# Marine weather: understand forecasts, compare apps and train

ISLADREAM · 2 November 2026 · https://isladream.pages.dev/en/journal/meteo/

Useful weather analysis leads to an explained decision: depart, wait, reduce sail, change course or keep shelter accessible. This guide combines French references, tool comparisons and original exercises. Prices checked on 2 November 2026; ISLADREAM has not personally tested these applications or attended these courses.

Documentary synthesis, without an onboard survey. Adapt decisions to the boat, current manuals and actual conditions.

## 1. Start with a sailing decision

Before opening an animated map, write down the proposed passage, realistic duration, alternatives and crew capability. The question is not “what wind tomorrow?” but “what conditions will we meet at departure, in the exposed area and on arrival if delayed?” A harbour forecast may not describe an inter-island channel or the return upwind.

Make a four-part sheet: available information, uncertainties, current decision and the condition that would change it. Illustrative example: if two scenarios make a late arrival uncomfortable, leaving earlier is not the only option; a shorter leg or waiting also deserves consideration. Avoid universal wind limits for every boat. Limits should account for familiar sail configurations, sea state, fatigue, visibility and nearby hazards. This framing allows tools to be compared for actual usefulness. A purchase is easier to justify when it resolves a named information gap rather than merely adding another weather screen.


## 2. Official bulletin, observation and model: three roles

Météo-France publishes marine bulletins by area and special warnings known as BMS. Read the area, issue time and validity before the detail, and save the bulletin covering the passage. An observation reports a past or recent measurement at a location. A model calculates a possible evolution. Putting these in separate columns avoids presenting a forecast as an observation.

Suggested routine: read the bulletin, examine the broad situation, compare passage variables and review relevant observations. If reality diverges from the selected scenario, investigate rather than opening more applications until one displays a reassuring value. Two applications may use the same model, so their agreement is not necessarily agreement between independent forecasts. Record model and cycle in the weather log, then explain the decision to the crew in one sentence. A coloured display should produce an explanation that still makes sense when the screen is switched off.


References for this section : [Météo-France — Bulletins météo marine et BMS](https://meteofrance.com/meteo-marine)

## 3. ARPEGE, AROME, ECMWF: choose the right scale

Météo-France distinguishes global ARPEGE from finer regional AROME. ECMWF provides ensemble forecasts representing multiple possible developments. Visual detail does not guarantee the exact position of a squall or a gust at the displayed time. Verify actual coverage: an AROME mainland-France product should not be assumed identical to a Caribbean product.

Compare the same variable at the same valid time, using cycles as close as practicable. First examine the scenario’s consistency over several hours, then coastal detail. A small grid does not remove forecasting limitations. If route options change substantially between models, treat that sensitivity as decision information. Prepare a plan that remains workable across several reasonable scenarios instead of treating the most favourable curve as a promise. Record the missing area or variable as a limitation rather than silently filling it with data from a different time or product.


References for this section : [Météo-France — Les modèles de prévision météo](https://meteofrance.com/meteo-a-z/les-modeles-de-prevision-meteo), [ECMWF — Medium-range forecasts and ensemble uncertainty](https://www.ecmwf.int/en/forecasts/documentation-and-support/medium-range-forecasts)

## 4. Read a GRIB without treating it as a bulletin

GRIB is a data format, not a safety certification. Before reading, check model, cycle, area, time step, units and final forecast time. The navigator should know whether the display uses local time or UTC. Including date and cycle in the file name avoids confusion when several downloads coexist. Smooth animation may interpolate between forecast times; it does not mean an observation exists for every frame.

Dockside exercise: download two files for the same area, disconnect and locate their first and last valid times without assistance. Compare one route point across three successive times. If a variable is missing, record the gap instead of inferring it from another layer’s colour. Keep warning text separately. The compact file supports spatial reasoning but does not remove the need to read relevant official information. A useful offline test includes reopening the application after restart, not merely viewing a file already cached on screen.


## 5. Wind, gusts and waves: compare like with like

Météo-France’s marine guide defines significant wave height from the highest third of waves; it is not maximum height. Mean wind and gusts are also different variables. A single number cannot establish that no larger wave or stronger wind will occur. Add wave direction and period and their relationship to the intended heading.

For a day sail, write three lines: what assists progress, what tires the crew and what makes arrival difficult. Situations with the same displayed wind may justify different decisions when sea state, route angle and shelter change. Method example: compare the leg in both directions before selecting a lunch anchorage. A pleasant outward trip does not guarantee an acceptable return. A table reduced to wind alone often omits the decisive question. Keep the crew’s practical limits visible beside the data, including whether they can move, rest and operate familiar sail controls in the expected conditions.


