Why the Forecast Is Always Wrong on Terrain-Influenced Water

Collection: Field Notes — Old Fashioned Seamanship

Subject: Why the forecast is always wrong on terrain-influenced water — the structural gap between what numerical weather prediction models can do and what actually happens when large-scale weather meets coastlines, hills, islands, and the thermal boundary between land and sea

The forecast said Force 3 to 4, backing southwest. You are motoring out of a Scottish sea loch into a Force 6 funnelled down from the hills with a confused chop at the entrance and a wind direction that has nothing to do with southwest. The forecast was not wrong in any conventional sense. The forecast was correct for the region. The region is not where you are.

This is the central problem of inshore weather forecasting, and it has nothing to do with forecasting incompetence. It is a structural consequence of how numerical weather prediction works, what it can and cannot represent, and the scale at which you actually experience weather when sailing in tidal coastal and estuarine waters surrounded by complex terrain.

Understanding why this gap exists — and what you can do about it from a small boat — is the foundation for everything else in this series.

What numerical weather prediction actually does

Modern weather forecasts are produced by numerical weather prediction models — mathematical descriptions of the atmosphere solved on a grid of points spread across the globe or a region. The Met Office's global model, like its equivalents at ECMWF, NOAA, and other agencies, divides the atmosphere into a three-dimensional grid and solves the equations of fluid dynamics and thermodynamics for each grid cell at each time step, advancing the state of the atmosphere forward in time from an initial observed state.

Simon Rowell explains in Weather at Sea that this process is remarkably sophisticated and has produced dramatic improvements in forecast skill over the past few decades — a five-day forecast today is more reliable than a two-day forecast was in the 1980s. The large-scale patterns: the position and movement of fronts, pressure systems, and jet stream configurations are forecast with genuine accuracy out to four or five days. Beyond that, skill degrades rapidly. Beyond ten days, the forecast is essentially climatological background with high-profile caveats.

But the limitation that matters most to a coastal sailor has nothing to do with forecast horizon. It is spatial resolution. A global model with a grid spacing of ten kilometres cannot represent a headland two kilometres wide. A regional high-resolution model running at one and a half kilometres — which is roughly what the Met Office UKV model achieves over the British Isles — can begin to represent that headland, but only as a smeared approximation. The actual wind acceleration around Portland Bill, the actual turbulence zone in the lee of the Mull of Kintyre, the actual channelling effect of the Pentland Firth: these operate at scales and with a complexity that even the best operational forecast models only approximate.

Rowell explicitly notes in Weather at Sea that sea breeze effects — one of the most consistently important local weather phenomena for inshore sailors — are generally not well forecast by standard models until resolution drops to around two kilometres or less. A forecast product derived from a ten-kilometre model will not show you the sea breeze. It will show you the synoptic flow that the sea breeze might or might not override. This is a different thing.

The two worlds Gooley describes

Tristan Gooley opens The Secret World of Weather with a distinction that is worth understanding precisely, because it explains the structural relationship between professional forecasting and what we actually experience.

The "known world" of meteorology is the large-scale world of synoptic systems: highs, lows, fronts, isobars, the broad wind fields associated with them. This world is well-mapped, well-understood, and well-forecast at the scales it operates on. When a professional forecaster says it will be windy in the southwest on Thursday, they are describing this known world with competence.

The "secret world" is the weather you actually experience: different on two sides of a tree, different at the cliff face and a mile offshore, different on the windward and lee sides of a hill, different at the entrance to a sea loch than at its head. This is what Gooley calls microclimate — though he notes with some impatience that meteorologists often use this word to dismiss the whole territory as not really being weather, when in fact it is precisely the weather you are living in!

The gap between these two worlds is not a failure of either. Professional meteorologists are producing forecasts for regions containing millions of people and cannot tailor them to every coastline feature. The secret world operates at scales the models cannot resolve. And the sailor operating in inshore waters with a complex coastline is operating almost entirely in the secret world, with only the known world as background context.

How the land creates its own weather

The fundamental mechanism is differential heating. Land heats faster than sea in sunlight and cools faster at night. This creates pressure differences, which generate winds. The wind generated by a sun-warmed coast is not in the synoptic forecast — the model's grid cell is too coarse to represent it. Neither is the cold air draining down a Scottish glen at dawn, nor the acceleration zone around a headland, nor the wind shadow in the lee of a hill chain.

Gooley describes the physics of this in The Secret World of Weather clearly and accessibly. The sun warms different surfaces at different rates: dark soil faster than light sand, dry land faster than moist, woodland faster than open water. Warm air rises, creating a local low pressure. Cooler air flows in to replace it from the nearest available source. If that source is the sea, you get a sea breeze. If the warming is over a hill, you get thermals and cloud development. If the cooling is over a high plateau, you get cold air drainage.

Every one of these effects is predictable in principle — because the physics is well understood — but is not in the forecast because the model cannot see the terrain that generates it. This is where local knowledge compensates directly for model limitation. A sailor who knows that a particular estuary produces a strong funnelled southwesterly at the entrance on summer afternoons whenever the synoptic wind is in the southwest, regardless of what the forecast says about Force 3, has information the forecast cannot provide. That knowledge is built by observation over repeated passages, not by reading a better forecast.

