water accounting is not a restored cycle — the ledger question versus the factory question (this post)
- a water certificate is not a wet acre — a claim on a volume versus the place that makes it
- the conserved acre-foot that leaked — recapture, and why accounting exists
- who buys a water credit — the payors already covering shortage
- hold the factory, not the ledger — the alternative purchase, stated plainly
Adjacent: what a biodiversity credit actually is on the sibling unit product, and the colorado river runs out of rules in 2026 on the allocation politics this accounting sits inside.