References for this section : [Météo-France — Guide marine, édition octobre 2022 : définitions](https://meteofrance.fr/sites/meteofrance.fr/files/files/editorial/Guide_marine_octobre2022.pdf)

## 6. Compare applications by intended use

This table summarises functions described by the consulted publishers. It does not measure forecast accuracy or rank applications. Windy.com and Windy.app are separate services: verify the publisher before paying. Apply the same test to each: identify the cycle, examine a familiar area, open data without connectivity and explain a decision to another crew member.

Choose the need first: broad weather context, a local outing, departure comparison or compact offshore downloads. Then check platform, coverage, charts, file size and renewal terms. A routing function does not replace examining the route on suitable nautical charts. “Verify” means that our consultation did not establish the information, not proof that a feature is absent. An application that meets a day-sailing need can still fail a specific offshore workflow, so test that workflow before committing the voyage to it.


| Tool | Documented use | Pre-departure check |
| --- | --- | --- |
| PredictWind Standard / Pro | Routing, departure planning, satellite weather access | Chosen tier and terminal/connection chain; airtime separate |
| Weather4D Routing & Navigation | iPhone/iPad: GRIBs, routing, nautical charts | Separate weather and GeoGarage charts; offline download |
| SailGrib WR | Android: GRIBs, routing, AIS/NMEA and charts | Licence, chart purchases, device and offline operation |
| Squid | GRIBs and area-specific model offers with tiered functions | Exact coverage, version and subscription for cruising area |
| Windy.com | Weather map; distinct from Windy.app | Displayed model, Premium functions and offline workflow to verify |
| Windy.app | Spot forecasts and wind/water sports | This publisher’s own Pro offer; subscriptions are separate |
| Windfinder | Different Pro and Plus products | Pro one-time purchase; Plus subscription; check store-specific functions |

References for this section : [PredictWind — Standard et Professional, fonctions annoncées](https://help.predictwind.com/en/articles/8563285-predictwind-subscriptions-differences-between-standard-and-professional), [Weather4D — Routage & Navigation, présentation éditeur](https://www.weather4d.com/fr/project/weather4d-routage-navigation/), [SailGrib — Applications Android, présentation éditeur](https://sailgrib.com/fr/), [Squid — FAQ officielle](https://www.squid-sailing.com/en/faq), [Windy.com — Distinction officielle avec Windy.app](https://community.windy.com/topic/22297/windy-vs-windy-app), [Windy.app — Présentation officielle](https://windy.app/), [Windfinder — Achats Pro et abonnement Plus](https://www.windfinder.com/help/subscriptions/troubleshooting)

## 7. Published prices: currency, period and scope

Checked on 2 November 2026. These figures come from official pricing pages and are not guaranteed quotations. PredictWind uses US dollars and may add location-dependent taxes. Squid displays euros including French 20% VAT, adjusted to the country in the cart. Weather4D warns that prices vary by country; the consulted comparison table does not explicitly state tax treatment. Confirm the final total and renewal before purchase.

Some applications deliberately have no consolidated price here: several in-app purchase amounts do not always identify duration, package and tax treatment clearly. We have not assigned them an invented single price. Keep application purchase, weather data, nautical charts and communications separate. A forecast subscription does not automatically pay for a terminal or satellite traffic. Match the payment period to the actual sailing season, then compare annual commitment and shorter packages without assuming they can be paused.


| Offer | Displayed price | Scope |
| --- | --- | --- |
| PredictWind Basic | USD 29/year or USD 12/ 3 months | Possible additional tax |
| PredictWind Standard | USD 249/year or USD 99/ 3 months | Possible additional tax |
| PredictWind Professional | USD 499/year or USD 249/ 3 months | Possible additional tax |
| Weather4D Lite / Routing / R&N | € 19.99 / € 46.99 / € 64.99 | Application purchase; separate editions |
| Weather4D weather / SHOM charts | € 45.99/year / € 59.99/year | Separate subscriptions; SHOM through GeoGarage |
| Squid Europe / US & Caribbean | € 9.98/month each | French 20% VAT included, adjusted by country |
| Squid Mediterranean / World & Satellite | € 19.99 / € 69.98 per month | French 20% VAT included, adjusted by country |
| Windy.com / Windy.app / Windfinder / SailGrib | Verify final price in local offer or store | Period and taxes not consolidated here |

