Skip the animated story A house drawn above an Arizona aquifer A continuous blueprint line draws a house, becomes a reservoir level, crosses a management boundary, and descends to a rural well. THE NEXT HOUSE THE LINE THE AQUIFER DOES NOT STOP HERE

ARIZONA · THREE WATER CLOCKS

The house
and the hundred years.

Arizona's water crisis is easiest to misunderstand from a kitchen sink. The water still comes out. The permit for the next house—and the depth of a rural well—tell the better story.

THE SYSTEMS DIAGRAM

There is no single Arizona water crisis.

There is a Colorado River allocation problem, an urban growth test, and a rural groundwater problem. They overlap. They do not fail in the same place, on the same day, or for the same person. The most consequential piece of water infrastructure may be a line on a map.

LAKE MEAD · AUGUST 29, 2026

First clock: the river.

The federal gauge read 1,038.87 feet—16.52 feet below the same date one year earlier. The chart follows one official observation every August 29 since 2000. Reclamation reported that Lake Mead reached its lowest recorded elevation in August 2026; the final observation is provisional.

THE RULES · 2027–28

The river got two years.

With no long-term consensus among the basin states, Reclamation adopted a 1.25-million-acre-foot annual reduction from normal Lower Basin apportionment for 2027 and 2028. The proposed state split requires implementing agreements. The framework lasts through 2036, but only 2027–28 have operating guidelines; Reclamation expects later guidelines in two-year intervals unless the parties reach a longer agreement.

Arizona was already under a Tier 1 shortage in 2026: a 512,000-acre-foot reduction equal to 30 percent of CAP's normal supply. CAP says nearly all of Arizona's reduction was taken within its priority system, leaving less Colorado River water for central Arizona agriculture.

MEAD ELEVATION
1,038.87 FT
MEAD · 2000
1,196.79 FT
CHANGE SINCE 2000
−157.92 FT
LOWER BASIN REDUCTION
1.25M AF / YR

ASSURED WATER SUPPLY

Second clock: one hundred years.

The clocks measure different things: two years is the life of the current federal operating rules; one hundred years is the horizon Arizona uses to test new development.

Inside an Active Management Area, a new subdivision generally must demonstrate a 100-year assured supply before lots can be sold. This is not a prophecy that every year will be wet. It is a legal stress test attached to the next house. Outside those boundaries, an inadequate supply can often be disclosed rather than prohibited.

RIVER GUIDELINES
2 YEARS
NEW-SUBDIVISION TEST
100 YEARS

PHOENIX AMA · 2023 MODEL

The model stopped the next house, not today's tap.

ADWR's 2023 Phoenix AMA projection found 4.86 million acre-feet of groundwater demand would go unmet over 100 years because modeled wells ran dry. Separately, parts of the projection exceeded the Assured Water Supply program's 1,000-foot depth-to-water limit.

In the Phoenix AMA, the state stopped approving new assured-supply determinations based only on groundwater; existing certificates and designated providers remained valid. ADWR's 2026 update changed annual inputs and technical details but did not recalibrate the 2023 model.

MODELED UNMET DEMAND
4.86M AF
HORIZON
100 YEARS
DEPTH-TO-WATER LIMIT
1,000 FT

RANEGRAS PLAIN · THE EIGHTH AMA

Cross the line. Change the rules.

The aquifer does not recognize a jurisdiction. The rules do. In Ranegras Plain, ADWR estimates roughly 42,000 acre-feet leaves the aquifer system each year while about 4,500 enters it—more than nine out for every one in. Agriculture used an estimated 40,083 acre-feet in 2023. Self-supplied homes used 70. A shorter shower was never going to close that ledger.

One monitored well fell more than 240 feet since the 1980s; parts of the basin subsided by up to 40 centimeters between 2010 and 2025. ADWR's director ordered Ranegras designated in January 2026, and the designation became final February 4. Most high-capacity wells now face metering and reporting; new irrigated acreage is generally prohibited while the AMA develops its management plan.

ONE WELL · SINCE THE 1980S
−240+ FT
SUBSIDENCE · 2010–25
UP TO 40 CM
AMA FINAL
FEB 4, 2026

WHAT THE BLUEPRINT SHOWS

Existing is not the same as next.

A smaller Colorado River apportionment, a subdivision that needs another source, and a domestic well that no longer reaches water are different events borne by different people. Any claim that “Arizona has water” or “Arizona is running out” needs four qualifiers: where, for whom, under what legal right, and for how long?

The river has two years of operating rules. The next house still needs a hundred-year supply.

REPORTING NOTES

How to read the lines

01

The archive date keeps this experiment below newer posts; the reporting is current through August 30, 2026. No 2026 observation or policy decision is presented as information available in 2024.

02

The Lake Mead line uses an observed elevation from August 29 of each year in the Bureau of Reclamation's RISE catalog. The 2026 value is provisional. Elevation is measured in feet above mean sea level and is not proportional to reservoir volume. Download the plotted data.

03

Reclamation's 2027–28 operating guidelines set the Lower Basin total. The allocation among Arizona, California, and Nevada remains conditional. Its decision page distinguishes the two-year guidelines from the 2027–36 framework. CAP's Colorado River operations page supplies the 2026 Tier 1 figures and explains where the reduction lands in its priority system.

04

Arizona's Assured and Adequate Water Supply overview explains the 100-year test and its geographic limits. The Phoenix AMA updates explain the 4.86-million-acre-foot projection and why it constrained new groundwater-based determinations rather than existing certificates or designated providers.

05

The Ranegras figures come from ADWR's AMA designation announcement, its basin assessment, and the designation order. The well decline is one monitored well, not a basin average. The use estimates are basin-scale and some unmetered pumping is modeled.