Charted: Colorado River Compact Allocations vs Actual Flows

November 1922, Bishop's Lodge near Santa Fe. The light is thin, the whiskey is neat, and seven states are about to divide a river whose gauged record spans barely two decades — years hydrologists would later recognize as unusually wet. The bargain still governs the basin. At Lees Ferry — the Compact Point dividing Upper and Lower Basins — Reclamation's naturalized water-year series shows what the river actually delivers before storage operations and depletion accounting. Since 2000, that natural flow has averaged about 12.4 million acre-feet (maf) per year against the ~16.4 maf supply the Compact-era planners assumed. The gap is about 4.0 maf per year, or 24% — not a rounding error, and not a single dry winter. In this basin, the past is never dead; it's on the water ledger.

The dashboard frames the problem as a flow-versus-budget comparison, with switchable views for natural-flow overlays, era gaps, paper claims, and use-versus-entitlement, and adjustable windows running from the full post-1964 Glen Canyon record down to the last decade. The structural point survives every window: legal paper still adds to 16.5 maf while recent hydrology clears closer to 12–13.

What the Compact actually allocated

The Colorado River Compact of 1922 split the system at Lees Ferry. Article III(a) apportioned 7.5 maf of beneficial consumptive use to the Upper Basin and 7.5 maf to the Lower Basin. Article III(b) then granted the Lower Basin "the right to increase its beneficial consumptive use … by one million acre-feet per annum" — an unconditional statutory increment, though how it interacts with surplus and Mexico's share is still argued today [Colorado Compact 1922]. Article III(d) bound the Upper Division not to deplete Lees Ferry flow below 75 maf in any ten consecutive years (a 7.5 maf/yr equivalent floor on delivered flow, not on natural runoff).

The Compact did not invent Mexico’s share. The 1944 US–Mexico Water Treaty later set a 1.5 maf delivery obligation. Stack the two basin compact blocks with Mexico and the familiar 16.5 maf of paper claims appears. Negotiators in the early 1920s were working from a short, wet gage era; modern summaries put their implied mean near 16–17+ maf. Reclamation and CRS now publish a 1906–2024 natural average near 14.6 maf — already below the paper stack before the 21st-century drought [CRS Colorado River].

Lees Ferry natural flow is the right yardstick

Naturalized Lees Ferry flow estimates the runoff that would arrive without upstream storage changes, evaporation from reservoirs, and consumptive depletions. It is the hydrology clock. Gage flow below Glen Canyon Dam is the operations clock — releases, equalization, and drought contingency rules can hold gage volumes near 8–9 maf even when natural runoff collapses.

For budget stress, natural flow is the harder truth. A year like 2002 (~5.9 maf natural) or 2021 (~7.2 maf) cannot fund 16.5 maf of simultaneous claims no matter how cleverly Powell and Mead are managed. Wet spikes (2011 ~20.2 maf, 2023 ~17.4 maf) refill storage and buy political time; they do not rewrite the multi-decade mean.

GCDAMP hydrology tables compiling Reclamation natural-flow accounting put the 2015–2024 ten-year natural average near 12.6 maf — still about 3.8 maf under the compact-assumed 16.4 [Reclamation Natural Flow]. The millennial-drought window (2000–2024) averages 12.4. Those are not provisional blog estimates; they are the same order of magnitude Reclamation cites in post-2026 operations materials (stress-test ~13.0; millennial drought ~12.4).

Paper claims versus observed eras

The table is the desk version of the dashboard's Era gaps panel. Long-term hydrology was already short of paper; the 21st century opened a four-maf hole relative to the Compact's mental model. That hole is larger than Nevada's entire Lower Basin apportionment (0.3 maf) by an order of magnitude, and comparable to a large share of Arizona's 2.8 maf.

Lower Basin math is fixed; Upper Basin math is residual

The Boulder Canyon Project Act (1928) carved the Lower Basin’s 7.5 maf into fixed state shares: California 4.4, Arizona 2.8, Nevada 0.3. Those numbers are politically sticky because they are statutes, not percentages of leftover water. Shortage guidelines and Drought Contingency Plan tiers cut junior users first — Arizona’s Central Arizona Project and Nevada feel curtailment earlier than California’s senior contractors — but the headline apportionments do not shrink when Lees Ferry natural flow averages 12.4.

