TelltaleINTELLIGENCE FOR THE WATER Support ↗

METHODS / PROVENANCE / VALIDATION

How Telltale models wind.
How the results are tested.

Telltale combines the driving forecast with the terrain, shoreline and thermal structure of a sailing ground. Historical replay and station comparisons test different parts of that chain.

01 / TELLTALE COASTAL WIND

The landscape is
part of the calculation.

TCW combines terrain, the driving wind and atmospheric stability to calculate acceleration, channel flow, lee wakes and turning. A mass-consistent adjustment constrains the flow around land.

Thermal circulation adds a parametrised sea breeze, land breeze and nocturnal drainage. It is reduced where the parent forecast already resolves the circulation, to avoid counting the same effect twice.

GEOMETRY

Terrain & shorelines

Copernicus elevation and explicit water masks preserve islands, channels and inland water. The published grid depends on domain and computation budget.

FORCING

The parent stays visible

TCW-E identifies ECMWF forcing; the I and G naming conventions identify ICON and GFS. The replay shown here is a saved model run, with its source recorded.

RESOLUTION

State the actual scale

The configured field is about 500 m in latitude, with finer operational targets and coarser large-domain ramps. This is cell spacing, not an accuracy radius.

WIND / AROUND ISLANDS AND THROUGH GAPS

Where the wind turns.
Where it accelerates.

Watch the mechanism, then consider the sailing decision. These maps follow saved Telltale wind vectors around the real Ionian coastline. Select a numbered explanation to connect a local feature with the sailing decision.

02 / TELLTALE INLAND WIND

A lake has
its own geography.

Water above sea level must still be treated as water.

Explicit lake masks preserve the real lake surface elevation. Shore contrasts, surrounding slopes and valley openings then have the right geometry for over-water flow and thermal effects.

North American lake observations now inform regional forecast selection. Alpine lake circulation and local corrections are evaluated separately.

INLAND WIND / THE LAKE AND THE SLOPES

Through valleys.
From slopes to shore.

COASTAL WAVES / FOLLOW THE ENERGY

Follow swell from open water
into shelter.

Start with the actual Telltale wave animation. Then isolate depth, shelter and returning energy in controlled model experiments.

CURRENTS & TIDES / FLOW THROUGH GEOGRAPHY

What the water does
around land.

Start with the saved regional current field. Then follow water around a complex coastline, through a narrowing passage and through flood, slack and ebb.

03 / THE VALIDATION RECORD

Historical replay.
Observed comparisons.

Historical reanalysis and archived operational forecasts answer different questions. TCW replays both through the production normaliser and downscaler, then compares at the reference’s own scale.

Evidence shown here: the documented October 2026 regional model-selection study. The current cross-region TCW validation record is being reconciled before publication.

45 US / Canadian observing stations60,015 paired forecast comparisons1–48 hours tested forecast leads

Choose the regional forecast
on measured performance.

The October 2026 study compared HRDPS and HRRR against lake buoys, shoreline stations and airport observations. HRDPS met the defined selection criteria for the Great Lakes aggregate and the Ontario / Ottawa study region. Telltale applies individual lake decisions where the evidence supports them. This tests the driving models; it is distinct from a TCW terrain-correction validation.

Global history

ERA5 provides the long historical baseline for replay. Dataset history and the period actually tested are reported separately.

European reanalysis

CERRA adds a regional comparison at its own scale. Replay can test terrain and weather regimes across many historical episodes.

US & Canada

Observed station and buoy comparisons inform which regional forecast drives each sailing ground.

Australia & Pacific

Validation in progress. Results will be published shortly.

REANALYSIS

Reconstruct the past

ERA5 supplies historical forcing. CERRA and the 9 km IFS analysis provide regional comparisons. Aggregate fine cells up to the reference scale; retain episode-level uncertainty.

OPERATIONAL REPLAY

Keep the forecast lead

Replay the model run that was available before the weather happened. A reconstructed historical field is not a forecast issued a day in advance.

OBSERVATIONS

Measure the missing scale

Airport reports constrain open-air forcing. Moorings and boat logs can test different settings. Sampling a point instrument is different from averaging into a model grid box.

Data sources and the reproducibility record

Each replay episode records its source, valid time, code revision and explicit parameter set. The archive keeps cached inputs and per-terrain-bin errors, speed bias, circular direction error and whole-episode bootstrap bands.

04 / KNOW WHAT THE PICTURE CAN SAY

Three layers.
Different claims.

Local wind

Diagnostic terrain and thermal response to a parent forecast. Small-scale features require small-scale observational validation.

Coastal waves

Directional propagation with refraction, shelter, diffraction and reflection. Not a full frequency spectrum, harbour-resonance solve or wave–current coupling model.

Currents

Upstream ocean-current context at its native scale. A dense display is not a bay-scale tidal-stream forecast. The current replay uses saved Copernicus Marine surface vectors; tidal mechanism diagrams are conceptual.