References for this section : [PredictWind — Tarifs officiels](https://explore.predictwind.com/pricing), [Weather4D — Tableau comparatif et tarifs](https://www.weather4d.com/fr/tableau-comparatif/), [Squid — Offres et tarifs](https://www.squid-sailing.com/en/offers), [Windy.com — Offre Premium](https://www.windy.com/subscription), [Windfinder — Achats Pro et abonnement Plus](https://www.windfinder.com/help/subscriptions/troubleshooting), [SailGrib — Applications Android, présentation éditeur](https://sailgrib.com/fr/)

## 8. Routing: test robustness, not just arrival time

A polar describes the performance assumed by the calculation; if it does not match the loaded boat and actual sailing style, the result may be impractical. Start from observed performance and document assumptions. Routing functions advertised by PredictWind, Weather4D or SailGrib do not remove the need to check the route and its constraints. An optimised line is not permission to pass.

Suggested method: calculate several departures, reduce assumed performance and compare weather scenarios. Record exclusion areas, acceptable wind and sea conditions, night-sailing preferences and arrival constraints. If a small adjustment produces a very different route, do not simply select the fastest. Investigate the change. Then manually check coastal legs on suitable charts. Sometimes the useful result is “three scenarios preserve access to shelter” rather than “arrival eight minutes earlier”. Keep the assumptions with the saved route so another watchkeeper understands why it was chosen and what might invalidate it.


References for this section : [PredictWind — Standard et Professional, fonctions annoncées](https://help.predictwind.com/en/articles/8563285-predictwind-subscriptions-differences-between-standard-and-professional), [Weather4D — Routage & Navigation, présentation éditeur](https://www.weather4d.com/fr/project/weather4d-routage-navigation/), [SailGrib — Applications Android, présentation éditeur](https://sailgrib.com/fr/)

## 9. Offshore downloads: design a degraded mode

First check the entire chain: application, weather subscription, terminal, communications contract, antenna, energy and user. Compatibility with a satellite system does not mean airtime is included. Test receipt of a bulletin followed by a limited file, offline opening and recovery after interruption. Record actual transferred size and elapsed time on the selected equipment.

Reduce a request through area, forecast times and useful parameters, without arbitrarily removing the diversion area. Consider a broad context file and a more detailed selection for the immediate passage. Test these choices before departure because options and limits vary by service. Invented sizing example: four 250 kB files daily for twenty days total 20,000 kB before transmission overhead. This describes no particular satellite tariff; it shows why request frequency matters as much as file size. Reserve a clearly documented priority order for weak connections rather than repeatedly trying a large failed request.


## 10. Coastal case: examine the return and critical times

Training scenario, not a real forecast: a family plans a 35-mile leg past an exposed headland with two possible ports. Prepare three headland times using normal progress, slower progress and a delayed departure. For each, compare mean wind, gusts, sea state and access to the alternative. Do not automatically align the best weather value with ideal boat speed.

Compare the map application, area bulletin and relevant observations. The decisive question becomes: does the plan remain acceptable if the headland is reached two hours late? If the answer depends on an unverified detail, seek an update or change the leg. Afterwards, compare saved forecasts with onboard observations, matching position and time. One disappointing trip does not establish that an application is always poor; it identifies what needs better documentation. Keep this exercise short enough to repeat before ordinary outings, not only ambitious passages.


## 11. Caribbean case: squalls, channels and cyclone risk

For the French Caribbean, retain access to local Météo-France services and warnings. Sailing in an island’s lee, crossing a channel and approaching an exposed destination are not one uniform situation. Separate the broad wind pattern, observed cloud development, inter-island crossing and arrival anchorage conditions. Apparently regular trade winds do not remove the need for reassessment.

For cyclones, NHC’s cone describes the centre’s track uncertainty, not the complete extent of hazards. Never interpret being outside the cone as a guarantee. Prepare shelter options and official instructions in advance, then follow current bulletins. This guide supplies no cyclone-avoidance route. A useful exercise outside an alert situation is to list the time needed to prepare the boat, move the family and reach an authorised safe location. Compare that with realistic response capacity, not a reassuring animation. Decide who monitors official updates when the skipper is occupied with preparation.


