Key facts
| Fact | Detail |
|---|---|
| NWS marine wind | Wind at 10 m above the sea, averaged over two minutes |
| NWS marine gust | Highest 5-second wind inside a two-minute interval |
| Small NDBC buoys | Anemometer about 5 m above the waterline |
| Large NDBC buoys | Anemometer at 10 m |
| Speed amendment | Sustained wind change of 10 knots or more |
| Direction amendment, 20 knots and above | Change of more than 45 degrees |
| Direction amendment, below 20 knots | Change of more than 60 degrees |
| Working sample | Five comparable trips in one direction bucket |
A repeatable gap differs from a bad afternoon
The method needs one page, a fixed set of fields and a buoy or handheld anemometer for the observation. The decision it changes is simple: the wind speed at which you launch, and the direction bucket that applies. A forecast bias is systematic. An error is any one-off difference. A model can miss one afternoon because cloud formed early or a sea breeze arrived late; that error says nothing about the next trip. A model can also run 3 knots high every time the wind comes from the southwest at one headland. That second pattern repeats with the same inputs, so it can be corrected. Three mechanisms produce local bias. Grid resolution averages the wind over a box a few kilometres wide, mixing open water, beach and land into one number. Terrain accelerates flow over a headland, shadows it in the lee and channels it through a gap, at a scale no forecast grid resolves. Measurement height and exposure add the third: a forecast describes wind at a standard height over open water, while a buoy or a handheld instrument measures at whatever height it happens to sit. Height matters offshore too. A 10 m buoy carries its anemometer 10 m above the waterline. A 3 m or 6 m buoy carries the sensor about 5 m up. The National Data Buoy Center adjusts its winds to 10 m, and also to 20 m for comparison with ships, but the raw station page shows the unadjusted value. Part of a two-buoy disagreement is geometry rather than weather. Look for a gap that repeats with the same wind direction. A forecast of 18 knots against a buoy reading of 14 on three southwest days is information. The same gap on one day is weather, and the log should record it without changing any threshold. Two high days and three flat ones do not form a reliable pattern yet.
Compare like with like
The definitions decide whether a comparison is fair. A marine forecast describes wind at 10 m above the sea surface averaged over two minutes. A gust is the highest 5-second wind inside a two-minute interval. A handheld anemometer at deck height reads a different quantity: lower because it sits closer to the water, and shaped by a different averaging window. Over open water, wind speed rises with height along a logarithmic profile. A sensor at 5 m reads a few percent lighter than the 10 m wind, on the order of one-twentieth to one-tenth in neutral air. That is small, but it has the same sign on every trip, so it enters the offset as a constant. Use the height-adjusted product when the station publishes one, and note the height effect in the log when it does not. Average over the same window. A two-minute mean and a ten-minute mean can differ by several knots in gusty air, especially behind a front. Compare the forecast block with an observation from the middle of that block, not with the single highest reading of the day. A 20-knot gust inside a forecast of 12 to 18 knots is a gust error; the sustained forecast can still be right. Record the gust separately from the sustained wind. A limit stated as 15 knots means one thing for a two-minute average and another for the peaks inside it. Set a comfort limit on the gust and a drift or casting limit on the sustained number, then keep the other value as context. A 15-knot gust ceiling and a 15-knot sustained ceiling describe very different days, and the log should say which one the decision used.
Five trips make a block
Five is a working rule. It is enough samples to see whether a gap has a consistent sign, and few enough that the log does not become a research project. The trips must be comparable: same general wind direction, same season, same stretch of coast, and a forecast and observation taken at the same time of day. A 6 a.m. comparison and a 2 p.m. comparison measure different winds, because the sea breeze builds through the afternoon. A stable sign across the block points to a bias. If four of five trips show the forecast high by 2 to 4 knots and the fifth shows it low by 2, the block points to a high bias with one outlier. If the sign changes with every trip, no bias exists to correct, and the threshold should stay where it is. Do not change the threshold after one bad trip. Change the threshold by the size of the gap, not by the size of the worst day. A block that averages 3 knots high supports a 3-knot correction. The worst day in that block is an outlier, and a correction built on it makes every ordinary trip more conservative than the evidence supports. Round to whole knots; the comparison cannot support a decimal. Keep the block clean. A trip with a missing buoy record is not a sample. A trip in a different season is a different block. If a heat wave week and a frontal week both sit in the same five, the average hides two different errors and the correction fits neither. Start a new block when the season or the weather pattern changes.
