TidePush answers the practical question behind a tide table: when should I leave? This page explains what it actually computes, where its limits are, and what it does not do.
A current prediction tells you what the water will be doing at a particular place and time. Planning a passage means working out when your boat will actually reach each current-sensitive part of the route, what the stream will be doing when you get there, whether the wind will oppose it, whether a narrows will still be passable, and whether you will arrive in daylight.
TidePush integrates those pieces across the whole passage and recommends a departure window, rather than leaving you to work each one out separately.
A passage is not safe or efficient merely because the boat can hold positive speed over the ground. What matters is the interaction of current speed, current direction, wind, boat speed, timing, draft, sea state and local geography. Turning those interacting variables into one practical recommendation is the whole of what this app does.
TidePush breaks a route into legs and gives each leg the NOAA or CHS current station that actually governs that stretch of water — not the nearest one on a map. It then advances the boat through the route a minute at a time, evaluating the current at the position and time the boat genuinely reaches, rather than the current running when it left the dock.
Each leg also carries a tidal phase offset relative to its station and a rate at which that phase propagates along it. That is what lets a boat ride the turning stream down Admiralty Inlet, meeting water that turns as it goes, instead of one fixed current everywhere.
Coverage today: 83 curated passages across 61 current stations.
STW — speed through the water is the vessel's speed relative to the water it floats in. It is the cruising speed you enter, and the number you actually know about your boat.
SOG — speed over ground is what is left after the current has had its say, and it is what decides when you arrive. The difference between the two is a large part of what TidePush is modelling.
A gate is a part of a passage where timing stops being about convenience and becomes an operational constraint. TidePush distinguishes several kinds, because they fail for different reasons and only one of them is fixed by waiting for slack.
The question is whether the boat keeps enough propulsion margin against the stream. A six-knot boat meeting four knots on the nose is making two over the ground with nothing in reserve for a bad moment. Positive SOG alone does not make a transit prudent. Limits are set per passage and scale with cruising speed, so a slower boat is automatically held to a stricter limit.
Dominated by wind against current rather than current magnitude alone. Wind blowing against the direction the water is flowing stands the sea up short and steep, even when either the wind or the current on its own would be unremarkable.
TidePush evaluates wind direction relative to the direction the CURRENT is flowing — not relative to the boat's heading. That distinction is the common and dangerous confusion: a fair four knots through a narrows with a gale against it is the bad case, and the one most likely to be mistaken for a good slot because the current is helping.
Some passes are a reaction-time problem rather than a propulsion problem. At six knots of stream in a hundred-yard notch, an engine cough, a deadhead, a traffic conflict or a navigation error leaves very little room to recover.
Some passages are limited by how much water there is, not how fast it moves. A current window cannot solve a draft problem; only enough tidal height can. The Swinomish Channel is the example — see below.
Ferry tracks, VTS lanes, narrow dredged channels, buoyage that reverses partway through, and channel edges that are riprap rather than mud. These are not computed as mathematical gates. They appear as per-leg hazard notes, because they are judgement rather than arithmetic, and representing them as precise algorithmic limits would be a lie.
Every gate names two limits, and they differ on purpose.
Against the stream, the limiting factor is propulsion margin. With the stream, the limiting factor becomes control: four knots behind a five-knot boat is nine over the ground in a two-hundred-foot channel — the least steerage and the most speed of any boat in the fleet. A single symmetric number would be wrong in one direction whichever way it was set.
Figures below are for a boat cruising at seven knots; limits expressed as a fraction of cruising speed scale down for a slower one.
The strictest gate in the library: a few hundred feet wide under the bridge, past eight knots at springs, with standing waves and whirlpools below it. Very high stream velocity, narrow geometry and sharply reduced reaction time. It floods east into Skagit Bay and ebbs west into Rosario. Refused above 2.80 kt with you and 2.45 kt against.
The western approach matters too: a southerly against an ebb builds a severe rip outside the pass — water you meet before you get to decide about the pass itself.
Treated as its own gate rather than part of Rosario Strait, because it has its own tidal timing. The ebb runs to about three knots and is at its worst against a southerly, which is the prevailing summer afternoon wind.
The practical strategy is usually to enter on or shortly after the turn in your favour and let the building stream carry you out, rather than trying to overpower a large adverse stream. Busy with tows that cannot give way.
Different geometry from Deception or Dodd: wide enough that the issue is stream velocity against vessel speed rather than extreme confinement, so both its limits are equal.
On a short trip the timing benefit may not be time saved at all. The objective is simply to reach the passage while it is open, and the answer says so in those words rather than quoting a saving that does not exist.
