Pre-Project Klamath Lakes

Without Project Effects

This animation shows the Klamath lakes without the federal action, meaning the Klamath Reclamation Project (the Project), which was authorized on 15 May 1905. Agricultural development was already under way before then: irrigation companies had built canals, and the Project later acquired some of them (OAH Case 003, Findings 14 and 16). So the model keeps irrigation demand at 1911–1920 levels. It uses today's watershed with upper-watershed changes still in place. River flows and lake levels look very different without control structures.

Each water year runs day by day. Rain and snowmelt fill Upper Klamath Lake. When the lake is high enough, water flows down the Link River to Lower Klamath Lake. Water leaves Lower Klamath Lake over the Keno reef into the Klamath River. Both lakes also lose water to the air. Water year 1904 was the most exceptionally wet year in Klamath County's 1896–2026 record.

Pick a water year type or a featured year. The full 65-year sequential record is on a separate page: Sequential record, 1962–2026.

Type years:
Featured years:
Water Year 2016 · 1 October

Pre-Project Development.

Inflow0 cfs
Upper lake0 ft
Over Link River0 cfs
Lower lake0 ft
Over Keno0 cfs
Lost River Slough0 cfs
Out the LR Slough so far0 AF
To the air0 cfs
Inflow so far0 AF
UKL change since 1 Oct0.0 in
Past Keno so far0 AF

Change in Upper Klamath Lake stage through the water year

Hydrograph — maximum available over the Keno reef in the pre-Project setting

Flow-duration curve — Keno, this water year

What You Are Looking At

How to read the picture

The blue ribbon at the top shows water flowing into Upper Klamath Lake that day. A wider ribbon means more water. The two lake levels are in feet. The two bars below add up the year so far. The top bar is all the water that has arrived. The bottom bar is all the water that has passed Keno. The gap is water still stored in the lakes or lost to the air. The rain gauges show what fell at Klamath Falls and at Keno that day.

The three water years

Dry (1992), median (2016) and wet (1999) are real years chosen to show a range. They are labels, not percentile rankings. Upper Klamath Lake receives about 580,000 acre-feet in the dry year, about 1,010,000 in the median year and about 1,700,000 in the wet year. One acre-foot covers one acre to a depth of one foot.

How years start

Each type year starts on 1 October with Upper Klamath Lake at 4,139.46 feet, so any difference comes from that year's water, not the starting point. In the sequential record, each year starts from where the previous year ended.

The charts below the animation

The stage chart shows how far the lake rose or fell since 1 October, in inches. The dashed line is the Link River Dam ceiling, 4,143.3 feet, where the regulated lake has been held since 1922. The red line is the highest level on the USGS record, 4,144.98 feet, from high-water marks about 20 April 1904, before the dam. The hydrograph traces the year's flow into the lake, out of the Link River and past Keno. On the flow-duration curve, each point shows a flow and the share of days the river carried at least that much.

The Keno reef and the Lost River Slough

The Keno reef sets the lowest level Lower Klamath Lake can reach. The Lost River Slough carries water from Lower Klamath Lake toward Tule Lake when the lake is high enough. A dike built in 1889–1890, before the Project was authorized on 15 May 1905, closes that outlet. The dike is in place by default, so no water reaches the Slough. Remove it and the Slough reopens. The lake then spills into the Slough on 105 days in the median year and 215 days in the wet year. In the dry year the lake never rises above 4,085 feet, so nothing spills. Even when open, the Slough carries little water: about 24,400 acre-feet in the wet year, roughly 1.4% of that year's inflow.

Using the controls

Dry, Median and Wet each run one year from the same start. The featured buttons show two notable years. Play runs the selected year, and the slider moves through it by hand. Green sensitivity buttons are included in the Pre-Project setting; pink buttons are removed. Click any button to change it.

What this is not

This is a what-if, not a forecast. Today's watershed has changed over a century. Upstream diversions, irrigation on the Wood, Williamson and Sprague rivers, and denser forest all reduce the water that reaches the lake. The picture shows the lakes before the Project on today's basin, not as the basin was in 1905. The model also leaves out groundwater flowing into Lower Klamath Lake.

