By the numbers
Every passage re-planned on 365 days of 2027, to answer what the tide is actually worth.
Common questions
How much time does timing the tide actually save?
It depends entirely on the passage, and the range is enormous. Re-planning every route across 365 days of 2027, the difference between the best and worst cast-off of the same day runs from under a minute on the short Desolation Sound hops to 130 min on Campbell River → Chatham Point (Seymour Narrows). That figure comes with a catch worth reading in the same breath: 52 per cent of the cast-offs on that passage are refused outright, so it is not a saving lying there for the taking. The largest gain on a passage that refuses almost nothing is 68 min on Anacortes → Shilshole (Rosario).
Which passages can only be made at certain times of day?
The tidal rapids. On Octopus Islands → Big Bay (Hole in the Wall & Yucultas), 96 per cent of all possible cast-offs are refused outright — the water is past what the pass allows, and no boat speed changes that. There timing does not save you hours, it decides whether the passage happens at all. On an open crossing like Anacortes → Shilshole (Rosario) — which refuses 0 per cent — timing buys you 68 min instead.
Are there passages you cannot make in winter?
Yes, and it is daylight rather than current that closes them. For a boat making 6.5 knots, 23 of the 92 passages have at least one month with no plannable day, and the longest — Bainbridge → Echo Bay, Sucia (Rosario) at 75.8 nm — is shut for 4 months. A passage of seventy-odd miles does not fit between a December sunrise and sunset.
Does a faster boat change which passages close?
Completely, and this is the clearest thing the whole analysis shows. The same year was re-planned at 5 knots sail, 6.5 knots sail, 7 knots sail, 9 knots motor. At 5 knots sail, 37 of the 92 passages close for a month or more. At 9 knots motor, 0 do — a motor boat at that speed can make every passage in the library in every month of the year. Closing dates are daylight arithmetic, and speed buys daylight.
Can a faster boat get through the rapids more often?
No, and the numbers are flat enough to be worth stating plainly. Octopus Islands → Big Bay (Hole in the Wall & Yucultas) refuses 96 per cent of cast-offs at 5 knots sail — and 96 per cent at 9 knots motor. Across a speed range that empties the winter closure table entirely, the rapid does not move at all. Engine power is an answer to daylight and to an adverse stream in open water. It is not an answer to overfalls and whirlpools, and the gate policy gives it no credit there.
How often do two tidal gates on one passage conflict?
Often enough to matter. On Coal Harbour → Friday Harbor, the windows fail to line up on 45 per cent of days — each gate has its own slack at its own station, and the two drift apart by about fifty minutes a day. On those days no single departure serves both and something has to give.
What speed changes, and what it does not
The whole year was re-planned four times, once per boat. Two columns move together and one does not move at all.
| Boat | Passages that close | Passages with refusals | Octopus Is. → Big Bay refuses |
|---|---|---|---|
| 5 knots sail | 37 of 92 | 33 | 96% |
| 6.5 knots sail | 23 of 92 | 33 | 96% |
| 7 knots sail | 16 of 92 | 33 | 96% |
| 9 knots motor | 0 of 92 | 33 | 96% |
Every figure over all 365 days of 2027.
Closures collapse from 37 to 0. That column is daylight and nothing else: a slow boat needs more hours than a winter day contains, and a fast one does not.
The same 33 passages have refusals at every speed, and the rapid does not budge. Speed changes which DAYS work. It does not change which passages have a gate in them, or how often that gate is shut.
Why there is no "time saved by speed" column here
It would look like the obvious fourth column and it would be misleading, so it is left out on purpose. The saving is measured over the cast-offs available before dark — and that window is a different size for every boat. A 5-knot boat takes about eleven hours on a long passage against roughly six for a 9-knot one, so in summer the slow boat has perhaps five usable hours of departures where the fast boat has nine.
Two effects then pull against each other. A slow boat is more sensitive to the tide, because two knots of stream is a larger share of five than of nine — that widens the spread. But its departures are squeezed into a narrower window and so meet more similar water — that narrows it. The two cross over somewhere in the middle, and the resulting numbers rise and then fall rather than trending. They are four measurements of four different things, and putting them in a column would invite reading a trend that is not there.