Buys-Ballot's Law and what happens at the shore

Rowell describes a mechanism in Weather at Sea that is immediately practical: the behaviour of wind as it crosses the boundary between land and sea. The key principle is that surface friction over land is greater than over sea, so wind flowing from sea to land slows down, and wind flowing from land to sea speeds up. Because of the Coriolis effect, this change in speed also produces a change in direction.

In the Northern Hemisphere, when wind crosses from sea to land and slows, it backs — turns anticlockwise. When wind crosses from land to sea and accelerates, it veers — turns clockwise. The practical consequence is that a wind blowing offshore along a coast will veer as it passes over the water. If you are on starboard tack sailing offshore, you will be lifted. If you are on port tack sailing offshore, you will be headed. This is real and consistent and it is not in the forecast. It is a consequence of the land-sea friction boundary that operates at the scale of a few kilometres from shore.

The amount of this direction shift depends on the roughness of the land surface and the temperature contrast between land and sea. A built-up area produces more friction than open fields. Warm land in summer produces more vertical mixing, which reduces the effect. On a winter day with cold land and a warmer sea the effect is more pronounced and extends further offshore. None of this is in a forecast product derived from a regional model. All of it is physically predictable once you understand the mechanism.

Convergence and acceleration at headlands

Rowell identifies acceleration zones around headlands as one of the most consistently dangerous terrain effects for coastal sailors. Where wind is constricted by the local topography — squeezed around a headland or through the gap between islands — it accelerates, because the same mass of air must pass through a smaller cross-section in the same time and can only do so by moving faster. The visual signature of this on the water surface is a visible line — the wind line — where the texture of the water changes abruptly from a lighter, less-rippled area to a darker, rougher one as the wind picks up. In daylight this is often visible from some distance. At night it is invisible until you are in it.

The correlation between strong wind acceleration zones and strong tidal streams is not coincidental. Headlands produce both, for the same reason: any constriction in the flow path of fluid — whether air or water — produces acceleration. Portland Bill, Ardnamurchan, Duncansby Head, the Mull of Galloway: all are simultaneously tidal acceleration zones and wind acceleration zones. Wind against tide in these locations in moderate synoptic conditions can produce sea states that bear no relationship to the forecast Beaufort number. This is covered from the wave mechanics perspective in What Waves Know; the point here is that the meteorological component of the hazard is also invisible to the forecast model.

Katabatic winds and Scottish waters

Rowell describes katabatic winds specifically in Weather at Sea in the context of Norwegian fjords and the Adriatic, but the mechanism is directly relevant to Scottish west coast sailing.

A katabatic wind forms when cold, dense air accumulated on a plateau or hillside is pushed offshore by a light gradient wind, and flows down the slope under gravity, accelerating as it goes. The wind at the foot of the slope — and therefore on the water offshore — can be dramatically stronger than the gradient wind that triggered the event. Rowell gives an example of wind increasing from a gentle breeze to approximately sixty knots as a cold air mass builds momentum descending from high ground. In the Scottish context this is a known hazard in the lochs and sounds between the mainland hills and the outer islands, particularly after settled cold weather has allowed a cold air pool to accumulate over the higher ground.

These events are essentially unforecastable at the spatial resolution of an operational weather model. The model might show a light offshore gradient wind, which is accurate at the model's scale. The katabatic event that this gradient wind triggers occurs on the slope of a specific hill at a specific time and is over before the model's next output time step. A skipper who sees the grass moving at the top of a steep hillside on an otherwise calm day in cold settled weather has more information than the forecast provides.

What this means in practice for inshore sailing

Gooley's framing in The Secret World of Weather is useful here: the professional forecaster is describing a region, not your exact location. The forecaster knows this and has no alternative — they cannot produce a separate forecast for every headland and sea loch in the British Isles. The skill of the inshore sailor is to use the regional forecast as the background context and then apply local knowledge, observation, and the physical understanding of terrain effects to fill the gap.

This is not anti-technology. The synoptic forecast is essential background. A five-day synoptic forecast is genuinely useful for passage planning at the scale of a Channel crossing or a North Sea passage, where the large-scale patterns dominate. As soon as the scale narrows to the entrance of a harbour, a tidal race around a headland, or a coastal passage within a few miles of complex terrain, the forecast's predictive power diminishes rapidly and the observer's local knowledge and physical understanding become the primary instrument.

The subsequent posts in this series build the observational toolkit that closes this gap: What Hills Do to Wind covers terrain effects in detail, The Sea Breeze and the Land Breeze covers thermal circulation, Fog on Inland and Coastal Waters covers visibility, and Field Forecasting — Be Your Own Meteorologist brings these together into a practical short-range forecast framework.

The synoptic chart material that underpins this series — the depressions, fronts, highs, and jet stream behaviour that provide the large-scale context — is in the shared posts with the Passage Planning series, starting with How to Read a Synoptic Chart.

Tristan Gooley's The Secret World of Weather (Sceptre) is the primary source for the two-worlds framework, the physics of terrain-generated weather, and the observational approach to closing the gap between forecast and experience. Simon Rowell's Weather at Sea (Fernhurst Books) provides the professional forecasting perspective: what the models do well, where resolution limits their inshore predictive power, and the specific mechanisms of Buys-Ballot's Law, convergence and divergence, acceleration zones, and katabatic winds.

The full series index is at Weather Forecasting.