Key facts
| Question | What the sources support |
|---|---|
| Height effect | About 1 hPa per 8 m near sea level |
| Altitude error | A 100 m elevation error shifts pressure by roughly 12 hPa |
| Sensor limit | Consumer sensors are specified to about 0.1 hPa resolution, not laboratory accuracy |
| App source | Many apps display a nearby station's sea-level pressure instead of the raw sensor |
| Atmospheric tide | A semi-diurnal pressure oscillation of about 2 hPa in the tropics, smaller at mid-latitudes |
| Calibration | Compare at the same hour with an official station |
| Useful signal | The 3-hour tendency, not the absolute number |
| Placement | Keep the phone out of direct sun and out of sealed pockets |
What the sensor actually measures
A phone is an excellent relative-pressure logger and a poor absolute-pressure gauge, and almost every confusion about barometers on boats comes from mixing those two jobs. A barometer in a phone or watch is a micro-electromechanical pressure sensor: a small diaphragm that deflects as the surrounding air pressure changes, with the deflection read electrically. It responds to the absolute pressure of the air immediately around it. Meteorologists call that station pressure. It is a true local measurement, and it varies with weather, with altitude and with the environment around the phone. Station pressure is not what synoptic charts show. Charts use sea-level pressure, reduced to a common reference so that a coastal airport and an inland plateau can be compared on the same map. The reduction assumes a standard temperature profile through a fictitious column of air between the station and the sea, which is why an unusually warm or cold lower atmosphere introduces its own small error into every published sea-level value. The phone usually does not know its elevation precisely enough for that reduction, and the app may not use the raw sensor at all. Many weather apps display the sea-level pressure reported by the nearest airport or official station, which can be several kilometres away and at a different height. The number on the screen may therefore be a fair regional value and a poor local one.
Resolution is not accuracy
A phone typically reports pressure to 0.1 hPa because the sensor's internal conversion resolves that far, and that last digit invites more trust than it deserves. Resolution says how finely the number is divided; accuracy says how close it sits to the true value. The two are unrelated, and on a consumer sensor the second is several times the first. The practical error in the reading comes from three sources: the calibration offset, the temperature of the sensor cavity and the height term. None of them is visible in the display, and all of them can exceed the change you are trying to detect. Reading the third decimal as a meaningful signal is the fastest way to invent a pressure change that never happened. Drift is the slow version of the same problem. A sensor's offset can wander by a few tenths of a hectopascal over months, especially after repeated temperature cycling, and that drift is invisible within a single session. Recheck the calibration offset at the start of each season instead of carrying last year's number forward.
Height is the largest term
Near sea level, pressure falls by roughly 1 hPa for every 8 metres of rise. That is the hydrostatic relationship in its most useful form for coastal work: the atmosphere thins fast enough that a modest climb changes the reading more than most weather does. A 100 metre elevation error therefore shifts the reading by about 12 hPa. Compare that with a typical frontal passage, which might change the pressure by 10 to 20 hPa over a day. A cliff-top car park, the beach below it and a boat deck can differ by several hectopascals purely because of height, and none of that difference is weather. This is why comparing an uncorrected phone reading with a chart is misleading. The chart uses sea-level pressure; the phone uses station pressure. A cliff-top reading of 1000 hPa may correspond to a sea-level value near 1012 hPa, and the twelve-hectopascal gap is a geometry problem, not an atmospheric one. The correction is arithmetic, not mysterious. Subtract the height term, or let the app do it with a surveyed elevation. The rule must then be applied consistently: every reading in a comparison gets the same treatment. Mixing corrected and uncorrected values invents a gradient the atmosphere does not have. GPS altitude helps and is not sufficient. A consumer GPS fix is typically good to 10 to 30 metres vertically, which translates to 1 to 4 hPa — better than a guess, worse than a survey. For a fixed base station on a known jetty, use the map elevation. For a reading taken on a moving boat, assume the height term is unknown.