Technical description for hydrologists (click to open)

Status: a daily two-lake water balance built on published inputs. It is not validated against 1904–1911 gauge data. Open items are 3, 5, 9 and 12 (Section 5).

1. Model structure

The model steps daily through each water year, 1 October to 30 September. One cubic foot per second for one day equals 1.98347 acre-feet. Inputs and sources are in Section 2.

  • Upper Klamath Lake (UKL). Storage: V(t+1) = V(t) + I(t) − L(t). I is Reclamation's daily net inflow, already net of lake evaporation and precipitation, so no separate evaporation term. L is the Link River outflow over the reef, before any Klamath Canal Co. diversion. The diversion never reduces UKL storage, so the lake stage is the same with or without it. Stage comes from the storage table.
  • Link River outflow. L = 392.6 × (H − 4,138.5)^1.712 cfs, where H is UKL stage in feet. L = 0 at or below 4,138.5 ft. The Klamath Canal Co. diversion is taken from L: it is the lesser of the demand and L that day.
  • Lower Klamath Lake (LKL). Inflow = L minus the Klamath Canal Co. diversion. Losses: marsh evapotranspiration of 3.077 ft/yr on the wetted marsh and open-water evaporation of 3.50 ft/yr on open water. Precipitation of 0.885 ft/yr is credited to the wetted surface. Groundwater inflow = 0.
  • Wetted marsh fraction. 0.75 at 4,084.0 ft (no source; item 12), rising 0.125 per foot to 1.0 at 4,086.0 ft USBR. Open water 33,000 ac and marsh 52,120 ac (Section 2).
  • Keno outflow. Q = 181.66x² + 1,120.9x + 1,558.3 cfs, where x = H − 4,085.0 ft (2004 draft), at every stage. Zero at or below 4,082.9 ft. Capped at the water in the lake that day. The lake floor is 4,083.1 ft.
  • Lost River Slough. Q = 61.220 × (H − 4,085.0)^2.890 cfs above a 4,085.0 ft sill; zero below. Zero at every stage with the dike in place (default).

2. Inputs and sources

Values are used as written. Each row gives the value, its source and a note.

