Every departure decision I make starts with the forecast. Not with the forecast as a number to be accepted or rejected, but as a document to be read: a specific set of information produced by a specific model at a specific time, with specific strengths and specific limitations, and with a confidence level that varies depending on how far ahead you are looking and how stable the atmospheric pattern is in the region you are sailing through.
Most people who sail recreationally read the forecast the way they read a weather app on a phone: as a prediction of conditions to be accepted or doubted. The offshore sailor reads it differently: as a model output to be interrogated, compared against other model outputs, evaluated against the actual conditions at the time of reading, and used as one input among several in a decision that combines data with experience and local knowledge. This article describes how I do it and why the method matters more than the source.
The GRIB File
A GRIB file is a gridded binary file containing forecast data (Predictwind Offshore or Windy) from a numerical weather prediction model. For the offshore sailor, this typically means wind speed and direction, pressure, significant wave height, and swell period and direction, projected forward in time at regular intervals. The most commonly used model for offshore sailing is the GFS (Global Forecast System), which runs four times daily and produces forecasts out to sixteen days. The ECMWF (European Centre for Medium-Range Weather Forecasts) model is generally more accurate but more expensive to access in its full resolution.
On the Eighty2, I download GRIB files through the Starlink connection before every significant departure and at least once every 24 hours on a passage. The download takes seconds. The reading takes longer because the GRIB file contains more information than a single pass through it can absorb: the pressure pattern, the isobar spacing that indicates wind gradient, the forecast sea state and its relationship to the forecast wind, and the specific question of whether the model’s output for the next 24 hours is consistent with what is actually happening at the boat’s position at the time of reading.
Wind Speed versus Wind Gust
The forecast gives wind speed as a sustained average, typically over a ten-minute period. The gust is the peak wind speed within a shorter period, typically three seconds. The difference between the two is significant for sailing: the boat’s response to wind is primarily determined by the sustained wind, but the gust tests the rig and the helm in a way that the sustained wind does not. A forecast of 20 knots sustained with gusts to 28 is a different sailing proposition to a forecast of 20 knots with gusts to 22, even though the sustained wind is identical.
The gust factor, the ratio of gust to sustained wind, is highest in unstable air masses with convective activity: the trade winds produce lower gust factors than a cold front passage, and the Mediterranean summer Meltemi produces lower gust factors than the same Beaufort number at the edge of an Atlantic low. Understanding the atmospheric regime you are sailing in tells you as much about the gust factor as the forecast number does, and this is the kind of knowledge that comes from having sailed in a region across multiple seasons rather than from reading a single forecast.
Pressure and the Isobar Spacing
The isobar is the line connecting points of equal atmospheric pressure on a weather map. The spacing between isobars tells you about the pressure gradient: close isobars mean a steep gradient and strong winds, widely spaced isobars mean a shallow gradient and light winds. This relationship is direct and reliable in the middle latitudes where the gradient wind equation applies cleanly. In the tropics, where the thermal effects and local circulations are more significant relative to the synoptic pressure pattern, the isobar spacing is a less direct predictor of conditions, which is one of the reasons tropical weather forecasting is more uncertain than temperate latitude forecasting.
When I look at a synoptic chart before a departure, the first thing I assess is the pressure pattern over the area to be sailed: where the high is, where the low is, how the isobars are oriented relative to the route, and whether the pattern is stable or actively evolving. A stable high with widely spaced isobars over the route means a window that is likely to produce what the forecast says it will produce. A developing low with tightening isobars means uncertainty: the model may be capturing the development correctly or it may be running behind the actual evolution, and the confidence in the forecast is lower than the stable pattern provides.
The 72-Hour Window
For passage planning, the 72-hour forecast is the most reliable and the most relevant. The first 24 hours of a model forecast are typically accurate to within a Beaufort number in stable conditions. The second 24 hours introduce more uncertainty: the model’s representation of frontal passages and pressure system development at this range is good but not perfect, and the difference between a front arriving on schedule and arriving six hours late can be the difference between a comfortable passage and a difficult one. The third 24 hours should be treated as probabilistic rather than specific: the direction and approximate strength are likely to be approximately right, but the timing and the peaks require verification against the next available model run.
Beyond 72 hours, I use the forecast as context rather than as a decision input: to understand the synoptic situation that will exist at the end of a longer passage, to identify what systems might be in the area, and to make a provisional assessment of whether the weather is likely to be broadly favourable or unfavourable for the passage window. I do not make departure decisions based on the five-day forecast alone. I make them based on the first 72 hours, verified against at least two model runs on the day of departure, with the five-day providing the background that those 72 hours sit within.