References for this section : [Météo-France Guadeloupe — Cyclone et vigilance](https://meteofrance.gp/fr/cyclone), [NHC — Définition du cône de trajectoire cyclonique](https://www.nhc.noaa.gov/aboutcone.shtml)

## 12. Transatlantic case: different horizons, different decisions

An Atlantic crossing requires separating seasonal preparation, departure window and decisions over the next few days. A distant trend helps organise options; it does not provide an exact route for the whole crossing. ECMWF uses ensemble spread to express uncertainty, making several possible developments more informative than memorising one arrival time. Spread does not capture every possible forecasting-system weakness.

Suggested original routine: reserve a daily review, record recent observations, compare planned and actual progress and reconsider accessible options. Add a weather review when a developing event or discrepancy warrants it, without assuming one universal update frequency. Assess an extra day’s effect on water, food, energy and fatigue. A route also depends on the crew’s ability to maintain it. A fast but exhausting calculation is not automatically the best plan. Hand over the reasoning as well as the file so the next watch understands what would trigger reassessment.


References for this section : [ECMWF — Medium-range forecasts and ensemble uncertainty](https://www.ecmwf.int/en/forecasts/documentation-and-support/medium-range-forecasts)

## 13. French training: compare objectives and prices

Weather education and application training solve different needs. Before booking, ask about entry level, individual exercises, cruising areas studied and critical interpretation. Bringing an actual planned passage makes learning more concrete. The table documents public offers without a partnership, testimonial or first-hand quality endorsement.

Watch the dates: the quoted Navistage session is scheduled for 5–6 November 2026 with an application deadline of 21 October, already past at this guide’s date. It is a price reference, not an availability announcement. EFT figures are catalogue prices; tax treatment and additional expenses require confirmation. STW describes a 15% member discount, which should be confirmed for the particular course and session rather than automatically deducted below. Include travel and accommodation in the comparison, and distinguish a training certificate from evidence that both crew members can apply the method aboard their own boat.


| Course / resource | Published objective | Price / limitation checked 02 Nov 2026 |
| --- | --- | --- |
| Météo-France Navistage | Marine weather, bulletins, charts;2 days | € 550 excluding tax; session and deadline above |
| EFT offshore weather level 1 | Weather systems and strategic reading;2 days | € 369 displayed; taxes/extras to confirm |
| EFT offshore weather level 2 | Local effects and critical routing; prerequisites | € 369 displayed; taxes/extras to confirm |
| STW | Information and access to partner weather courses | Confirm membership terms; not assumed to be a separate additional course |
| Les Glénans | Cruising progression; select a suitable course | No single weather-course price established; check syllabus and shared expenses |

References for this section : [Météo-France — Navistage, programme et tarif 2026](https://services.meteofrance.com/index.php/formations-et-stages/formation-meteo-marine-navistage), [Escale Formation Technique — Météo marine hauturière niveau 1](https://www.escaleformationtechnique.com/fr/programme/m%C3%A9t%C3%A9o-marine-hauturi%C3%A8re-niveau-1), [Escale Formation Technique — Météo marine hauturière niveau 2](https://www.escaleformationtechnique.com/fr/programme/m%C3%A9t%C3%A9o-marine-hauturi%C3%A8re-niveau-2), [Escale Formation Technique — Catalogue tarifaire](https://www.escaleformationtechnique.com/), [STW — Formation météo, partenariat pédagogique et conditions](https://stw.fr/fr/formations/meteo), [Les Glénans — FAQ formations](https://www.glenans.asso.fr/faq)

## 14. Learn to compare without inventing a ranking

To assess a tool, retain forecasts before knowing the outcome. Record location, time, variable, model, lead time and the observation used. Comparing yesterday’s forecast with today’s measurement requires matching variable definitions; a boat’s apparent wind cannot be directly compared with meteorological wind without suitable treatment. A few memories of successful outings are not an accuracy trial.

Accessible exercise: during a month of preparation, repeatedly analyse the same area and write down what would have changed your decision. Count cases where missing information, interface problems or time confusion caused difficulty. These observations support a personal usability comparison without claiming scientific model superiority. For training, ask each crew member to explain a bulletin, identify an expired file and propose an alternative. These are observable outcomes, more useful than the number of installed applications. Save both the reasoning and the mistake so that the next practice session addresses a real weakness.


## 15. Build a complete weather budget

Separate equipment already owned, required purchases, recurring subscriptions, training and communications. Example calculated from displayed Weather4D prices: Routing & Navigation, annual weather and SHOM charts total € 64.99 +€ 45.99 +€ 59.99 =€ 170.97 in year one, before hardware and communications and subject to local checkout. This is not a commercial bundle. Later years should not automatically include repurchasing the app; check each subscription’s terms.