Labels are not interchangeable
The Weather Service glossary defines the altimeter setting as the pressure value used to set an aircraft altimeter so that it indicates altitude above mean sea level. It is the sea-level pressure in the form aviation uses, and it is what official stations report. An app may label its number "pressure", "sea-level pressure", "QNH" or "barometer". Those labels are not interchangeable. QNH is an altimeter setting; QFE is the pressure at the aerodrome itself, close to station pressure; a raw sensor value is station pressure; a model value is a forecast at a grid point, not an observation at your feet. Before comparing two devices, read the label. If one shows QNH and the other shows station pressure, they will disagree by the height term and neither is broken. If both show sea-level pressure but one comes from a station fifty kilometres inland, the difference may be real weather or may be distance, and there is no way to tell which without a third reading.
The atmosphere has its own tide
There is a pressure signal that has nothing to do with weather and that catches out anyone comparing readings at different times of day. The atmosphere has a semi-diurnal tide: pressure rises and falls twice a day, driven by solar heating of the whole atmospheric column. The amplitude is about 2 hPa in the tropics and falls to roughly 0.3 to 0.5 hPa at mid-latitudes, with maxima near local 10:00 and 22:00 and minima near 04:00 and 16:00. That is a small signal, but it is the same order as the change across a weak ridge, and it is perfectly repeatable. The practical consequence is a rule: compare readings at the same local hour, or correct for the tide. A three-hour tendency computed from 07:00 to 10:00 includes a rising tidal component that has nothing to do with the approaching low. A tendency computed from 22:00 to 01:00 includes a falling one. Neither invalidates the measurement; both require the same care that a tide table requires for water level.
Calibrate against an official station
Pick an official station within reach, preferably one that publishes hourly observations. At the top of the hour, compare its sea-level pressure with your device, after applying the height term. If the offset is stable across two or three visits, it is calibration. Record it, and subtract it. If the offset changes by more than a hectopascal between visits, something else is happening: a different location, a different label, a different sensor state, or a real pressure difference between two places. Repeat the comparison at the kind of location you actually use. A phone calibrated in an air-conditioned car park may behave differently on a hot boat deck, because temperature affects the sensor's internal reference as well as the phone's electronics.
Use the tendency, not the absolute value
Pressure tendency is the character and amount of change over a set interval, most often the previous three hours. That interval is long enough for weather to show and short enough to be operationally useful, and it is the quantity a phone measures well, because the height offset cancels when both readings come from the same place. The absolute number, once corrected, tells you which air mass you are in. The three-hour change tells you what the atmosphere is doing. For a decision about whether to go, the change is the more informative half. A phone that stays in the same pocket for three hours gives a usable tendency even when its absolute calibration is unknown. A phone carried from a cliff to a beach between readings gives a tendency dominated by altitude and worth nothing.
Placement changes the reading
Consumer barometers are designed for altitude and indoor navigation as much as for weather. They are good at relative change across minutes and hours, less good at absolute accuracy, and their readings drift with temperature and with the phone's own thermal state. Placement matters more than most people expect. A phone in direct sun heats up, and the sensor reports the pressure of a warmer cavity. A phone in a sealed waterproof pouch can trap air, lagging the ambient change by minutes. A phone against the body is warmer than the air and responds slowly. For a three-hour tendency, slow response is acceptable. For a squall passage, it is not. If the pressure is needed to confirm a fast change, take the phone out, shade it, wait for the reading to settle, and then compare.
What the phone cannot do
It cannot measure wind. It cannot measure sea state. It cannot define a gradient, because one station is one point. Those require other instruments, other stations or a chart. It also cannot replace an official station for a safety decision. The phone is a convenient trend logger and a second opinion. The official observation is the reference, and the gap between them is almost always explainable by height, distance, hour or label rather than by a broken sensor.
A protocol that works
Fix the phone at one location on the boat, out of the sun, in open air rather than a sealed pouch. Note the location and the time of every reading. Read it at the top of each hour and write the number down. Compute the three-hour difference rather than trusting an app's trend arrow, because the arrow may use a different interval or a model value. Correct the reading once, using a known elevation and an official station at the same hour, and record the offset. From that point on, work with the corrected tendency and ignore the absolute number for anything except confirming which air mass is overhead. That is the whole method, and it turns a consumer sensor into a useful instrument: one fixed place, one fixed hour, one correction, and a tendency that means something.
Sources
- Air Pressure — NOAA JetStream · T1
- National Weather Service Glossary — Altimeter Setting — NOAA National Weather Service · T1
- National Weather Service Glossary — Pressure Tendency — NOAA National Weather Service · T1