What was measured
Every distinct passage in the library was re-planned on every one of the 365 days of 2027 — for a sailing boat making 6.5 knots through the water. That is the app's own planner producing its own answers, not a model of it.
What one measurement actually is
A passage here means one simulated departure, integrated forward along the route a minute at a time against the changing stream. Departures are not tried every minute, though: the planner lays out candidate cast-offs on a 15-minute grid from first light, and keeps them as far as the last one that still gets in before dark.
Each of those cast-offs is run 11 times, not once. Five things are uncertain — boat speed, tidal phase, current strength, when the sea breeze fills, and how hard it blows — and a sigma-point transform propagates all five together. The duration used below is the middle of those 11 runs, which is why every time on this site comes with a band rather than a decimal place.
The saving is then one subtraction. Take every cast-off that day the planner would actually let you use, and subtract the fastest passage from the slowest. That is the day's number. The figure in the table is the median of those daily numbers across the year — median rather than average, so a single freak spring tide cannot set the headline.
The scale, counted rather than estimated. Planning the library for one year at one speed laid out 1,145,069 candidate cast-offs; across all 4 boats it was 4,668,043. Each was run 11 times, which is 12.6 million integrated passages — and since the average passage here is about 331 one-minute steps, roughly 4.2 billion steps of the same arithmetic a phone does when you plan a single passage. It ran on one laptop core in about three quarters of an hour.
What the grid means. A true best departure can fall between two 15-minute slots, so the real saving is a little larger than shown — but the arrival estimate at a gate thirty miles ahead is not good to five minutes, and a finer grid would be false precision rather than a better answer.
Every day, not a sample. This began at one day in four, which was defensible — the spring–neap cycle is 14.77 days, so a four-day stride precesses through springs and neaps rather than landing on the same phase — but it meant a month with eight sampled days could be called shut when unsampled days would have worked. Running all 365 days removed the argument. For the record it barely moved the numbers: the median passage changed by 0.4 minutes and the set of passages that close was identical, which is the sample vindicating itself on its way out.
One year, and it matters slightly. These are 2027 tides. The moon's orbit drifts on an 18.6-year cycle that moves current strengths by a few per cent, so the same analysis for a different year would shift these minutes a little. The shape — which passages care about timing and which do not — is not sensitive to that.
Wind is deliberately absent. No forecast reaches a year ahead, so every figure here is the app's tides-only mode: what the water alone does to a passage. That is what makes months comparable, and it means a real summer afternoon in the Strait of Juan de Fuca will be slower than these numbers suggest. One boat, one year, no weather — the shape is right and the minutes are not promises.
Timing is worth two completely different things
The obvious measure is hours: the gap between the best and worst cast-off of the same day. It tells the whole story on an open passage and almost none of it at a rapid, where the planner has already refused every departure but the few near slack, so the survivors all take about the same time. Octopus Islands → Big Bay (Hole in the Wall & Yucultas) measures 1 min of spread and refuses 96 per cent of its cast-offs. Timing there is not saving you time. It is the difference between going and not going.
Where the right hour saves the most time
| Passage | Distance | Best vs worst cast-off |
|---|---|---|
| Campbell River → Chatham Point (Seymour Narrows) | 20.4 nm | 130 min |
| Refuge Cove → Big Bay (the Yucultas) | 20.8 nm | 79 min |
| Anacortes → Shilshole (Rosario) | 58.4 nm | 68 min |
| Friday Harbor → Shilshole * | 59.4 nm | 67 min |
| Everett → Friday Harbor (Admiralty) * | 58.0 nm | 66 min |
| Refuge Cove → Dent Island Lodge ** | 24.3 nm | 66 min |
| Gig Harbor → Olympia (the Narrows) | 28.1 nm | 65 min |
| Bainbridge → Friday Harbor * | 65.1 nm | 64 min |
| Everett → Port Townsend | 34.2 nm | 59 min |
| Pirates Cove → Nanaimo (2nd stage) | 11.0 nm | 56 min |
| Sucia Island → Shilshole * | 72.6 nm | 56 min |
| Edmonds → Echo Bay, Sucia (Rosario) * | 66.0 nm | 54 min |
Median across the sampled days, comparing only cast-offs the planner would actually allow.