Compare the arrangement with the need, not merely another total. Coastal preparation may prioritise learning public bulletins. Offshore sailing adds distinct download, chart and communications requirements. Review prices annually and date every comparison. Saving on an unused subscription can be sensible; removing a function required by the tested onboard procedure is a different decision. Allocate practice time alongside money, because an unfamiliar paid feature may provide less practical value than a modest tool that both crew members understand.


## Key point

The best weather arrangement turns dated information into revisable decisions understood by the crew and workable across several scenarios.

## Your practical checklist

- [ ] Record area, issue time, validity and next weather review.
- [ ] Read official bulletins and warnings separately from models.
- [ ] Examine wind, gusts, waves, period and a delayed arrival.
- [ ] Identify models and cycles; explain disagreements to the crew.
- [ ] Actually open offline data on onboard equipment.
- [ ] Define shelter, alternative departure and a condition for cancelling.
- [ ] Separate price, currency, tax, renewal, charts and communications.

## Sources & documents

- [Météo-France — Bulletins météo marine et BMS](https://meteofrance.com/meteo-marine) — 2026-11-02
- [Météo-France — Les modèles de prévision météo](https://meteofrance.com/meteo-a-z/les-modeles-de-prevision-meteo) — 2026-11-02
- [ECMWF — Medium-range forecasts and ensemble uncertainty](https://www.ecmwf.int/en/forecasts/documentation-and-support/medium-range-forecasts) — 2026-11-02
- [Météo-France — Guide marine, édition octobre 2022 : définitions](https://meteofrance.fr/sites/meteofrance.fr/files/files/editorial/Guide_marine_octobre2022.pdf) — 2026-11-02
- [PredictWind — Tarifs officiels](https://explore.predictwind.com/pricing) — 2026-11-02
- [PredictWind — Standard et Professional, fonctions annoncées](https://help.predictwind.com/en/articles/8563285-predictwind-subscriptions-differences-between-standard-and-professional) — 2026-11-02
- [Weather4D — Tableau comparatif et tarifs](https://www.weather4d.com/fr/tableau-comparatif/) — 2026-11-02
- [Weather4D — Routage & Navigation, présentation éditeur](https://www.weather4d.com/fr/project/weather4d-routage-navigation/) — 2026-11-02
- [SailGrib — Applications Android, présentation éditeur](https://sailgrib.com/fr/) — 2026-11-02
- [Squid — Offres et tarifs](https://www.squid-sailing.com/en/offers) — 2026-11-02
- [Squid — FAQ officielle](https://www.squid-sailing.com/en/faq) — 2026-11-02
- [Windy.com — Distinction officielle avec Windy.app](https://community.windy.com/topic/22297/windy-vs-windy-app) — 2026-11-02
- [Windy.com — Offre Premium](https://www.windy.com/subscription) — 2026-11-02
- [Windy.app — Présentation officielle](https://windy.app/) — 2026-11-02
- [Windfinder — Achats Pro et abonnement Plus](https://www.windfinder.com/help/subscriptions/troubleshooting) — 2026-11-02
- [Météo-France Guadeloupe — Cyclone et vigilance](https://meteofrance.gp/fr/cyclone) — 2026-11-02
- [NHC — Définition du cône de trajectoire cyclonique](https://www.nhc.noaa.gov/aboutcone.shtml) — 2026-11-02
- [Météo-France — Navistage, programme et tarif 2026](https://services.meteofrance.com/index.php/formations-et-stages/formation-meteo-marine-navistage) — 2026-11-02
- [Escale Formation Technique — Météo marine hauturière niveau 1](https://www.escaleformationtechnique.com/fr/programme/m%C3%A9t%C3%A9o-marine-hauturi%C3%A8re-niveau-1) — 2026-11-02
- [Escale Formation Technique — Météo marine hauturière niveau 2](https://www.escaleformationtechnique.com/fr/programme/m%C3%A9t%C3%A9o-marine-hauturi%C3%A8re-niveau-2) — 2026-11-02
- [Escale Formation Technique — Catalogue tarifaire](https://www.escaleformationtechnique.com/) — 2026-11-02
- [STW — Formation météo, partenariat pédagogique et conditions](https://stw.fr/fr/formations/meteo) — 2026-11-02
- [Les Glénans — FAQ formations](https://www.glenans.asso.fr/faq) — 2026-11-02