Okanagan Basin Water Board

How low can Okanagan Lake go?

The 2026 freshet was the weakest in years of record. This is a conditional stress test of what a second dry year would do — not a forecast.

Where the lake stands

Elevations are metres CGVD28, the datum Okanagan Lake is operated against. Full pool is m; the normal operating band is m deep.

The premise

2026 delivered the weakest freshet on record

Okanagan Lake fills almost entirely in May and June — those two months alone explain of the variance in annual net inflow, across the years with a complete annual total. In 2026 that freshet came in , and the lake never reached full pool.

WY2026 WY2017 (flood year) WY2003 (driest year) 10th–90th percentile, 1944–2025
Weekly net inflow, by week of the water year (Oct 1 → Sep 30). Net inflow = change in lake storage + release at Penticton Dam.
Lowest May–June net inflow, 1944–2026

Why this is not a forecast

A dry year does not predict the next one

Across consecutive water years (1944–2025, year-to-year pairs), the correlation between one year's net inflow and the next is . No year-ahead skill is detectable in this record — but detectable is not the same as absent: an 81-pair record has only 78% power to find a correlation of 0.30, and the 95% interval on this estimate reaches +0.34, above the 0.30 ceiling the model itself treats as material. Short records systematically under-detect multi-year clustering, so we do not claim successive years are independent. We assert conditionality: this tool asks what a second dry year does given it occurs, and is agnostic about how likely that is. If anything, undetected persistence makes a dry-follows-dry sequence more plausible, not less — which is the case this stress test exists to examine. There is no year-to-year persistence in Okanagan Lake supply, so at a twelve-month lead there is no forecast skill to have.

That is why this tool asks a conditional question instead. You state next year's supply; it states the consequence. A second dry year is not a prediction — it is a scenario worth being ready for, and 2026 has already shown the range is wider than the record suggested.

Consequence — shoreline

What drawdown actually exposes

Measured from the 2021 topobathymetric lidar across two mapped reaches. Exposure accelerates as the lake falls: the beach flattens with depth, so each additional centimetre dewaters more ground than the one before.

Kelowna – Bear Creek Peachland
Area exposed relative to full pool, for the two mapped reaches only — not the whole lake.

Consequence — fish provisional

What the winter drawdown rule costs

Okanagan Lake operations carry a shore-spawning kokanee constraint: drawdown from 15 October to April should stay under 0.2 m. Kokanee spawn in October, so an egg laid at depth d below the October surface dewaters once the lake falls further than d. Intersecting 1,359 mapped spawner segments with the lidar puts a number on that rule.

Kelowna – Bear Creek Peachland
Share of shore-spawning habitat dewatered, referenced to a median 15 October stage, assuming a 3 m spawning depth band.

Read this as provisional. The depth at which kokanee actually deposit eggs in Okanagan Lake is not yet sourced, and it moves the answer a long way — a 1 m band gives 15–17%, a 5 m band about 5%. The comparisons between scenarios are far more robust than the levels. This section will be revised once that parameter is cited.

Two things survive that caveat, because both are ratios rather than levels: the rule is a tolerance rather than a zero-harm threshold, and it is least protective in a dry October — a lower autumn stage puts the eggs on a flatter part of the beach, so the same drawdown dewaters more of them.