On Friday 21 August the Bureau of Reclamation published its August 2026 24-Month Study twice. Two distinct PDFs went up on the Upper Colorado region's server carrying Last-Modified timestamps thirteen seconds apart, each one labelled an official Most Probable projection: one assumes Lake Powell releases 6.0 million acre-feet in water year 2027, the other assumes 7.0 maf. Through April 2027 the two documents give Lake Mead the identical elevation, 1,018.28 feet. Then they diverge, and they diverge across a number the same day's brand-new rulebook had just made operative. The 6.0 maf study shows Lake Mead at 1,005.63 feet in July 2027, below the 1,010-foot elevation at which the Secretary of the Interior “shall consult and coordinate with the Basin States and Basin Tribes”; the 7.0 maf study never falls below 1,013.92 feet.
Reclamation did not decide to publish twice. Table 1 of Section 5.1.C.2 of the 2027-2028 Operating Guidelines, signed by Interior Secretary Doug Burgum the same day, orders it in the August row: “Perform at least 2 ‘official’ August 24-Month Studies”, where “Runs assume a Water Year release of 6.0 maf and 7.0 maf”, and the study is to “State that the expected Water Year release will be between 6.0 and 7.0 maf”. Both studies comply to the letter. Each narrative says the expected WY 2027 release “will be in the range between 6.00 million acre-feet (maf) to 7.00 maf”, and neither names a single volume — a first for an August study, every one of which from 2013 through 2025 named exactly one. Reclamation's own Upper Colorado study index, marked last updated 8/21/26, now carries two entries in the August row where every other month carries one.
The clause the two documents split over is Section 5.3.A.4 of the same guidelines, which reads: “Should the Most Probable projection of any 24-Month Study show Lake Mead falling below 1,010 feet at any time in the subsequent 12-month period, the Secretary shall consult and coordinate with the Basin States and Basin Tribes, and other affected parties as appropriate”. On the 6.0 maf table, Lake Mead's end-of-month elevation sits below 1,010 feet for six consecutive months, June through November 2027, reaching 1,005.63 feet in July. On the 7.0 maf table the same twelve months never take it below 1,013.92 feet. The clause turns on “the Most Probable projection of any 24-Month Study”, and the guidelines designate both runs official Most Probable studies; nothing on the face of either document says which one governs, and neither resolves the question against the other.
After April 2027 the two tables separate steadily: 1,011.16 feet against 1,015.13 in May, 1,006.46 against 1,013.92 in June, 1,007.83 against 1,021.99 in August. By 31 December 2027 they stand 11.61 feet apart, 1,011.59 against 1,023.20. That gap is precisely what the April 2027 determination decides, because the guidelines defer the actual water-year release volume to that point. Upstream the two runs are deliberately identical for longer — they share the same October-through-April release pattern and bottom out at the same 3,511.70 feet at Lake Powell in February 2027, because Table 1 requires both to hold a 3,510-foot minimum target through March — and only pull apart by 30 September 2027, at 3,533.24 feet under the 6.0 maf assumption against 3,517.34 feet under 7.0.
Naming no release volume does not make the study indeterminate about 2027 deliveries. Both narratives state that “a shortage condition consistent with Section 5.3.A will govern the operation of Lake Mead for calendar year (CY) 2027”, and the projections put Lake Powell at 3,515.89 feet on 1 October 2026, below the 3,540 feet that places water year 2027 in the Low Elevation Infrastructure Protection Range. Section 5.3.A is where the new rulebook departs most sharply from the old one: it is elevation-independent. Section 5.3.A.1 has the Secretary determine that a Shortage Condition exists for both 2027 and 2028 with no trigger elevation anywhere in it, and Section 5.3.A.2 then apportions the Lower Basin a flat 6.25 maf — Arizona 2.04 maf, California 3.96 maf, Nevada 250,000 acre-feet. That is a reduction of 1.25 million acre-feet, with a further 700,000 acre-feet of System Conservation spread across 2026 to 2028 under Section 5.3.B.
