For two months in the summer of 1983, the people running Glen Canyon Dam had the opposite of today’s problem. Too much water was arriving, the spillways built to pass it were destroying themselves, and the emergency fix involved sheets of plywood.
The short answer
An unusually cold spring stacked snow in the Rockies into late May 1983, then a heat wave and heavy rain melted it all at once. Lake Powell was already near full. Glen Canyon Dam opened its spillway tunnels on June 2, and within four days high-velocity water had begun ripping the concrete lining apart through cavitation. Operators bolted plywood, then steel, onto the spillway gates to store water they could no longer safely release. The lake crested at 3,708.34 feet on July 14, 1983 — still the all-time high — and the dam held.
The elevations that mattered
Everything in 1983 turned on a few fixed marks. None of these are today’s reading, which is in the callout above.
| Elevation | What it is |
|---|---|
| 3,715 ft | Top of the dam (structural crest) |
| 3,708.34 ft | The 1983 peak — the all-time high, July 14 |
| 3,708 ft | Top of the 8-ft steel gate extensions |
| 3,704 ft | Top of the 4-ft plywood flashboards |
| 3,700 ft | Full pool, and the top of the closed radial gates |
| 3,648 ft | The concrete spillway sill under the gates |
At its highest, the reservoir sat 6.66 feet below the dam crest, per Reclamation’s hydraulic elevation records. The full record book, including the modern low that sits 188 feet below that 1983 mark, is in Lake Powell’s water level records.
How the forecast missed
The January 1983 runoff forecast sat at 112% of average, and by March it had actually fallen to 96%. A wet March pushed April back to 114%, and by mid-May forecasters were at 120% (USBR 1983 Annual Operating Plan). None of those numbers called for an emergency.
The melt’s timing is what caught Reclamation off guard. Unusually cold temperatures kept snow accumulating at mountain elevations until roughly May 21, so the snowpack peaked weeks late and then had to come off fast (USGS Water-Supply Paper 2300). A first rapid melt hit May 23–28. The first two weeks of June turned cold again, stalling it. Then a heat wave on June 18–23 put mountain temperatures in the 60s while Grand Junction reached 95°F, and widespread thunderstorms landed on top of the melt during June 24–28 (USGS, Water Resources Data — Colorado, 1983).
The forecast chased it the whole way. Reclamation raised its April–July projection to 11.3 million acre-feet on June 14, 13.3 MAF on June 18, and 14.6 MAF on June 22. Final April–July inflow came in around 14.5 MAF, roughly 190% of average and the largest since 1920. Peak inflow hit 111,500 cubic feet per second on June 27 (USBR spillway-operation paper, PAP-0714). Modern melt seasons move the lake in the other direction entirely — how much snowpack raises Lake Powell walks through today’s math.
The 1983 timeline
| Date | What happened |
|---|---|
| June 2 | Left spillway opened at 10,000 cfs |
| June 5 | Left spillway raised to 20,000 cfs |
| June 6 | Loud rumbling from the left tunnel; holes found in the invert. Plywood flashboards begun that evening |
| June 10–11 | Plywood installed on the right-gate side |
| June 27 | Peak inflow, 111,500 cfs |
| June 28 | Left tunnel raised to 32,000 cfs, then cut to 20,000 as concrete and sandstone came out the outlet |
| June 29 | Peak flow past Lees Ferry, 97,300 cfs |
| July 4 | Eight-foot steel extensions begin replacing the plywood |
| July 7 | Gates closed for the first full inspection of both tunnels |
| July 14 | Lake Powell crests at 3,708.34 ft |
What cavitation did to the tunnels
Cavitation is what happens when water moves fast enough that its own pressure drops far enough to boil. Vapor bubbles form, travel downstream, and collapse against the concrete with enough force to break it. Each collapse leaves a slightly rougher surface, which makes the next collapse worse.
At Glen Canyon it started with a calcite deposit 0.25 inches high on the tunnel floor, near the point where the 41-foot tunnel bends toward horizontal. After roughly 72 hours at 20,000 cfs, operators heard rumbling on the morning of June 6 and found several large holes in the invert. About 50 cubic yards of concrete were already gone.
It got much worse before it stopped. Reclamation’s dam-safety documentation records a final crater in the left tunnel measuring 35 feet deep, 134 feet long, and 50 feet wide, with about three-quarters of the liner circumference missing at the deepest point (USBR Best Practices, Chapter F-3). The right tunnel lost its invert liner for about 175 feet, with sandstone excavated as much as 12 feet deep, and exposed reinforcing steel that inspectors described as looking like spaghetti.