Where timing decides whether you go at all
| Passage | Distance | Cast-offs refused |
|---|---|---|
| Octopus Islands → Big Bay (Hole in the Wall & Yucultas) * | 8.3 nm | 96% |
| Porpoise Bay → Egmont (Skookumchuck) | 17.8 nm | 81% |
| Prideaux Haven → Dent Island Lodge ** | 31.2 nm | 79% |
| Refuge Cove → Dent Island Lodge ** | 24.3 nm | 77% |
| Refuge Cove → Octopus Islands (Surge Narrows) | 20.1 nm | 77% |
| Egmont → Chatterbox Falls | 33.7 nm | 77% |
| Refuge Cove → Big Bay (the Yucultas) | 20.8 nm | 71% |
| Campbell River → Chatham Point (Seymour Narrows) | 20.4 nm | 52% |
| Deer Harbor → Nanaimo (via Bedwell) | 54.3 nm | 40% |
| Friday Harbor → Nanaimo (via Bedwell) | 56.1 nm | 36% |
| Reid Harbor → Nanaimo (via Bedwell) | 46.3 nm | 36% |
| Sidney BC → Nanaimo | 43.1 nm | 35% |
Share of departures refused outright for current, depth or sea state, counted only over cast-offs that get in before dark — the planner draws a couple of hours past that to show the door closing, and counting those made long passages look fussier than they are. 33 of 92 passages refuse anything at all; the other 59 can be taken at any hour the light allows.
The passages that close
23 of 92 passages have at least one month in which no departure works. The cause is daylight, not current: the tides run on lunar cycles and do not care what month it is, but a seventy-mile passage at 6.5 knots does not fit inside a December day.
| Passage | Distance | Closed |
|---|---|---|
| Bainbridge → Echo Bay, Sucia (Rosario) * | 75.8 nm | Jan, Feb, Nov, Dec |
| Bainbridge → Sidney BC | 79.0 nm | Jan, Feb, Nov, Dec |
| Sidney BC → Shilshole (Haro) | 75.9 nm | Jan, Feb, Nov, Dec |
| Port Townsend → Neah Bay *** | 79.7 nm | Jan, Feb, Nov, Dec |
| Reid Harbor → Shilshole * | 68.9 nm | Jan, Nov, Dec |
| Sucia Island → Shilshole * | 72.6 nm | Jan, Nov, Dec |
| Edmonds → Echo Bay, Sucia (Rosario) * | 66.0 nm | Jan, Nov, Dec |
| Bainbridge → Roche Harbor (Haro) | 72.5 nm | Jan, Nov, Dec |
| Edmonds → Sidney BC | 69.2 nm | Jan, Nov, Dec |
| Bainbridge → Victoria | 69.3 nm | Jan, Nov, Dec |
How often the gates disagree
A passage with more than one gate has more than one slack to hit, and they drift apart by about fifty minutes a day. These are the passages where that becomes a problem, and how often.
| Passage | Gates | Days the windows conflict |
|---|---|---|
| Coal Harbour → Friday Harbor * | 2 | 45% |
| Coal Harbour → Sidney BC * | 2 | 40% |
| Coal Harbour → Roche Harbor * | 2 | 39% |
| Refuge Cove → Dent Island Lodge ** | 3 | 1% |
| Octopus Islands → Big Bay (Hole in the Wall & Yucultas) * | 2 | 0% |
| Prideaux Haven → Dent Island Lodge ** | 3 | 0% |
The three-rapid runs above Desolation Sound never conflict: the Yucultas and Gillard share a current station, so their slacks are minutes apart rather than hours. Conflict is a property of gates far enough apart to have independent tides.
Plan a passage — back to TidePush →