Both studies also agree on the number that has carried the story for a decade. Lake Mead is projected at 1,034.74 feet on 1 January 2027 — identical in the 6.0 and 7.0 maf runs, with a Probable Minimum of 1,034.19 and a Probable Maximum of 1,035.63, and down 2.57 feet from the 1,037.31 feet the July edition carried. Re-derived from the seventeen archived August studies, that is the lowest 1-January Lake Mead projection any August study has printed since 2010. The previous low was 1,040.78 feet, in August 2022, and that year's operative determination applied an adjustment treating 0.480 maf as delivered and placed Mead “below 1,050 feet and above 1,045 feet”, so on operative figures the gap is wider still.
Under the rulebook that expired, 1,034.74 feet would have been an input. Section 2.D.1.b of the 2007 Interim Guidelines — below 1,050 feet, at or above 1,025 — would have cut Arizona 400,000 acre-feet and Nevada 17,000, and California nothing; Table 1 of Exhibit 1 to the Lower Basin Drought Contingency Plan, row “at or below 1,035 and above 1,030”, would have added Arizona 240,000, Nevada 10,000 and California 300,000. Combined, 967,000 acre-feet. The new Section 5.3.A instead produces 1,250,000 acre-feet, 283,000 more and 29 percent above the counterfactual, and arrives there without reading the elevation at all. The most closely watched gauge in the basin now sits at a record low in a document that no longer uses it to decide anything.
One consequence of that flat apportionment deserves precise phrasing. Sections 2.D.1.a, 2.D.1.b and 2.D.1.c of the 2007 Interim Guidelines all leave California apportioned 4.4 maf: at no shortage level did the old rulebook reduce California's apportionment. Section 5.3.A.2 apportions California 3.96 maf, making the 2027-2028 guidelines the first federal Colorado River operating rule to cut it. That is a reduction in apportionment rather than a contribution, and the distinction matters — California could already owe DCP contributions with Mead below 1,045 feet, and at this elevation the old pairing would have asked it for 300,000 acre-feet.
This publication tracked the same 1-January figure across the 2026 editions on 16 August: 1,060.13 feet in January, 1,056.36 in April, then 1,037.03 in May, a 19.33-foot fall in a single edition, and now 1,034.74, down 25.39 feet on the year. The framing then was that the rulebook which would have turned that elevation into a water cut expires first. It did expire. A successor was signed on 21 August, and it did turn the year into a water cut — 1.25 maf, larger than the expiring rules would have produced — just not through that number.
The August study is also, five days on, missing from the archive built to hold it. As of Tuesday evening UTC the Lower Colorado region's own Most Probable slot for the month, /lc/region/g4000/24mo/2026/AUG26.pdf, still returns HTTP 404 to a plain request and to a Chrome user agent alike, and the archive index page — itself last modified in February 2024 — still builds exactly that URL for its Most Probable button. The page's “current month's study” link serves a file whose SHA-256 is byte-for-byte the July study. The August Probable Minimum, Probable Maximum and Lake Mead chart all posted to that same 2026 directory on 21 August, and the region's river operations page now links the two new studies directly. The 16 August observation that the slot was empty remains literally true and now supports the opposite inference: the study is out, and the region's archive cannot serve it.
One apparent discrepancy in the studies is not one. Both narratives say Lake Powell “is projected to decline below 3,510 feet during water year (WY) 2027” while neither published table goes below 3,511.70 feet, because Section 5.1.C.2 tests a separate unadjusted 7.0 maf “Exhibit Run” that is not published, and both published runs already carry the corrective October-to-April release pattern that holds the 3,510-foot target. The 1,010-foot question at Lake Mead is a different matter, and it does not dissolve on inspection. Either the April 2027 determination collapses the two studies into one, or the Secretary has to read Section 5.3.A.4 against a pair of documents his own department published as equals, thirteen seconds apart, on the day the section took effect.