The plywood that bought over a million acre-feet
With the spillways eating themselves, operators needed to stop releasing water through them. That meant storing more, which meant raising the height the radial gates could hold.
The decision came around 6 p.m. on June 6. Crews started about two hours later and finished roughly 22 hours after that: four-foot flashboards, built from upright 4×8 sheets of plywood, bolted to the tops of the gates. The right-side gates got theirs on June 10–11. In early July, eight-foot reinforced steel extensions replaced the plywood, and structural analysis showed they would only marginally overstress the gates — accepted as a temporary emergency condition.
Those extensions bought roughly 1.2 to 1.4 million acre-feet of storage that the dam had no other way to hold, depending on which Reclamation figure you use.
Was the dam ever going to fail?
Not according to the people who ran it, then or now. Reclamation’s own position is direct: during the events of 1983, Glen Canyon Dam “was never in danger of failing” (Bureau of Reclamation). Later geological review and a Safety Evaluation of Existing Dams reassessment reported no changes to the dam or its foundation. Reclamation’s operations manager at the time, Tom Gamble, told reporters plainly that the dam was not going to break.
What was at risk was narrower. Operators could have lost controlled spillway operation — the left tunnel’s flip bucket was eroding and might have stopped functioning, and the right tunnel’s bend sits nearer the dam foundation, which is why they protected it as long as they could. Water overtopping the gates was a live concern. Water overtopping the dam was not; the lake stayed nearly seven feet under the crest.
Popular retellings tend to collapse those two into one. Reclamation’s modern dam-safety guidance notes that cavitation inside a confined tunnel is less likely to progress to dam failure than cavitation in an open spillway, because the flow stays directed. The 1983 crisis was real, and no source puts a failure probability on it.
What changed afterward
The repair was an engineering redesign. Crews cut aeration slots into each inclined tunnel, upstream of the bend where the damage started. A small ramp lifts the water sheet just enough to draw air underneath it, and that compressible air layer keeps collapsing vapor bubbles off the concrete instead of against it (ASDSO case study).
The work ran roughly $30 million and was finished ahead of the next runoff season. On August 12, 1984, the repaired left spillway was tested at 50,000 cfs, then dewatered and inspected. No cavitation damage was found.
Downstream, the releases were their own event. Flow past Lees Ferry peaked at 97,300 cfs on June 29, 1983, the largest post-dam flood at that gage, and stayed above 31,500 cfs for 68 consecutive days (USGS Circular 1366).
Why 1983 still sets the ceiling
Forty-plus years later, 3,708.34 feet remains the number every Lake Powell record is measured against, and nothing since has come within several feet of it. The lake was last logged as essentially full — 99% — in June 1987, covered in when Lake Powell was last full. The opposite bookend came on April 13, 2023, at 3,519.92 feet.
In 1983 the emergency was getting water out of a reservoir that would not stop filling. The modern emergency runs the other way: Reclamation now cuts releases to the legal minimum to hold elevation, and the thresholds operators watch sit below the surface rather than above it. The dead pool tracker shows the cushion above 3,370 feet today, the water level chart puts the current reading against the full 365-day series, and will Lake Powell fill back up covers what a return toward those 1980s levels would actually require.
Sources
- U.S. Bureau of Reclamation — Operation of Glen Canyon Dam Spillway, Summer 1983 (PAP-0714) (spillway chronology, cavitation initiation, peak inflow, flashboards)
- U.S. Bureau of Reclamation — 1983 Annual Operating Plan, Colorado River (PDF) (runoff forecasts, April–July inflow, peak elevation and release)
- U.S. Bureau of Reclamation — Dam Safety Best Practices, Chapter F-3: Cavitation Damage (measured crater dimensions, failure-progression guidance)
- U.S. Bureau of Reclamation — Glen Canyon Dam, Upper Colorado Region (“never in danger of failing”; post-1983 safety reviews)
- U.S. Bureau of Reclamation — Glen Canyon Dam hydraulic report HL-2011-02 (PDF) (dam crest, gate and spillway sill elevations)
- U.S. Geological Survey — Water Resources Data, Colorado, Water Year 1983 (PDF) (melt timing, June heat wave and storms)
- U.S. Geological Survey — Circular 1366, Colorado River in Grand Canyon (PDF) (Lees Ferry peak flow, 68 days above 31,500 cfs)
- Association of State Dam Safety Officials — Glen Canyon Dam (Arizona, 1983) case study (incident summary, aeration slot function)
Hero photo: 2018 High-flow Release, Glen Canyon Dam by Grand Canyon NPS / M. Quinn, licensed CC BY 2.0, via Wikimedia Commons (cropped). It shows a 2018 experimental release, not the 1983 flood.