InputValue usedSourceNote
UKL storage–stage4,136.00–4,145.00 ft; 37 breakpoints at 0.25 ft (within 86 AF of the 0.01-ft table)Reclamation 2026 operations tracker, sheet UKL_Volume, column "UKL storage with Agency Lake and Barnes units of the UKNWR reconnected"Paired with the pre-dam outflow rating. Source to confirm; see assumption 9.
UKL net inflowDaily, WY1962–2026Net inflow as the BA PA model uses it, before the Agency–Barnes term. WY1981–2022: Reclamation operations-viewer daily series (UKLNI_raw; WY2022 from run IOP_extend_Nov2022). WY2025–2026: daily net-inflow column of Reclamation tracking workbooks (WY2026 total 807,379 AF). WY2023–2024: K.I.D. lake balance scaled to Reclamation annual totals (no daily column). WY1962–1980: USGS daily lake balance, unscaled.Already net of lake evaporation and precipitation.
WY2026 classificationWhole water year: 805,415 AF through 30 Sep (provisional), 8th lowest of 46 (WY1981–2026); period-of-record median 992,164 AF (WY2002). April–September: 4th driest of 65; 54% of record median.Whole year: Reclamation slide 16. April–September: K.I.D. WY2026 Dryness Evaluation §3.Menu labels use the April–September rank. Whole-year totals are on the 46-year record, not the 65-year daily record.
Link River ratingCoefficients in Section 1E. C. Koppen, pre-dam rating of outflow "in a state of nature" over the Link River reefChecked against Reclamation's 1915 paired stage and discharge: within 1.5% on the annual mean; 88–116% by month. K.I.D. deck gives a natural sill of 4,137.8 ft; the 4,138.5 ft zero-flow elevation is kept.
Keno ratingCoefficients in Section 1Reclamation, Undepleted Natural Flow Study, Part 2 (draft, 26 July 2004); coefficients transcribed from the rating figureReef crest 4,083.1 ft, outside the 1931 and 1965 cuts (K.I.D. deck). The 4,084.0 ft older record is not used. Not validated against 1904–1911 gauge data. See assumption 10.
Keno points cited, not fitted300 cfs at 4,084.0 ft2003 Natural Flow Study, as summarized in K.I.D.'s History of Keno Reef deck (28 August 2026), slide 7The quadratic gives 619 cfs there. The 730 cfs minimum in the 2025 final (§3.3.2.1) is not used: its source (Henshaw and Dean, 1915) is not in the files, and the USGS daily record shows lower 1904–1906 minimums (287, 230 and 535 cfs). See assumption 10.
UKL record high4,144.98 ft, about 20 April 1904, from high-water marks (USGS record, 1904–1950)USGS Water-Supply Paper 1315-B, Part 11-A, p. 826, station 368Treated as USBR datum (assumed); see item 5.
Klamath County precipitationW4 share by water year, 1896–2026; 1904 highest (27.9%); 1904–1908 highest five-year meanCounty SPI file (273eb216); BA review Figure 43 cites Drought.govPrecipitation, not inflow. Used in the introduction.
WY2026 snowpackPeak snow water equivalent 4.1 in (inches, not percent), 24 February 2026; season peak on the NRCS plot (18 sites)NRCS Klamath basin SWE plot: https://nwcc-apps.sc.egov.usda.gov/awdb/basin-plots/POR/WTEQ/assocHUC6/180102_Klamath.html?showYears=2026Featured button label. The live plot dated 9 October 2026 shows percent of median as N/A, so no percent-of-median figure is used.
1984 Keno peak3,979 cfs on the sequential record; eighth highest daily peak since 1962; follows wet WY1983Daily export, keno_cfs, WY1962–2026Model output. Highest peak is WY1965 (4,828 cfs).
1992 April–September inflow193,153 AF, driest of 65 years (Reclamation year-type record)K.I.D. WY2026 Dryness Evaluation §2Menu label. The model's daily series gives 140,676 AF for the same window (WY1992 total 592,932 AF). The 52,477 AF gap is not explained in the files.
Sequential UKL carryoverModeled UKL 1 October 2025: 4,139.84 ft; K.I.D. observed 4,139.28 ftModel run, sequential record; observed: K.I.D. WY2026 Dryness Evaluation §5About 0.6 ft above observed. See assumption 14.
Lost River SloughCoefficients in Section 1; sill 4,085.0 ft; 1,200 cfs at 4,087.8 ftUndepleted Natural Flow Study, Part 2; rating from HEC-RAS backwater on the early-1900s channelWet-year total is sensitive to the sill: 0.1 ft lower raises it 25%; 0.1 ft higher lowers it 21%. See assumption 13.
Dike closing the SloughIn place by default; removed in the sensitivityUndepleted Natural Flow Study (2004 draft: connection closed with a dike in 1890); Voorhees (1913); Abney (1964); 2023 Reclamation deck, slide 32; OAH Case 003 Corrected Partial Order of Determination, Finding 14Built 1889–1890 (sources differ). Predates the 1905 authorization, so it is in Pre-Project Development (K.I.D. position). Zero Slough outflow; no crest elevation in the files, so overflow is not modeled. See assumption 15.
Lower Klamath Lake geometryOpen water 33,000 ac; marsh 52,120 ac (85,120 ac in all); linear capacity 422,545 AF from 4,083.1 to 4,088.0 ftOpen water: 2025 final §2.2.2.1. Capacity and 2004 areas (34,994 open; 55,842 marsh): Undepleted Natural Flow Study, Part 2Linear relation, not a surveyed curve. The 85,120-acre footprint is as specified for this analysis.
Klamath Canal Co. demand (toggle)The monthly mean A Canal diversion for each calendar month, water years 1911–1920, March–October, less 50 cfs; cap 700 cfs (not reached; highest month 235 cfs, June)USGS Water-Supply Paper 1315-B, p. 825 (A Canal monthly diversions in acre-feet, converted at 1.98347 AF per cfs-day)Project-era record as a proxy for pre-Project demand. A judgment, not a measured pre-Project diversion. Taken from the flow over the Link River reef, not from UKL storage.
EvapotranspirationOpen water 3.50 ft/yr (decided 12 October 2026); marsh 3.077 ft/yrOpen water: 2004 Draft Natural Flow Study, about 3.5 ft gross (92,000 AF/yr net over about 35,000 ac, plus 0.885 ft precipitation). Marsh: USGS Scientific Investigations Report 2013-5014Open water inside the 3.25–3.60 AF/ac band. Marsh rate measured at Upper Klamath Lake and transferred; part of item 12.
Precipitation0.885 ft/yr; monthly shape: 63% October–March, 10% July–SeptemberNOAA GHCN-Daily, Tulelake, station USC00049053; mean of 79 complete yearsCredited to the wetted surface.
Groundwater to LKLAbout 72,400 AF/yrUndepleted Natural Flow Study, Part 2Not included (set to zero).
Rain gauges (display)Klamath Falls Airport USW00094236 (WY1999, WY2016); Klamath Falls 2 SSW USC00354506 (WY1992); Keno USC00354403NOAA GHCN-Daily (ncei.noaa.gov)Shown on the page; not used in the water balance.
Reference gaugesUSGS 11507001 (Upper Klamath Lake); 11507500 (Link River); 11509500 (Klamath River at Keno); 11507200 (A Canal)USGS National Water Information System (waterdata.usgs.gov)Used for checks. K.I.D.'s 2026 data audit compared the net inflow record with these gauges.

