Arizona · Three water clocks
The tap is on. The permit is not.
The kitchen sink is a terrible water gauge. It answers only whether this house has water today—not whether the next subdivision can prove a supply, or whether a rural well will reach the same aquifer tomorrow.
Arizona does not run out of water all at once. Models and laws decide which demand counts, in which basin, under which right, and in which year.
- Phoenix model revision
- −48%
- unmet demand · 2023 to 2026 versions
- Regulatory verdict
- Same
- regional run · default route still fails
- Ranegras draft goal
- 50 / 50
- half the overdraft · fifty years
01 · The estimate moved
The model's shortfall estimate fell nearly in half. The default route still failed.
In 2023, Arizona's Phoenix groundwater model projected 4.862 million acre-feet of unmet demand over one hundred years. A 2024 update cut that estimate to 3.604 million. Version 3, released in June 2026, cut it again—to 2.5145 million acre-feet.
The policy conclusion did not move. Inside the Phoenix Active Management Area, a proposed subdivision generally must demonstrate a 100-year assured supply. The regional Version 3 run still does not, by itself, support a new groundwater-based determination. A project needs a designated provider, another approved supply, an eligible demand-reduction credit, or an accepted project-specific model revision.
This is what a model is supposed to do: change when assumptions and inputs change. It is also what a threshold does: produce the same legal answer even when the estimate moves a long way.
Source: ADWR, Phoenix AMA Groundwater Model and 100-Year AWS Projection, Version 3, June 2026, Executive Summary ES-3.
02 · The model is a priority map
Not every modeled shortfall counts the same way.
Version 3 divides its 2.5145-million-acre-foot shortfall across legal categories. 836,500 acre-feet falls in assured-water, municipal, and long-term-storage-credit sectors that must be satisfied for the assured-supply program. The remaining 1.678 million falls in “unprotected” agricultural and industrial sectors outside that test.
“Protected” is a term in this model, not a promise that every person or well is safe. But the split reveals the system: the model is not only hydrology. It is hydrology processed through a statute that asks whether adding a particular kind of demand is permissible.
The rule generally attaches to proposed subdivisions of six or more lots inside an AMA. A developer may qualify through a designated provider or another approved supply; “the next house needs one hundred years” is a useful shorthand, not the law's complete shape.
Sources: ADWR Version 3 memo, ES-3; ADWR Assured and Adequate Water Supply overview.
Version 3 · 2.5145M AF unmet demand
The legal stress test
A qualifying groundwater supply must survive the model's full projection without unmet protected demand.
03 · Cross the line
The aquifer changed first. The legal boundary arrived later.
Ranegras Plain offers a different crisis. One monitored well—not a basin average—fell more than 240 feet from the 1980s to the 2020s. An October 2025 ADWR hydrologic memo estimated roughly 42,000 acre-feet leaving the aquifer system each year and about 4,500 entering it: more than nine out for every one in.
On February 4, 2026, Ranegras became Arizona's eighth Active Management Area. Most non-exempt wells now face metering and reporting, and the expansion of irrigated acreage is permanently prohibited subject to limited exceptions. Historic users may qualify for grandfathered rights, but they must apply. Designation started accounting and constrained new demand; it did not erase the deficit.
As of September 1, the management goal was still being drafted. The proposal was to reduce groundwater overdraft by 50 percent in 50 years—too slow for some residents, too onerous for some agricultural users.
Sources: ADWR Ranegras hydrologic memo, October 24, 2025; AMA designation materials and draft management-goal page. The use estimates come from separate analyses and are not combined here.
04 · Who absorbs scarcity?
A river shortage is also a priority system.
On August 29, 2026, Lake Mead's federal gauge read a provisional 1,038.87 feet—157.92 feet below the same date in 2000 and below the August point on this chart from 2022.
Arizona's household taps did not fall by the same proportion. Under the 2026 Tier 1 shortage, Arizona lost 512,000 acre-feet, 30 percent of the Central Arizona Project's normal supply. Nearly all the reduction fell on CAP users; the CAP priority system left less river water for central-Arizona agriculture.
Reclamation's 2027–28 guidelines fix a 1.25-million-acre-foot Lower Basin reduction in each year. A proposed implementing agreement would allocate that reduction; absent one, the Secretary apportions it under applicable law. A broader framework points toward 2036 and allows longer operating periods if the states agree. Two years is the current rule, not a law of nature.
Sources: Bureau of Reclamation RISE gauge and 2027–28 decision; Central Arizona Project Colorado River operations page.
The useful question
Arizona has water. Arizona is running out. Both sentences are incomplete.
Complete the claim with four questions:
- 01Which basin?
- 02Which demand?
- 03Under which right?
- 04In which year?
The tap can stay on while the permit stops. A model can improve while the verdict holds. And a legal boundary can arrive decades after the water falls.
Reporting notes
Measured, modeled, estimated, and legal
The current Phoenix figures come from ADWR's June 2026 Version 3 technical memorandum. It reports 2,514,500 AF of unmet demand, including 836,500 AF in AWS, municipal, and long-term-storage-credit sectors. The 2023 and 2024 numbers are previous model versions, not observed aquifer deficits. Download the chart data.
ADWR's Assured and Adequate Water Supply overview explains the 100-year test, geographic limits, and designated-provider route. “Protected” and “unprotected” are model/program categories used in the Version 3 memo; they are not general guarantees of service. ADWR's Ag-to-Urban program documents one eligible demand-reduction-credit path.
Ranegras hydrology comes from ADWR's October 2025 memo. The well decline is one monitored well, not a basin average or proof that a domestic well ran dry. The designation requirements and draft 50-in-50 goal come from ADWR's Ranegras AMA page. The competing stakeholder views are recorded in ADWR's April 2026 comment memorandum.
The Lake Mead line uses one August 29 observation per year from the Bureau of Reclamation's RISE catalog. The 2026 observation is provisional; elevation is not proportional to volume. Download the annual series.
Reclamation's 2027–28 guidelines set the Lower Basin reduction. CAP's operations page supplies the 2026 Tier 1 figures and explains how priority directs the cut.