SelmaRacing Mode
— — — —

No race selected

—
BoatSOGknVMCkn DTWnmETAΔ vs menm 5′15′Age

Event log

Tacks, mark roundings and alerts will appear here.

Measured now

No position yet
Model Windkn · Δ Gustkn · Δ Dir° · Δ Score Updated
No models selected — choose models in Setup → Weather models.

Δ = forecast − measured. Score ranks the selected models against each other by recent agreement with your B&G instruments at this position (wind, gusts, direction). Higher is better — but it reflects skill here, now, not necessarily downstream where you're routing to.

Rolling accuracy · window min

Model Wind biasknRMSEkn Gust biaskn Dir bias°Dir RMSE°n
No comparison cycles recorded yet.

Bias = mean signed error (a steady offset across all models usually means instrument calibration, not the model). Forecast snapshots stored for future lead-time scoring: 0.

Choose which models to rate, and tune the comparison, in Setup → Weather models.

Corrected leaderboard

Corrected = (elapsed + projected time to finish at current VMC) × ToT. Finished boats use their actual elapsed. Lower is better; ratings in Fleet.

#BoatMarkDTWnm To finnmSOGknVMCkn ToTETA finCorrected

Replay

--:--:--
Select a race with recorded data.
BoatSOGVMCDTWMark

Races

My boat

Not configured.

Competitors

Scan nearby AIS

Listens to all AIS traffic around your current position and lets you add the closest vessels as competitors — useful on the start line.

Connection

disconnected

Position source for my boat: AIS (own MMSI) and/or this device's GPS — set in "Edit my boat".

Weather models

Pick which models to pull and rate against your B&G instruments (Weather tab). Global models always cover you; regional high-res models appear only when you're inside their domain.

Weather comparison

ⓘ How the height correction works

Weather models report wind at the standard 10 m reference height. Your anemometer sits at the masthead (e.g. 20–28 m), where the wind is stronger because of the vertical wind gradient (shear) over the sea. Comparing the two raw values would blame the model for what is really a height difference.

When enabled, both the base wind (window mean) and the gust (window max) are scaled down to their 10 m equivalent using the power-law profile:

V₁₀ = Vmast × (10 / H)α

  • H — anemometer height above water (the field below)
  • α = 0.11 — power-law exponent for neutral open-sea conditions

Wind direction is not adjusted — veer with height is small. The same factor is applied to base and gust, before any model comparison or scoring.

Example at H = 28 m: factor = (10/28)^0.11 ≈ 0.891, so 13.0 kn at the masthead → 11.6 kn at 10 m. Lower mast → smaller correction; at H = 10 m the factor is 1.0 (no change).

Note: this is a single-exponent approximation; it ignores atmospheric stability and is a first-order fairness adjustment, not an exact boundary-layer model.

Stream debug

Alerts

Events

Tack / gybe detected
Mark rounded / finished
Boat within nm of a mark
Speed drop below % of the boat's speed min ago
Repeat the sound every s until my boat is back up to speed (own boat only), intensity
Show a browser notification (works offline when this tab is open)
AIS signal lost (no fix for 10 min)
Boat speed below % of the polar target for min — polar % averaged over s

Averaging window: 15–20 s flat water / upwind trimming (quick feedback) · 30–45 s moderate sea · 60–90 s downwind in waves, where surfing cycles last 10–15 s and a short window just follows the waves. The alert judges this same averaged value, so size the minutes above accordingly.

Repeat the sound every s until the boat is back on target, intensity
Show a browser notification (works offline when this tab is open)

Heading more than ° off a set bearing of ° for s
Repeat the sound every s until back within tolerance, intensity
Show a browser notification (works offline when this tab is open)

Fleet changes (boats added from scan / map)
Info messages (imported, connecting…)

Unticked events show no banner and play no sound; they are still written to the event log. Repeat rate: the interval applies right at the threshold and shortens as the discrepancy grows; intensity 1 = steady; for the polar alert 50 halves the interval per 10 points below the threshold and 100 cuts it to a third; heading and speed react harder — per 10° or 1 kn beyond the threshold, 50 cuts the interval to about a third and 100 to about an eighth. Never below 0.25 s. Use Test to hear it. Mute silences the current episode.

Sounds

Volume
Vibrate (phones)

Sound clips: Kenney "Interface Sounds", CC0 (public/sounds/LICENSE.txt). The first sound needs one tap anywhere on the page (browser rule).

Data

Browser storage used—

Everything is stored locally in this browser (localStorage, about 5 million characters). Export regularly as backup; a warning appears at 70 %. Auto-export rewrites one chosen file on a timer (Chrome / Edge only; the choice lasts until the tab is closed).