3. Steps to recreate a run

  1. Load the daily UKL net inflow (Section 2) and the UKL storage table. Interpolate stage from storage.
  2. Set starting conditions. Type years: UKL 4,139.46 ft and LKL 4,084.10 ft on 1 October. Sequential record: carry both lakes from the previous 30 September (assumption 14).
  3. Each day: UKL stage, then Link River outflow L (Section 1), then update UKL storage with V(t+1) = V(t) + I(t) − L(t). The Klamath Canal Co. diversion does not enter this step.
  4. Route L to LKL, less the Klamath Canal Co. diversion taken from L. Compute the wetted marsh fraction, marsh and open-water losses and the precipitation credit. Update LKL storage and get LKL stage from the LKL storage relation.
  5. Compute Keno outflow and Slough overflow from LKL stage. Remove both from LKL storage.
  6. Record each day. Sample every 5/3 days to build the animation frames (217 per year, day 0 to day 360).

Steps 4 and 5 follow Part 2 of the Natural Flow Study. Confirm the order of operations and same-day routing against the source script.

4. Checks

  • Mass balance: for each lake, storage change equals inflow minus outflows and losses, every day and every water year.
  • Rating reproductions: Keno gives 619 cfs at 4,084.0 ft and 1,558 cfs at 4,085.0 ft. The Slough gives 1,200 cfs at 4,087.8 ft. Link River gives about 786 cfs at 4,140.0 ft.
  • Link River: compare with Reclamation's 1915 paired records; expect agreement within 1.5% on the annual mean.
  • Gauges: compare the UKL and Link River balances with USGS 11507001 and 11507500 over overlapping years.
  • Historical matrix: 189,895 AF on the BiOp method (storage change plus measured outflow, Ady canals excluded; 219,731 AF with them), 1.7% below Reclamation.