Split the log by wind direction
A single average across all directions hides the pattern. The same headland accelerates a southwest wind and shelters a northeast wind, so one direction runs low while the other runs high on the same day. Grouping by direction separates a useful correction from a number that makes every trip worse. Four buckets, north, east, south and west, are enough to start. Use the direction the wind comes from, matching the way marine forecasts report it. A southwest wind blows from the southwest, and the log records the source bearing rather than the heading it pushes toward. That convention matters at a headland, where the local wind bends around the point and arrives at the ramp from a different bearing than the forecast zone. Log the direction at the water as well as the forecast direction when the two differ. Check whether the bias appears in one bucket and disappears in another. A southwest bucket running 3 knots high and a northeast bucket running 1 knot low gives two local corrections, not one global offset. Apply each only in its own bucket. When the wind sits between buckets, use the nearer one and treat the correction as approximate, because the transition is where the terrain effect changes fastest.
The log fits on one page
Six fields and one row per trip cover the method: date and time, forecast wind speed and direction, forecast gust, observed wind speed and direction, sea state, and the go or no-go decision. The field list is based on a 2019 log-keeping guide: wind in knots and the direction it comes from, gusts, seas and water clarity. Add a final column for the difference in knots and the difference in degrees. Write the row within an hour of returning. A row written from memory a week later records an impression, not the day. A usable row reads without interpretation. Forecast 15 southwest with a gust of 20; buoy 12 southwest with a gust of 16; difference plus 3. That row can be compared with the next one. Windy, and the forecast was off, cannot. Keep the fields identical from row to row, whether the page lives in the boat bag or on a phone, because a changed field breaks the comparison at the point where the log starts to matter. Record sea state and water clarity beside the row; they explain whether a building breeze or a passing front produced the gap. Photograph the buoy page at the ramp on the way out. The station archive holds the record, but the photograph pins the observation to your launch time. It also shows whether the station was reporting at all, which a missing row cannot. Keep the images with the log rows for the season, then delete them when the block closes. A dated image settles any later question about which record the row used.
Apply the offset to the threshold
The correction is arithmetic on the forecast. With a 3-knot high bias in the southwest bucket, a forecast of 18 knots means planning for 15. If the personal limit is 15 knots, 18 in the text is the ceiling for that direction. A low bias works the other way: a forecast of 10 with a 3-knot low bias means planning for 13, so a 15-knot limit leaves less margin than the text suggests. An observation outranks the offset. If the forecast is 12 knots, the buoy reads 18 and the limit is 15, the trip is off. The correction adjusts a forecast, and the measurement stands on its own. Apply the offset to the decision threshold, so the rule reads as one line: southwest, subtract 3, compare with 15. A rule that needs a discussion at the ramp will not be applied at the ramp. The weather service uses a similar scale for a material change. It amends a marine forecast when the sustained wind changes by 10 knots or more. It also amends on direction: more than 45 degrees when winds are 20 knots or greater, and more than 60 degrees when they are below 20 knots. Use those scales to decide when the log should treat a day as a new regime. Round the offset. A 2 to 4 knot correction describes the evidence better than 2.7, and a whole-knot rule survives a windy morning. Write the rounded number on the log's first page. Review it at the start of each season and compare it with the first rows of the new block, because a correction that has not been checked against a fresh block is a memory of last year's weather.
Reset the sample when the inputs change
A bias belongs to a set of inputs, and it dies when they change. A model upgrade can reset it overnight, because new resolution and new physics produce a different error. A buoy relocation or a sensor change resets the comparison, even when the station keeps its identifier. A move to a different launch puts a different headland between the forecast point and the water, which changes the terrain effect that created the bias. Distance to the buoy bounds the correction. A station across the bay describes different water from the launch, and a station 30 nautical miles offshore can report a wind that never reaches the beach. The smaller the distance and the more similar the exposure, the more the gap is model bias rather than spatial difference. Log the station identifier so a later reader can check which comparison produced the number. A warning outranks any local offset. A small craft advisory or a gale warning cancels the correction for the day, and the log should record that the rule was overridden rather than applied. The offset is a tool for the marginal middle of the range, where the forecast is close to the limit and the decision is genuinely uncertain. It has no business arguing with a warning. Keep the current block at the front of the log and archive the old ones. Rebuild after a model change, a season change or a station change, and note the date of the rebuild. Five trips later, the new offset replaces the old one. The page stays useful because it gives one number for one direction in a single glance.
Related reading
Planning a weather window Gusts versus sustained wind Wind speed thresholds for small boats
Sources
- NWS Instruction 10-303: Marine and Coastal Services Standards and Guidelines — NOAA National Weather Service · T1
- Does NDBC adjust C-MAN and buoy wind speed observations to a standard height? — NOAA National Data Buoy Center · T1
- Keeping A Log — The Fisherman · T3