Draft is the primary constraint; current still affects ETA. TidePush compares your draft plus a 25% margin against the predicted water level at the gauge inside the cut, and reports whether the tide is rising or falling during the transit — four feet under the keel and rising is an ordinary day; four feet and falling, with no room to turn, is not.
A maintained project depth is not a guaranteed observed depth. The cut is dredged to twelve feet at MLLW and shoals between dredgings, which is exactly why the check carries a margin.
The clearest example of why current magnitude and sea-state risk must be separate ideas. Point Wilson is genuinely dangerous — a westerly over an ebb stands up short and steep right where you turn the corner — but it is not a narrows, and refusing it on current magnitude would block days that are perfectly fine. So it carries no current cap, and the risk is caught where it actually lives: in the wind-against-current check.
Different in kind from a narrows. The problem is extended exposure rather than a current bottleneck — a crossing puts a boat in twenty-four to forty-plus miles of open water with no pinch point anywhere in it, where the current is a knot or so and almost never the thing that hurts you. What hurts you is what the wind has built.
The gate therefore reads significant wave height, wave period, the angle the sea makes with the course, wind direction, the interaction with current, and the vessel's rig.
Period matters as much as height. A short-period sea of a given height is much harder work than a long-period sea of the same height: 1.2 m at 4 s is a slamming chop, 1.2 m at 10 s is a swell you barely notice. Reading height alone refuses the comfortable day and passes the miserable one.
The Strait of Juan de Fuca carries the same kind of gate, tuned tighter. Georgia is an enclosed basin whose sea was built there and dies when the wind does. Juan de Fuca is open to the Pacific at its western end, so ocean swell reaches most of the way to Port Angeles whether or not there is local wind to explain it, and the whole strait is a tidal funnel.
This is about different vessel dynamics, not about toughness.
Under sail: cruising speed remains a hard limit — favourable wind relieves the engine and pays the diesel, never the clock. Sail pressure damps roll, and a boat heeled and driving is steadier than the same hull lying to a swell. A beam reach is the best point of sail there is, so beam-on is where a sailing boat wants the wind. Dead downwind is rollier than it looks, and dead to windward is the worst angle of all, because you cannot sail there — you are motoring into a head sea with the rig doing nothing for you.
Under power: wind is largely a speed and comfort penalty. A beam sea is the worst roll, a head sea pounds, and a following sea is easiest. Waves begin cutting effective speed as conditions build.
| Sea from | Powerboat | Sailboat |
|---|---|---|
| Astern | Easiest | Rolly, sails slatting |
| Quarter | Easy | Broad reach — comfortable |
| Beam | Worst roll | Beam reach — favourable |
| Ahead | Pounding | Worst: motoring into it |
One condition overrides all of it. When the wind is against the stream, or running across a sea it did not make, a sailing rig gets no allowance at all: that water is short, steep and out of phase with the wind driving you, and sail pressure does not damp it.
TidePush evaluates candidate departures across the day and prefers one that is close to the day's fastest passage, avoids unacceptable gate conditions, preserves daylight margin, and stays acceptable across the uncertainty range rather than only at its best case.
Where two departures produce substantially the same passage, it recommends the later one. There is no reason to send a skipper off at dawn for a passage that goes just as well at nine.
The recommended time is only the centre of the answer; the box around it matters more. The window holds the departures that satisfy substantially the same criteria as the recommendation, and the recommendation always sits inside it.
Leave anywhere inside the window and the passage still works on the same terms. Tidal predictions, weather forecasts and vessel performance do not justify minute-perfect precision, and a single time pretends otherwise.
Rather than returning no answer, TidePush relaxes constraints in a deliberate order: first it gives up near-optimal passage time, then it tolerates some increased foul-pass risk, then it reduces the daylight buffer. A recommendation that required relaxation is marked visibly so you know the day was a compromise.
Some things are never relaxed. A crossing judged too rough does not become recommendable because nothing else that day worked, and a passage whose expected arrival is after dark is not offered simply because nothing better exists.
TidePush propagates uncertainty in vessel speed, tidal timing, current strength, wind timing and wind strength, and reports the result as a band rather than one exact arrival.
A single arrival time would imply more certainty than the data deserves.
| Horizon | What is used |
|---|---|
| Days 0–3 | Full wind and wave forecasts. |
| Days 4–7 | Weather forecasts, with widened uncertainty. |
| Beyond | Tides only. Timing is still meaningful; weather is not — recheck closer to departure. |
TidePush compares the best and worst departures on the same day. "Timing worth" is what leaving at a poor part of the tidal cycle costs you against a favourable one — which is a more useful number than comparing your passage with an imaginary still-water one you were never going to get.
Written down deliberately. A planner that implies it has checked something it has not is more dangerous than one that names its edges. None of the following is currently computed.
Related: About TidePush · Version history · llms.txt