5. Open items

#Issue
3Dike dates and Slough history. Built 1889–1890: Voorhees (1913) gives 1889; Abney (1964) and the 2023 Reclamation deck (slide 32) give 1890; the K.I.D. statement says late 1889 and early 1890. The 2004 draft's average Slough outflow of 6,000 AF/yr is a simulation (p. 46), not a measurement; this run's sequential-record average with the dike removed is 10,763 AF/yr. The 1888 flood predates the dike, so the draft's simulated 114 cfs peak is not a like-for-like test of its 800–1,000 cfs estimate for that flood (p. 26). The sources do not support "significant periods" of overflow, missing flood records, or 25-ft banks.
5Datum. Elevations are on the USBR datum. The 2025 study gives NAVD88 offsets of +2.01 ft (upper lake), +1.86 ft (Lower Klamath Lake) and +1.89 ft (Tule Lake), and 1.78 ft between Reclamation and USGS datums. Four 2023-draft points, converted from NAVD88, match the page or memo within 0.06 ft. The 1950s USGS compilation gives two upper-lake gage datums, 4,100.00 ft and 4,098.22 ft, in garbled OCR text. The 2004 draft labels its 4,088 ft capacity elevation "USRS datum" (item 12), which conflicts with the 2025 figure unless the datums are settled.
9Year totals and readout. The inflow record is Reclamation's daily series for WY1981–2022 and WY2025–2026, the USGS balance for WY1962–1980, and the K.I.D. balance scaled to Reclamation totals for WY2023–2024. The WY1992 total of 592,932 AF matches the 2024 BiOp (p. 61). The WY1999 total of 1,767,443 AF is 7,729 AF above the BiOp's 1,759,714 AF. The readout stops at day 360 and runs 0.1–0.7% below full-year totals, checked for every dataset.
12Lower Klamath Lake marsh base and wetland evapotranspiration. The wetland rule applied is marsh 100% wetted at 4,086.0 ft USBR. Two inputs remain open. (a) Marsh base: the 75% wetted fraction at 4,084.0 ft has no source in the files. The 2025 final (§2.2.2.1) puts the onset of marsh inundation at 4,088 ft NAVD88, or 4,086.14 ft USBR (NAVD88 less 1.86 ft), with full wetting about one foot higher. If the marsh stays dry below that onset, dry-year Keno rises 9.5% and median-year Keno rises 8.1% (dike in place). With the dike removed, the median-year Slough total falls from 1,481 to 1,407 AF. An earlier run gave a rise from 331 to 563 AF; on the rebased record the sign reverses, and the cause is not yet traced. (b) Wetland ET: the 3.077 ft/yr rate was measured at Upper Klamath Lake (USGS SIR 2013-5014). The Lower Klamath Lake rate needs the ET component memo, which is not in the files. The 2025 final's total of about 78,000 acres one foot above onset differs from the 85,120-acre footprint; the footprint is kept.

6. Decisions and assumptions

Numbers are kept so existing cross-references still work.

