Marine weather routing · Rust isochrone engine

Every passage hasone perfectdeparture.route.

OpenSail tests half a million candidate routes against real bathymetry, real forecast gusts, and your crew’s limits — then tells you which line to sail, when to leave, and exactly why. Built by sailors who anchor a kilometer off the beach.

Plan a passage — free Watch the engine thinkNo card. Real routing on the free tier.

ECMWF IFS global open data

NOAA GFS 0.25° global

HRRR 3 km US coastal

Copernicus Marine currents + waves

GEBCO 15″ coastal bathymetry

The engine

Watch it think.

This isn’t a line drawn on a weather map. It’s a native-Rust isochrone solver expanding time-layered reachability fans — every candidate tested against wind, waves, current, depth, and land. Scroll to run a real solve: Tampa Bay → Nassau.

SOLVER · isochrone_v2phase: EXPAND FRONTIERt+00:00 · budget 45s

states pruned

depth rejections

land rejections

—

passage time

STEP 01

The fan opens.

Consistently along the route — and at the critical intersections — the solver expands the reachable frontier: thousands of candidate positions propagated through your boat’s polar, the forecast wind field, waves, and ocean current. No garbage collector in the hot loop. No hand-waving.

STEP 02

The sea says no. We listen.

GEBCO 15-arc-second bathymetry (~460 m cells) enforces your draft to roughly a mile from shore. Every leg is checked against high-resolution land polygons — the actual segment geometry, not grid cells. Real forecast gusts gate against your crew’s wind ceiling, not a guessed 1.25× multiplier.

STEP 03

One route survives.

24.8 hours, beam reach, max gust 22 kt. The faster-looking rhumb line dies in the shallows off the bank — and the app tells you that in plain language. When no safe route exists at all, you get an enumerable reason, never a spinner that gives up silently.

The intelligence layer

It picks the route —
and the moment.

The numbers pick a route. The briefing tells you why. A solved passage already contains the decision — which route, which departure — buried in a table of durations, comfort scores, and gusts. The intelligence layer reads those numbers and hands you the call in plain language: one recommendation, the trade-offs that justify it, and an honest rating of how much daylight there is between the options.

Compare all modes

Four routes solved. One recommended. Reasons attached.

Compare All Modes runs fastest, balanced, family, and comfort+ as four complete solves against the same forecast. Then the intelligence layer reads all four and commits: take family — 2.6 extra hours buy 2.3 comfort points and halve the heel. Not a table you decode at the dock. A decision with its reasoning shown.

Every claim traces to a solve

The briefing is generated from your four solved routes — durations, comfort, heel, gusts, notable events. It never invents weather it wasn't handed.

It talks you out of overpaying

When a gentler mode buys almost nothing — 48 minutes for 0.2 comfort points — the briefing says so, instead of upselling you on calm you already have.

Deep dive: Compare All Modes →

Ft. Lauderdale → Freeport4 modes · 1 recommendation

Which Route?

Take Family — 2.6 extra hours buy 2.3 comfort points and halve the max heel on the Stream crossing.

All four modes sail Ft. Lauderdale → Freeport against the same forecast: an 18–22 kt northeaster easing overnight as the front clears the Stream. Fastest and balanced cross the axis at peak gradient and pay for it in heel; family and comfort_plus delay the crossing about three hours and meet 1.1 m seas instead of 1.9 m. Comfort_plus offers almost nothing beyond family here — 48 minutes for 0.2 points — so the honest pick is family: the calmest crossing of the set at a cost the crew will never notice.

By Mode

Fastest24.9 h and heel to 21° — quickest across, but you wear every knot of the northeaster on the Stream axis.
Balanced25.3 h, comfort 6.8 — shaves the worst heel for 24 extra minutes, still crosses the axis at peak gradient.
Family27.5 h, comfort 8.4, heel to 10° — waits out the front and crosses behind it. The pick.
Comfort+28.3 h for comfort 8.6 — 48 more minutes buy 0.2 points beyond Family. Not worth the trade tonight.

Briefing generated from the four solved routes — planning aid only, not for navigation.

The actual panel, frozen. In the app, “View Family route” loads the recommended track onto the chart in one tap.

Departure briefing

Not just the best window — the argument for it

The departure planner solves your full passage for every leaving time you're weighing. The briefing explains the winner: which windows almost won and why they lost, what the losing hours would have cost in wind and heel, and a confidence rating that reflects how close the race actually was.

Confidence you can calibrate on

High, moderate, or low — scored from the spread between windows and any arrival caveats, so you know when the recommendation is a landslide and when it's a coin flip worth re-checking against the next model run.

The runners-up get named

“Thursday is faster but meets more wind” is a sentence you can sanity-check. A recommendation that names its rivals is one you can trust — or overrule with cause.

Deep dive: Departure Planner →

8 windows · Thu–Satwhy this departure

Recommended departure

Sat 02:00 UTC

37.7 h · comfort 4.6 · max wind 19 kt

DepartureDurationComfortWind

Thu 08:0036.6 h4.420 kt

Thu 14:0034.8 h4.419 kt

Fri 14:0033.9 h3.823 kt

Sat 02:0037.7 h4.619 kt

Depart Saturday 02:00 UTCmoderate confidence

Saturday's window gives the best comfort of the set (4.6) with max wind 19 kt, seas to 1.1 m, and a 24° max heel over a 37.7-hour passage. The Thursday slots are close behind at 4.4 and slightly faster, but the earlier one meets more wind and the later one still trails on comfort. Friday pushes wind to 23 kt and heel to 25° — a shorter passage that costs more than it saves. All eight windows are feasible; the trade is an extra hour at sea for the calmest crossing of the set.

Briefing generated from the computed windows — planning aid only, not for navigation.

Four of eight computed windows shown. Every number in the briefing comes from a full route solve of your boat, your draft, your routing mode.

The honesty layer

When the models disagree,
you’ll know first.

Serious sailors hand-roll uncertainty by flipping between apps. OpenSail re-samples every route against a second global model with an identical interpolator and renders the divergence on your track — a verdict, not a vibe.

Model agreement, on your route

ECMWF and GFS along the Nassau track. Agreement holds to +30h — then the band opens.

VERDICT: models agree through +30h, then diverge — leave in the early windowHow model agreement works →

Refusal is a feature.

When no safe route exists, OpenSail doesn’t guess — it explains. Rejection clusters, counted and mapped. Try getting a “why” out of a black box.

"status": "no_feasible_route",
"reasons": {
"shallow_rejections": 1_842, "land_rejections": 967,
"gust_ceiling_hits": 214 // NW 34kt fills the Stream
},
"suggestion": "Window opens Fri 18:00 — beam reach, max gust 22kt"

depth_verified is per-segment. Not a vibe.

Where bathymetry isn’t trusted, the route says so — leg by leg. Open data in (NOAA, ECMWF CC-BY, Copernicus), attributed out. You can check our inputs against the source.

Every data source, attributed →

The passage toolkit

Everything between
you and the dock lines.

OFFSHORE

Email in. Route out. At 2.4 kbps.

Send a route request as an email subject line over Iridium — get a plain-text plan plus a chartplotter-safe GPX back. Corridor GRIBs budgeted for satellite airtime.

SLEEP AID

Anchor Check

Drop a pin, mark your open sectors, get an hour-by-hour verdict for the night.