Upper Basin states live under a different geometry. The 1948 Upper Colorado River Basin Compact assigns percentages of whatever remains after Lee Ferry duties (Colorado 51.75%, Utah 23%, Wyoming 14%, New Mexico 11.25%, plus Arizona’s small fixed 50 kaf Upper Basin slice). In practice, Upper Basin consumptive use has recently hovered near ~4.5 maf — well under the 7.5 paper entitlement — precisely because meeting the non-depletion delivery culture and protecting Powell elevations leaves less water to develop. The Lower Basin, by contrast, has historically run much closer to a full 7.5 take before ICS, conservation programs, and shortage tiers forced cuts.

That asymmetry matters for post-2026 talks. A basin that “balances” on paper by leaving Upper Basin entitlement unused is not in surplus; it is using under-development as a hidden buffer. If Upper Basin use rises toward 7.5 while Lower Basin refuses to permanently reduce its take, the 16.5 paper stack collides harder with 12-ish hydrology.

Reservoirs deferred the reckoning — they did not cancel it

Lake Powell and Lake Mead were built as multi-year batteries. When natural flow runs below consumptive use plus losses, storage falls. The 2000s–2020s drawdown sequence — Mead and Powell both visiting elevations that triggered shortage tiers and emergency releases — is the physical expression of the table above. Wet years 2011 and 2023 proved the battery can still charge; they did not prove the Compact’s assumed mean has returned.

Evaporation and system losses add another quiet debit. Basin accounting often treats ~1+ maf of combined reservoir evaporation and phreatophyte / operational losses as real water that never reaches a contractor. That volume sits outside the neat 7.5 / 7.5 / 1.5 story even when deliveries look “full” on a Lower Basin water-accounting report.

What the rolling ten-year lens shows

Compact Article III(d)’s 75 maf / 10 yr floor is a delivery constraint on Upper Division depletion, not a promise that nature will supply 7.5 maf of virgin flow forever. The dashboard’s rolling ten-year natural mean stays well above 7.5 in most windows — natural runoff averaged over a decade is usually 12–16 maf even in drought decades — which is why the non-depletion clause can be met while the allocation stack still exceeds supply. The political crisis is not that Upper Basin states failed the letter of III(d); it is that simultaneous full use of all paper claims is hydrologically impossible under millennial-drought means.

Desk takeaway: do not confuse "the Compact delivery floor was met" with "the Compact budget is solvent." The first is a legal test on cumulative Lee Ferry volumes. The second is an arithmetic test on whether 16.5 can be satisfied from 12.4. Out here, arithmetic always collects.

Caveats and limits

Natural-flow series are model reconstructions, revised when consumptive-use-and-losses data are updated. GCDAMP / Reclamation tables can differ slightly from a given Excel vintage; treat individual years as ± a few tenths of an maf and focus on multi-year means. Compact “assumed supply” (~16.4) is a historiographical figure — what negotiators thought the river would yield — not a single line in Article III. Mexico’s 1.5 maf is a treaty delivery with Minute flexibility under extreme shortage; it is not identical to a U.S. state apportionment. Upper Basin “~4.5 maf use” is an approximate recent consumptive band, not a certified CUL year. Shortage tiers, Intentionally Created Surplus, and system-conservation programs change who gets cut without rewriting the underlying hydrology gap. This post is an analytical comparison of published flow means against statutory paper — not an operating guideline, legal opinion, or forecast of post-2026 alternatives.

Sources and method

Primary flow path: Reclamation natural-flow accounting as compiled in GCDAMP hydrology tables for Lees Ferry water-year naturalized volumes (1964–2025 sample used in the dashboard). Apportionments: 1922 Compact text; Boulder Canyon Project Act Lower Basin splits; 1944 Mexico Treaty. Era averages: Reclamation / CRS published summaries for 1906–2024 (~14.6), stress-test (~13.0), and millennial drought (~12.4). Gap percentages divide (16.4 − era mean) by 16.4. Use-versus-entitlement bars are illustrative recent levels for Upper Basin under-use versus Lower Basin near-full take — labeled as such in the data module.