  1. Evapotranspiration. Open-water rate 3.50 ft/yr, inside the 3.25–3.60 AF/ac band (decided 12 October 2026). The marsh rate is covered in item 12.
  2. No Upper Klamath Lake evaporation term. Net inflow is already net, so the model cannot draw UKL below the Link River zero-flow elevation once outflow stops. A natural lake would keep evaporating. No action needed.
  3. Keno reef crest 4,083.1 ft (decided). The 1906 survey placed the reef between 4,083.1 and 4,084.0 ft. Copco cut a channel to a 4,080.3 ft invert in 1931, and the reef was lowered again in 1965. The 4,083.1 ft crest is outside those cuts (K.I.D. deck, slides 2, 6, 13 and 19). A lower reef gives more natural Keno flow, so 4,083.1 ft understates the Project's contribution. Kept unless a newer survey is found.
  4. Upper Klamath Lake storage table (decided). The BiOp-based table reproduces the 2024 BiOp active storage: 645,633 AF through 4,143.3 ft, against 645,627 AF in the BiOp. It runs 1–12% above the 2025 final's tables over 4,139–4,142 ft. The Koppen table differs by −41% to +10% across the range. The table behind the BiOp total is not in the files; Reclamation should supply it.
  5. Keno rating: 2004 quadratic at every stage (applied 12 October 2026). It replaces the 2003 power law. Where the quadratic exceeds the water in the lake (at the 4,083.1 ft floor it gives about 84 cfs), outflow is capped at that day's water, so the model creates no water. Coefficients are transcribed from the draft's rating figure and should be confirmed against the draft text. The quadratic gives 619 cfs at 4,084.0 ft, where the 2003 study gives 300 cfs. The 1906 low-water test (Jacobs: "over 4,084 even at low water") favors the power law, which gives 4,084.00 ft at 300 cfs; the quadratic gives 4,083.52 ft. The quadratic gives about 6,100 cfs at 4,087.8 ft. Not used as check points: 730 cfs (1904–1906 minimum; source not in the files) and 8,400 cfs (June 1904; the USGS daily maximum is 2,696 cfs). The curve is not validated against 1904–1911 gauge data, and early flows were measured without a weir.
  6. Lost River Slough formula, as modeled (Section 1). The 2004 draft's rating gives about one-ninth of this flow at the same stage and is not used. With the dike removed and the current Keno relation, the draft's rating would cut the sequential-record average from 10,763 to about 1,382 AF/yr. The 2023 draft (slide 49) shows Slough flow also depends on Link River discharge, which this formula does not capture.
  7. Lower Klamath Lake open water (applied). 33,000 ac, the area before the marsh floods (2025 final §2.2.2.1). Marsh 52,120 ac; footprint 85,120 ac (Section 2).
  8. Slough sill at 4,085.0 ft (assumed resolved; formerly open item 4). The 2003, 2004 and 2005 natural flow studies and the 2025 final use 4,085.0 ft as the overflow threshold. No survey of the Slough entrance is in the files. With the dike removed and the sill moved ±0.1 ft: median-year Slough total 857–2,378 AF (base 1,481); wet-year 19,244–30,493 AF (base 24,438); sequential-record average 8,710–13,192 AF/yr (base 10,763). Keno changes by less than 0.5%.
  9. 1 October storage (sequential record; counsel's decision). The sequential record carries both lakes over each 1 October. The handoff record run instead resets UKL to 4,139.46 ft on each 1 October, 0.18 ft above observed. Carrying over puts 1 October 2025 at 4,139.84 ft, about 0.6 ft above observed, and gives a WY2026 Keno total of 671,142 AF with the dike in place; the reset gives 637,280 AF. Which method counsel adopts is open.
  10. Dike across the Slough (in place by default). It predates the 15 May 1905 authorization (OAH Case 003, Finding 14), so Pre-Project Development includes it; switching it off restores the natural Slough. The model gives zero Slough outflow with the dike in place. No crest elevation is in the files, so flow over the dike is not modeled.

7. Files and sources

  • Model script: without_project_effects_model.py (its header lists the source of each parameter).
  • Stage export: the 23 August 2026 export (WY1961–2025) is superseded by the current inflow record; a WY1962–2026 export is pending.
  • Source studies: Reclamation, Undepleted Natural Flow Study, draft of 26 July 2004 (Part 2); Natural Flow Study, Part 2 (2003), as quoted in the 2024 BA review and the K.I.D. deck; Reclamation KBAO, History of Keno Reef, 21 November 2019; USGS Scientific Investigations Report 2013-5014; NOAA NCEI, Global Historical Climatology Network–Daily.
  • Reclamation data: operations viewer (UKLNI_raw, WY1981–2022; WY2022 from run IOP_extend_Nov2022); 2026 operations tracker (UKL_Volume sheet and daily net inflow); tracking workbooks for WY2025–2026; K.I.D. lake balance scaled to Reclamation annual totals for WY2023–2024. Pre-1981 net inflow: USGS daily gages 11507001, 11507500 and 11507200, balance method (K.I.D. derivation, 3 October 2026).
  • W4 classification: Klamath County monthly Standardized Precipitation Index file (1895–2026). A water year is W4 when its mean share of the county in W4 is 25% or more. Only water year 1904 qualifies in 1896–2026.
  • Context studies: Meko, Woodhouse and Touchan (2014), Klamath tree-ring reconstructions, California DWR; Malevich, Woodhouse and Meko (2013), corrigendum to J. Hydrol. 495 (corrected Klamath Falls reconstruction).